supervised_learning.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 5,
  "metadata": {
    "colab": {
      "provenance": [],
      "collapsed_sections": [
        "76f80fe8-a126-4cb7-bb10-695e971d9dc5",
        "d6c8364f-c90d-48ef-be41-6ff4465c9cd7",
        "aa6b8b5d-c38d-48bf-a263-4e4120ad1ce9",
        "023e34c2-f58c-40f7-92c7-d9e63ad5e9d9",
        "5e5e7fcc-6880-45d8-8611-34873ee22921",
        "76b16d87-9867-4639-a41c-1c7a3ac211dc",
        "fe947726-38dc-4e2c-be0a-dca6056d8437"
      ],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "27410c2c-7a8a-4d35-81b1-baae05054e7f"
      },
      "source": [
        "# ΕΡΓΑΣΙΑ 1"
      ],
      "id": "27410c2c-7a8a-4d35-81b1-baae05054e7f"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n9LCgD5KxO0J"
      },
      "source": [
        "* Ομάδα 53\n",
        "* Μέλη ομάδας:\n",
        "  * Αθανασίου Ιωάννης, 03117041\n",
        "  * Καραβαγγέλης Αθανάσιος, 03117022\n",
        "  * Τσιτσής Αντώνιος, 03117045"
      ],
      "id": "n9LCgD5KxO0J"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qOiMjhrRx5FF"
      },
      "source": [
        "## Εγκαταστάσεις πακέτων"
      ],
      "id": "qOiMjhrRx5FF"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ee1215b5-e4fd-412a-a068-786989b4895e",
        "outputId": "f5834b2e-f24b-499a-8086-c7453474e7da"
      },
      "source": [
        "!pip install --upgrade pip #upgrade pip package installer\n",
        "!pip install scikit-learn --upgrade #upgrade scikit-learn package\n",
        "!pip install numpy --upgrade #upgrade numpy package\n",
        "!pip install pandas --upgrade #upgrade pandas package\n",
        "!pip install --upgrade matplotlib # Κάνουμε update την matplotlib\n",
        "!pip install optuna"
      ],
      "id": "ee1215b5-e4fd-412a-a068-786989b4895e",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Requirement already satisfied: pip in /usr/local/lib/python3.7/dist-packages (21.3.1)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n",
            "Requirement already satisfied: scikit-learn in /usr/local/lib/python3.7/dist-packages (1.0.1)\n",
            "Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn) (3.0.0)\n",
            "Requirement already satisfied: scipy>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn) (1.4.1)\n",
            "Requirement already satisfied: numpy>=1.14.6 in /usr/local/lib/python3.7/dist-packages (from scikit-learn) (1.21.4)\n",
            "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn) (1.1.0)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (1.21.4)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n",
            "Requirement already satisfied: pandas in /usr/local/lib/python3.7/dist-packages (1.3.4)\n",
            "Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas) (2.8.2)\n",
            "Requirement already satisfied: pytz>=2017.3 in /usr/local/lib/python3.7/dist-packages (from pandas) (2018.9)\n",
            "Requirement already satisfied: numpy>=1.17.3 in /usr/local/lib/python3.7/dist-packages (from pandas) (1.21.4)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.7.3->pandas) (1.15.0)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.7/dist-packages (3.5.0)\n",
            "Requirement already satisfied: numpy>=1.17 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (1.21.4)\n",
            "Requirement already satisfied: pillow>=6.2.0 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (7.1.2)\n",
            "Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (21.3)\n",
            "Requirement already satisfied: fonttools>=4.22.0 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (4.28.2)\n",
            "Requirement already satisfied: pyparsing>=2.2.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (3.0.6)\n",
            "Requirement already satisfied: python-dateutil>=2.7 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (2.8.2)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (0.11.0)\n",
            "Requirement already satisfied: setuptools-scm>=4 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (6.3.2)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.7/dist-packages (from matplotlib) (1.3.2)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.7->matplotlib) (1.15.0)\n",
            "Requirement already satisfied: tomli>=1.0.0 in /usr/local/lib/python3.7/dist-packages (from setuptools-scm>=4->matplotlib) (1.2.2)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.7/dist-packages (from setuptools-scm>=4->matplotlib) (57.4.0)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n",
            "Requirement already satisfied: optuna in /usr/local/lib/python3.7/dist-packages (2.10.0)\n",
            "Requirement already satisfied: colorlog in /usr/local/lib/python3.7/dist-packages (from optuna) (6.6.0)\n",
            "Requirement already satisfied: cmaes>=0.8.2 in /usr/local/lib/python3.7/dist-packages (from optuna) (0.8.2)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from optuna) (4.62.3)\n",
            "Requirement already satisfied: cliff in /usr/local/lib/python3.7/dist-packages (from optuna) (3.10.0)\n",
            "Requirement already satisfied: scipy!=1.4.0 in /usr/local/lib/python3.7/dist-packages (from optuna) (1.4.1)\n",
            "Requirement already satisfied: sqlalchemy>=1.1.0 in /usr/local/lib/python3.7/dist-packages (from optuna) (1.4.27)\n",
            "Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.7/dist-packages (from optuna) (21.3)\n",
            "Requirement already satisfied: alembic in /usr/local/lib/python3.7/dist-packages (from optuna) (1.7.5)\n",
            "Requirement already satisfied: PyYAML in /usr/local/lib/python3.7/dist-packages (from optuna) (3.13)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from optuna) (1.21.4)\n",
            "Requirement already satisfied: pyparsing!=3.0.5,>=2.0.2 in /usr/local/lib/python3.7/dist-packages (from packaging>=20.0->optuna) (3.0.6)\n",
            "Requirement already satisfied: greenlet!=0.4.17 in /usr/local/lib/python3.7/dist-packages (from sqlalchemy>=1.1.0->optuna) (1.1.2)\n",
            "Requirement already satisfied: importlib-metadata in /usr/local/lib/python3.7/dist-packages (from sqlalchemy>=1.1.0->optuna) (4.8.2)\n",
            "Requirement already satisfied: importlib-resources in /usr/local/lib/python3.7/dist-packages (from alembic->optuna) (5.4.0)\n",
            "Requirement already satisfied: Mako in /usr/local/lib/python3.7/dist-packages (from alembic->optuna) (1.1.6)\n",
            "Requirement already satisfied: PrettyTable>=0.7.2 in /usr/local/lib/python3.7/dist-packages (from cliff->optuna) (2.4.0)\n",
            "Requirement already satisfied: pbr!=2.1.0,>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from cliff->optuna) (5.8.0)\n",
            "Requirement already satisfied: cmd2>=1.0.0 in /usr/local/lib/python3.7/dist-packages (from cliff->optuna) (2.3.3)\n",
            "Requirement already satisfied: stevedore>=2.0.1 in /usr/local/lib/python3.7/dist-packages (from cliff->optuna) (3.5.0)\n",
            "Requirement already satisfied: autopage>=0.4.0 in /usr/local/lib/python3.7/dist-packages (from cliff->optuna) (0.4.0)\n",
            "Requirement already satisfied: typing-extensions in /usr/local/lib/python3.7/dist-packages (from cmd2>=1.0.0->cliff->optuna) (3.10.0.2)\n",
            "Requirement already satisfied: pyperclip>=1.6 in /usr/local/lib/python3.7/dist-packages (from cmd2>=1.0.0->cliff->optuna) (1.8.2)\n",
            "Requirement already satisfied: attrs>=16.3.0 in /usr/local/lib/python3.7/dist-packages (from cmd2>=1.0.0->cliff->optuna) (21.2.0)\n",
            "Requirement already satisfied: wcwidth>=0.1.7 in /usr/local/lib/python3.7/dist-packages (from cmd2>=1.0.0->cliff->optuna) (0.2.5)\n",
            "Requirement already satisfied: zipp>=0.5 in /usr/local/lib/python3.7/dist-packages (from importlib-metadata->sqlalchemy>=1.1.0->optuna) (3.6.0)\n",
            "Requirement already satisfied: MarkupSafe>=0.9.2 in /usr/local/lib/python3.7/dist-packages (from Mako->alembic->optuna) (2.0.1)\n",
            "\u001b[33mWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\u001b[0m\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gIqPcc4RT8xc"
      },
      "source": [],
      "id": "gIqPcc4RT8xc",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cad74aaa-6be6-4900-8cca-11f1023aac1b"
      },
      "source": [
        "## ΜΕΡΟΣ 1"
      ],
      "id": "cad74aaa-6be6-4900-8cca-11f1023aac1b"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "76f80fe8-a126-4cb7-bb10-695e971d9dc5"
      },
      "source": [
        "### Dataset: U02 - Εισαγωγή και επισκόπηση"
      ],
      "id": "76f80fe8-a126-4cb7-bb10-695e971d9dc5"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d6c8364f-c90d-48ef-be41-6ff4465c9cd7"
      },
      "source": [
        "#### Σύντομη παρουσίαση"
      ],
      "id": "d6c8364f-c90d-48ef-be41-6ff4465c9cd7"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5e63714b-1d02-4678-b55b-166e24d0ec60"
      },
      "source": [
        "* To dataset αφορά το classification δύο πνευματικών καταστάσεων με βάση καταγεγραμμένα ηλεκτροεγκεφαλογραφικά σήματα.\n",
        "* Οι πνευματικές καταστάσεις με τις οποίες ασχολούμαστε είναι η ηρεμία και η νοητική προετοιμασία μίας κίνησης."
      ],
      "id": "5e63714b-1d02-4678-b55b-166e24d0ec60"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aa6b8b5d-c38d-48bf-a263-4e4120ad1ce9"
      },
      "source": [
        "#### Διάβασμα dataset"
      ],
      "id": "aa6b8b5d-c38d-48bf-a263-4e4120ad1ce9"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "a2f40095-6832-47b8-b494-69c63eef5834",
        "outputId": "dd0452f5-063c-43c1-8f5d-c39a84f45fa2"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "data = pd.read_csv(\"plrx.txt\", sep=\"\\t \", header=None)\n",
        "print(data)"
      ],
      "id": "a2f40095-6832-47b8-b494-69c63eef5834",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "          0         1         2         3   ...        9         10        11   12\n",
            "0   -0.17936 -0.207000 -0.209710 -0.097260  ... -0.032886  0.003383 -0.334250  1.0\n",
            "1   -0.14659 -0.164940  0.249870 -0.073985  ...  0.050759 -0.017511  0.066959  1.0\n",
            "2   -0.13131  0.518160  0.432810  0.397890  ... -0.040303  0.088057  0.631200  2.0\n",
            "3    0.64940  0.418780  0.148350  0.333490  ... -0.367230  0.364680  0.571720  1.0\n",
            "4   -0.40628  0.115030  0.073336  0.070066  ...  0.153750 -0.126950  0.184570  1.0\n",
            "..       ...       ...       ...       ...  ...       ...       ...       ...  ...\n",
            "177 -0.55005  0.276290 -0.273240  0.113190  ... -0.014462  0.002273 -0.100480  2.0\n",
            "178 -0.78399  0.254600 -0.700990 -0.299170  ...  0.020018 -0.020915  0.663940  1.0\n",
            "179  0.29877 -0.179620  0.032859 -0.899010  ... -0.411130  0.246830 -0.389160  2.0\n",
            "180 -0.37812 -0.522870  0.215410 -0.243590  ... -0.039117  0.007591  0.476030  1.0\n",
            "181  0.37913 -0.046562 -0.123360 -0.063958  ... -0.242290  0.173690  0.490630  2.0\n",
            "\n",
            "[182 rows x 13 columns]\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:3: ParserWarning: Falling back to the 'python' engine because the 'c' engine does not support regex separators (separators > 1 char and different from '\\s+' are interpreted as regex); you can avoid this warning by specifying engine='python'.\n",
            "  This is separate from the ipykernel package so we can avoid doing imports until\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "023e34c2-f58c-40f7-92c7-d9e63ad5e9d9"
      },
      "source": [
        "#### Επεξεργασία dataset"
      ],
      "id": "023e34c2-f58c-40f7-92c7-d9e63ad5e9d9"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "11eac077-7d3e-4a51-b63d-cb22df0e17b4"
      },
      "source": [
        "* Παρατηρώ ότι η στήλη 12 παίρνει ως τιμές μόνο το 1.0 και το 2.0, και επομένως η στήλη αυτή θα περιέχει τα labels.\n",
        "* Θα διαχωρίσω τα features από τα labels.\n",
        "* Θα μετατρέψω τα labels ώστε να παίρνουν τιμές 0 και 1 ως εξής: 1.0->0, 2.0->1."
      ],
      "id": "11eac077-7d3e-4a51-b63d-cb22df0e17b4"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "e518356c-46ff-4edc-98ea-8a7d0ef0bbc5",
        "outputId": "c6393dd2-c326-48ae-b294-ce1f4a4f6e14"
      },
      "source": [
        "import numpy as np\n",
        "labels = data.iloc[:, [12]].values\n",
        "def normalize(label):\n",
        "    assert label==2. or label==1.\n",
        "    if label==1.:\n",
        "        return 0\n",
        "    if label==2.:\n",
        "        return 1\n",
        "vectorized_norm = np.vectorize(normalize)\n",
        "labels = np.array([vectorized_norm(row) for row in labels])\n",
        "\n",
        "features = data.iloc[:, :12].values\n",
        "\n",
        "print(\"labels shape:\", labels.shape)\n",
        "print(\"features shape:\", features.shape)"
      ],
      "id": "e518356c-46ff-4edc-98ea-8a7d0ef0bbc5",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "labels shape: (182, 1)\n",
            "features shape: (182, 12)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5e5e7fcc-6880-45d8-8611-34873ee22921"
      },
      "source": [
        "#### Παρατηρήσεις"
      ],
      "id": "5e5e7fcc-6880-45d8-8611-34873ee22921"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "568dbc82-cb31-45a6-8f14-d042d41546d9"
      },
      "source": [
        "* Κάθε example έχει 12 features και 1 label.\n",
        "* Έχουμε 182 examples.\n",
        "* Δεν υπάρχει αρίθμηση γραμμών.\n",
        "* Δεν υπάρχουν επικεφαλίδες.\n",
        "* Δεν υπάρχουν απουσιάζουσες τιμές.\n",
        "* Δεν υπάρχουν μη διατεταγμένα χαρακτηριστικά."
      ],
      "id": "568dbc82-cb31-45a6-8f14-d042d41546d9"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "caa40642-779f-40df-b4e9-b1e6659832fc",
        "outputId": "f2032493-e592-41a3-d5b9-887932e2894b"
      },
      "source": [
        "count = np.bincount(labels.flatten())\n",
        "print(f\"Class-0: {count[0]} examples, {round(100*count[0]/sum(count), 3)}%\")\n",
        "print(f\"Class-1: {count[1]} examples, {round(100*count[1]/sum(count), 3)}%\")"
      ],
      "id": "caa40642-779f-40df-b4e9-b1e6659832fc",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Class-0: 130 examples, 71.429%\n",
            "Class-1: 52 examples, 28.571%\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f604ab6a-48d7-41e5-9789-f879d69c5875"
      },
      "source": [
        "* Μετράω ποσοστά 71.429% και 28.571% και επομένως καταλήγω στο συμπέρασμα ότι το dataset δεν είναι ισορροπημένο."
      ],
      "id": "f604ab6a-48d7-41e5-9789-f879d69c5875"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "76b16d87-9867-4639-a41c-1c7a3ac211dc"
      },
      "source": [
        "### Προετοιμασία"
      ],
      "id": "76b16d87-9867-4639-a41c-1c7a3ac211dc"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fe947726-38dc-4e2c-be0a-dca6056d8437"
      },
      "source": [
        "##### Διαχωρίζω σε train και test set με αναλογίες 70-30"
      ],
      "id": "fe947726-38dc-4e2c-be0a-dca6056d8437"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "00d7f123-d41f-40e4-944c-4a6b8c7dad97",
        "outputId": "767eee0f-310c-4c52-a3dd-565449ac6824"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "train, test, train_labels, test_labels = train_test_split(features, labels, test_size=0.3)\n",
        "train_pipeline, test_pipeline, train_labels_pipeline, test_labels_pipeline = train_test_split(features, labels, test_size=0.3)\n",
        "train_grid, test_grid, train_labels_grid, test_labels_grid = train_test_split(features, labels, test_size=0.3)\n",
        "print(train.shape, train_labels.shape)\n",
        "print(test.shape, test_labels.shape)"
      ],
      "id": "00d7f123-d41f-40e4-944c-4a6b8c7dad97",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(127, 12) (127, 1)\n",
            "(55, 12) (55, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d5ba8d41-72e5-46af-9cbd-d07340d09c46"
      },
      "source": [
        "* Δεν υπάρχουν απουσιάζουσες τιμές, ούτε κατηγορικά/μη διατεταγμένα χαρακτηριστικά."
      ],
      "id": "d5ba8d41-72e5-46af-9cbd-d07340d09c46"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4611403f-173d-44d9-9a82-5edf42409231"
      },
      "source": [
        "### Ταξινόμηση"
      ],
      "id": "4611403f-173d-44d9-9a82-5edf42409231"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aa092efb-baf2-4b69-93ce-286507f4a829"
      },
      "source": [
        "#### Επιδόσεις out-of-the-box"
      ],
      "id": "aa092efb-baf2-4b69-93ce-286507f4a829"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "48b6e0bb-6ed8-4769-8d4e-a05387ad030e",
        "outputId": "d53dae79-9069-41d3-beb9-d5effd459c03"
      },
      "source": [
        "from sklearn.dummy import DummyClassifier\n",
        "from sklearn.naive_bayes import GaussianNB\n",
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from sklearn.model_selection import cross_val_score\n",
        "from sklearn.metrics import accuracy_score, f1_score\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "\n",
        "dummy = DummyClassifier()\n",
        "gnb = GaussianNB()\n",
        "knn = KNeighborsClassifier()\n",
        "log_reg = LogisticRegression()\n",
        "\n",
        "models = [\n",
        "    {\n",
        "        \"name\": \"dummy\",\n",
        "        \"model\": dummy,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"bayes\",\n",
        "        \"model\": gnb,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"Knn\",\n",
        "        \"model\": knn,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"log_regr\",\n",
        "        \"model\": log_reg,\n",
        "    },  \n",
        "]\n",
        "\n",
        "def compute_simple_scores(models, train, train_labels):\n",
        "    accuracy = dict()\n",
        "    f1 = dict()\n",
        "    for m in models:\n",
        "        name, model = m.values()\n",
        "        # print(model)\n",
        "        model.fit(train, train_labels.ravel())\n",
        "        preds = model.predict(test)\n",
        "        accuracy[name] = round(accuracy_score(test_labels, preds), 3)\n",
        "        f1[name] = round(f1_score(test_labels, preds), 3)\n",
        "    return accuracy, f1\n",
        "\n",
        "out_of_the_box_accuracy, out_of_the_box_f1 = compute_simple_scores(models, train, train_labels)\n",
        "print('accuracy:', out_of_the_box_accuracy)\n",
        "print('f1-score:', out_of_the_box_f1)"
      ],
      "id": "48b6e0bb-6ed8-4769-8d4e-a05387ad030e",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "accuracy: {'dummy': 0.727, 'bayes': 0.6, 'Knn': 0.636, 'log_regr': 0.709}\n",
            "f1-score: {'dummy': 0.0, 'bayes': 0.0, 'Knn': 0.091, 'log_regr': 0.0}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9df73777-4721-444d-a016-81d1ff2fb79e"
      },
      "source": [
        "##### Πίνακας σύγκρισης"
      ],
      "id": "9df73777-4721-444d-a016-81d1ff2fb79e"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "d88aeca7-9a67-44d2-9670-0303fa26f3d9",
        "outputId": "ce5e3752-099e-4d98-de14-c6388a35a2ca"
      },
      "source": [
        "def print_scores_table(models, accuracy, f1, train_times=None, test_times=None):\n",
        "    if train_times:\n",
        "        print(\"Model      \\t| accuracy\\t| f1-score\\t| train_time(ms)\\t| test_time (ms)\")\n",
        "    else:\n",
        "        print(\"Model      \\t| accuracy\\t| f1-score\")\n",
        "    print(\"-------------------------------------------\")\n",
        "    for m in models:\n",
        "        name, model = m.values()\n",
        "        acc, f1_score = accuracy[name], f1[name]\n",
        "        if train_times:\n",
        "            time = test_times[name]\n",
        "            print(f\"{name}      \\t| {acc}      \\t| {f1_score}\\t\\t| {train_times[name]}\\t\\t| {test_times[name]}\")\n",
        "        else:\n",
        "            print(f\"{name}      \\t| {acc}      \\t| {f1_score}\")\n",
        "            \n",
        "\n",
        "print_scores_table(models, out_of_the_box_accuracy, out_of_the_box_f1)"
      ],
      "id": "d88aeca7-9a67-44d2-9670-0303fa26f3d9",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model      \t| accuracy\t| f1-score\n",
            "-------------------------------------------\n",
            "dummy      \t| 0.727      \t| 0.0\n",
            "bayes      \t| 0.6      \t| 0.0\n",
            "Knn      \t| 0.636      \t| 0.091\n",
            "log_regr      \t| 0.709      \t| 0.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "f66d308e-10c5-426b-8e1b-d4b54a4acdc5"
      },
      "source": [
        "<table>\n",
        "<thead>\n",
        "  <td>Model</td>\n",
        "  <td>Accuracy-score</td>\n",
        "  <td>f1-score</td>\n",
        "</thead>\n",
        "<tbody>\n",
        "<tr>\n",
        "  <td>Dummy</td>\n",
        "  <td>0.727</td>\n",
        "  <td>0.0</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Bayes</td>\n",
        "  <td>0.6</td>\n",
        "  <td>0.0</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Knn</td>\n",
        "  <td>0.636</td>\n",
        "  <td>0.091</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Log.regression</td>\n",
        "  <td>0.709</td>\n",
        "  <td>0.0</td>\n",
        "</tr>\n",
        "</tbody>\n",
        "</table>"
      ],
      "id": "f66d308e-10c5-426b-8e1b-d4b54a4acdc5"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d6765368-0705-4d89-8b4c-e48b89b8cba5"
      },
      "source": [
        "##### Bar plots σύγκρισης"
      ],
      "id": "d6765368-0705-4d89-8b4c-e48b89b8cba5"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 228
        },
        "id": "b357f6b3-56a9-4e9b-9b64-ebd6e52838c7",
        "outputId": "35391705-42a6-4795-e391-f082cfef9913"
      },
      "source": [
        "def plot_scores(accuracy, f1, fig_index=1, title_prefix=\"\", color=\"blue\"):\n",
        "    fig, axs = plt.subplots(1, 2, figsize=(20, 3), sharey=True)\n",
        "    axs[0].bar(accuracy.keys(), accuracy.values(), color=color)\n",
        "    axs[0].set_title(f\"{title_prefix} Accuracies\")\n",
        "\n",
        "    axs[1].bar(f1.keys(), f1.values(), color=color)\n",
        "    axs[1].set_title(f\"{title_prefix} F1-scores\")\n",
        "\n",
        "\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1)"
      ],
      "id": "b357f6b3-56a9-4e9b-9b64-ebd6e52838c7",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "924e7773-c448-4770-abe3-122e441bccaa"
      },
      "source": [
        "##### Συμπεράσματα"
      ],
      "id": "924e7773-c448-4770-abe3-122e441bccaa"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8ed909eb-f58b-48b0-a8f0-212f6e033403"
      },
      "source": [
        "* Παρατηρώ ότι όλα τα μοντέλα έχουν, στο testing set, σχετικά ικανοποιητικά accuracie-scores.\n",
        "* Ωστόσο, στην μετρική του f1-score, παρατηρούνται αντίθετα αποτελέσματα, αφού όλα τα μοντέλα έχουν μικρά score.    "
      ],
      "id": "8ed909eb-f58b-48b0-a8f0-212f6e033403"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3a9f0065-6767-420f-8198-e0eb0b1ce0cd"
      },
      "source": [
        "#### Βελτιστοποίηση με προεπεξεργασία και χρήση pipelines"
      ],
      "id": "3a9f0065-6767-420f-8198-e0eb0b1ce0cd"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3832488c-82ea-4cdf-9cdf-0c3506df8635"
      },
      "source": [
        "##### Για αρχή, κατασκευάζουμε ένα pipeline και χρησιμοποιούμε τις default τιμές των υπερπαραμέτρων των μετασχηματιστών και των εκτιμητών"
      ],
      "id": "3832488c-82ea-4cdf-9cdf-0c3506df8635"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fc888583-cf47-4394-9344-084f9f2eca72"
      },
      "source": [
        "* Έχω ένα μοντέλο για κάθε ταξινομητή, αφού δεν δοκιμάζω διαφορετικές τιμές για τις υπερπαραμέτρους\n",
        "* Επομένως, θα εκπαιδεύσω και πάλι κάθε μοντέλο σε όλο το training set και θα υπολογίσω την επίδοσή του στο testing set"
      ],
      "id": "fc888583-cf47-4394-9344-084f9f2eca72"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "045125a5-5166-427f-8375-23690c2d33e6",
        "outputId": "5f94dd78-8d82-47ca-bb82-6c682c18a408"
      },
      "source": [
        "from imblearn.pipeline import Pipeline\n",
        "from sklearn.feature_selection import VarianceThreshold\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from imblearn.over_sampling import RandomOverSampler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "dummy = DummyClassifier()\n",
        "gnb = GaussianNB()\n",
        "knn = KNeighborsClassifier()\n",
        "log_reg = LogisticRegression()\n",
        "\n",
        "models = [\n",
        "    {\n",
        "        \"name\": \"dummy\",\n",
        "        \"model\": dummy,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"bayes\",\n",
        "        \"model\": gnb,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"Knn\",\n",
        "        \"model\": knn,\n",
        "    },\n",
        "    {\n",
        "        \"name\": \"log_regr\",\n",
        "        \"model\": log_reg,\n",
        "    },  \n",
        "]\n",
        "\n",
        "def compute_pipeline_scores(models):\n",
        "    pipeline_accuracy = dict()\n",
        "    pipeline_f1 = dict()\n",
        "    \n",
        "    for modell in models:\n",
        "        name, model = modell.values()\n",
        "        selector = VarianceThreshold()\n",
        "        scaler = StandardScaler()\n",
        "        ros = RandomOverSampler()\n",
        "        pca = PCA()\n",
        "        pipe = Pipeline(\n",
        "            steps=[\n",
        "                ('selector', selector),\n",
        "                ('scaler', scaler),\n",
        "                ('sampler', ros),\n",
        "                ('pca', pca),\n",
        "                (name, model),\n",
        "            ]\n",
        "        )\n",
        "        \n",
        "        #pipeline_accuracy[name] = round(cross_val_score(pipe, train_pipeline, train_labels_pipeline.ravel(), cv=10, scoring='accuracy').mean(), 3)\n",
        "        #pipeline_f1[name] = round(cross_val_score(pipe, train_pipeline, train_labels_pipeline.ravel(), cv=10, scoring='f1').mean(), 3)\n",
        "\n",
        "        pipe.fit(train_pipeline, train_labels_pipeline)\n",
        "        preds = pipe.predict(test_pipeline)\n",
        "        pipeline_accuracy[name] = round(accuracy_score(test_labels, preds), 3)\n",
        "        pipeline_f1[name] = round(f1_score(test_labels, preds), 3)\n",
        "\n",
        "    return pipeline_accuracy, pipeline_f1\n",
        "\n",
        "pipe_accuracy, pipe_f1 = compute_pipeline_scores(models)\n",
        "print(pipe_accuracy)\n",
        "print(pipe_f1)"
      ],
      "id": "045125a5-5166-427f-8375-23690c2d33e6",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'dummy': 0.727, 'bayes': 0.473, 'Knn': 0.618, 'log_regr': 0.473}\n",
            "{'dummy': 0.0, 'bayes': 0.171, 'Knn': 0.364, 'log_regr': 0.216}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a3f5e016-a004-4e93-aa94-87ab3cb4b729"
      },
      "source": [
        "##### Εκτυπώνω τα αποτελέσματα σε πίνακα:"
      ],
      "id": "a3f5e016-a004-4e93-aa94-87ab3cb4b729"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0d08e0d7-2056-41ce-acb8-e8a2915e2f78",
        "outputId": "8bce23e1-50d6-45fa-a95d-bf1111e3c34c"
      },
      "source": [
        "print_scores_table(models, pipe_accuracy, pipe_f1)"
      ],
      "id": "0d08e0d7-2056-41ce-acb8-e8a2915e2f78",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model      \t| accuracy\t| f1-score\n",
            "-------------------------------------------\n",
            "dummy      \t| 0.727      \t| 0.0\n",
            "bayes      \t| 0.473      \t| 0.171\n",
            "Knn      \t| 0.618      \t| 0.364\n",
            "log_regr      \t| 0.473      \t| 0.216\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a8c5ec01-7883-4396-b5a3-efe049d67d2c"
      },
      "source": [
        "<table>\n",
        "<thead>\n",
        "  <td>Model</td>\n",
        "  <td>Accuracy-score</td>\n",
        "  <td>f1-score</td>\n",
        "</thead>\n",
        "<tbody>\n",
        "<tr>\n",
        "  <td>Dummy</td>\n",
        "  <td>0.727</td>\n",
        "  <td>0.0</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Bayes</td>\n",
        "  <td>0.473</td>\n",
        "  <td>0.171</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Knn</td>\n",
        "  <td>0.618</td>\n",
        "  <td>0.364</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>Log.regression</td>\n",
        "  <td>0.473</td>\n",
        "  <td>0.216</td>\n",
        "</tr>\n",
        "</tbody>\n",
        "</table>"
      ],
      "id": "a8c5ec01-7883-4396-b5a3-efe049d67d2c"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0d754c54-a8ac-45b4-8192-18afcbff2d91"
      },
      "source": [
        "##### Κατασκευάζω και τα bar plots:"
      ],
      "id": "0d754c54-a8ac-45b4-8192-18afcbff2d91"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 228
        },
        "id": "04e8d24a-0a01-4c5d-81b5-567fc97d502b",
        "outputId": "09213908-3783-4fac-c82a-6f3f2ba2d465"
      },
      "source": [
        "plot_scores(pipe_accuracy, pipe_f1)"
      ],
      "id": "04e8d24a-0a01-4c5d-81b5-567fc97d502b",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c7a5bd54-b806-4e59-8a29-0918ddea1d2c"
      },
      "source": [
        "* Παρατηρώ ότι, σε σύγκριση με τις out-of-the-box επιδόσεις χωρίς την χρήση pipelines:\n",
        "    * Μόνο το f1-score του dummy παραμένει μηδενικό.\n",
        "    * Όλα τα υπόλοιπα f1-scores έχουν αυξηθεί σημαντικά.\n",
        "    * Τα accuracy-scores παρουσιάζουν κάποιες μειώσεις, αλλά συνεχίζουν να κινούνται στο ίδιο εύρος με το προηγούμενο πείραμα."
      ],
      "id": "c7a5bd54-b806-4e59-8a29-0918ddea1d2c"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "98810e41-b3c8-45e8-9187-cd5c45959a55"
      },
      "source": [
        "#### Βελιστοποίηση με χρήση pipelines και gridsearchCV"
      ],
      "id": "98810e41-b3c8-45e8-9187-cd5c45959a55"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "be72ce46-1ab1-4a1e-b054-79611a1d77e4"
      },
      "source": [
        "##### Ορίζω τους ταξινομητές"
      ],
      "id": "be72ce46-1ab1-4a1e-b054-79611a1d77e4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "696a99c0-ec35-447a-9bda-ff8f448fa6ac"
      },
      "source": [
        "* Τώρα, θα δοκιμάσω, με την μέθοδο του grid search, διάφορες τιμές των υπερπαραμέτρων του κάθε ταξινομητή.\n",
        "* Θα κατασκευάσω, δηλαδή, πολλά μοντέλα για κάθε ταξινομητή, και θα βρω το βέλτιστο του κάθε ταξινομητή, ώστε να ελέξγω την επίδοσή του στο testing set.\n",
        "* Όσον αφορά την προεπεξεργασία των δεδομένων, για να αποφασίσω τις τιμές στις οποίες θα κινείται το threshold του selector, μετράω τις εξής διακυμάνσεις:"
      ],
      "id": "696a99c0-ec35-447a-9bda-ff8f448fa6ac"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fa64bd32-28dd-45c4-887e-f855269822cb",
        "outputId": "bc08819d-080a-4538-f24c-e11149ef7bff"
      },
      "source": [
        "train_variance = train_grid.var(axis=0)\n",
        "print(train_variance)\n",
        "print(np.max(train_variance))"
      ],
      "id": "fa64bd32-28dd-45c4-887e-f855269822cb",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0.22825624 0.15829012 0.11811736 0.16605055 0.2178019  0.17289506\n",
            " 0.21141897 0.13344763 0.23172825 0.17260755 0.11340507 0.10812781]\n",
            "0.23172825088761193\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "755a24f2-3f06-4032-baa0-afde51eac826"
      },
      "source": [
        "* Θα δοκιμάσω άρα λίγες τιμές στο περίπου εύρος 0.10-0.20.\n",
        "* Ορίζω τους εκτιμητές των μοντέλων που θα εκπαιδεύσω:"
      ],
      "id": "755a24f2-3f06-4032-baa0-afde51eac826"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "be2da9c3-8e3b-41e4-820b-73bd808ca6dd"
      },
      "source": [
        "from sklearn.model_selection import GridSearchCV\n",
        "\n",
        "def generate_estimators():\n",
        "    dummy = DummyClassifier()\n",
        "    gnb = GaussianNB()\n",
        "    knn = KNeighborsClassifier()\n",
        "    log_reg = LogisticRegression()\n",
        "\n",
        "    models = [\n",
        "        {\n",
        "            \"name\": \"dummy\",\n",
        "            \"model\": dummy,\n",
        "            \"hyper_param_name\": \"dummy__strategy\",\n",
        "            \"hyper_param_values\": [\"stratified\", \"most_frequent\", \"prior\", \"uniform\"],\n",
        "        },\n",
        "        {\n",
        "            \"name\": \"bayes\",\n",
        "            \"model\": gnb,\n",
        "            \"hyper_param_name\": None,\n",
        "            \"hyper_param_values\": None,\n",
        "        },\n",
        "        {\n",
        "            \"name\": \"knn\",\n",
        "            \"model\": knn,\n",
        "            \"hyper_param_name\": \"knn__n_neighbors\",\n",
        "            \"hyper_param_values\": [1, 2, 5, 8, 11, 14],\n",
        "        },\n",
        "        {\n",
        "            \"name\": \"log_regr\",\n",
        "            \"model\": log_reg,\n",
        "            \"hyper_param_name\": None,\n",
        "            \"hyper_param_values\": None,\n",
        "        },  \n",
        "    ]\n",
        "    return models"
      ],
      "id": "be2da9c3-8e3b-41e4-820b-73bd808ca6dd",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "620800fa-7da8-4a52-91f6-e28dd8089ef0"
      },
      "source": [
        "##### Ορίζω τις συναρτήσεις εκπαίδευσής τους και αξιολόγησής τους στο test set"
      ],
      "id": "620800fa-7da8-4a52-91f6-e28dd8089ef0"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "593c0dd0-0bcd-4cd7-9927-cba5b6586748"
      },
      "source": [
        "* Για κάθε ταξινομητή, εντοπίζω το καλύτερο μοντέλο του, με αναζήτηση πλέγματος με διασταυρούμενη επικύρωση."
      ],
      "id": "593c0dd0-0bcd-4cd7-9927-cba5b6586748"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "a22cd245-8aeb-477d-acf8-490ceddfe094"
      },
      "source": [
        "from sklearn.metrics import accuracy_score, f1_score\n",
        "import time\n",
        "\n",
        "def pick_best_models(useSelector=True, useScaler=True, useSampler=True, usePca=True):\n",
        "    '''\n",
        "    * for each estimator type,\n",
        "      * generate a pipe\n",
        "      * use grid search to decide the best hyper-params\n",
        "    * return a list of dictionaries describing the best models, as they were decided by grid search\n",
        "    '''\n",
        "    models = generate_estimators()\n",
        "    grid_accuracy = dict()\n",
        "    grid_f1 = dict()\n",
        "    best_estimators = []\n",
        "    for modell in models:\n",
        "        name, model, hyper_params_name, hyper_params_values = modell.values()\n",
        "        \n",
        "        steps = []\n",
        "        hyperparams = dict()\n",
        "\n",
        "        if useSelector:\n",
        "            selector = VarianceThreshold()\n",
        "            steps.append(('selector', selector))\n",
        "            hyperparams['selector__threshold'] = [0.10, 0.14, 0.18]\n",
        "        if useScaler:\n",
        "            scaler = StandardScaler()\n",
        "            steps.append(('scaler', scaler))\n",
        "        if useSampler:\n",
        "            ros = RandomOverSampler()\n",
        "            steps.append(('ros', ros))\n",
        "        if usePca:\n",
        "            pca = PCA()\n",
        "            steps.append(('pca', pca))\n",
        "            hyperparams['pca__n_components'] = [i for i in range(1, 3)]\n",
        "        \n",
        "        steps.append((name, model))\n",
        "\n",
        "        if hyper_params_name:\n",
        "            hyperparams[hyper_params_name] = hyper_params_values\n",
        "\n",
        "        pipe = Pipeline(\n",
        "            steps = steps,\n",
        "            memory='tmp',\n",
        "        )\n",
        "                \n",
        "        estimator = GridSearchCV(\n",
        "            pipe,\n",
        "            hyperparams,\n",
        "            cv=10,\n",
        "            scoring=\"f1_macro\"\n",
        "        )\n",
        "                \n",
        "        estimator.fit(train_grid, train_labels_grid.ravel())\n",
        "        best_estimators.append({\n",
        "            'name': name,\n",
        "            'model': estimator.best_estimator_,\n",
        "            'params': estimator.best_params_,\n",
        "        })\n",
        "    return best_estimators\n",
        "\n",
        "\n",
        "def reTrainEstimator(name, params, useSelector=True, useScaler=True, useSampler=True, usePca=True):\n",
        "    '''\n",
        "    * take as input the name and the params of a model\n",
        "    * this model is one of best, as decided by grid search\n",
        "    * generate a new NON-FITTED model with these params\n",
        "    * re-train it on the while dataset\n",
        "    '''\n",
        "    steps = []\n",
        "\n",
        "    # re-define the tranformers\n",
        "    if useSelector:\n",
        "        threshold = params[\"selector__threshold\"] if params.get(\"selector__threshold\") else 0\n",
        "        selector = VarianceThreshold(threshold=threshold)\n",
        "        steps.append(('selector', selector))\n",
        "\n",
        "    if useScaler:\n",
        "        steps.append(('scaler', StandardScaler()))\n",
        "\n",
        "    if useSampler:\n",
        "        steps.append(('sampler', RandomOverSampler()))\n",
        "\n",
        "    if usePca:\n",
        "        n_components = params[\"pca__n_components\"] if params.get(\"pca__n_components\") else None\n",
        "        pca = PCA(n_components)\n",
        "        steps.append(('pca', pca))\n",
        "\n",
        "    # re-define the estimator\n",
        "    if name==\"dummy\":\n",
        "        strategy = params[\"dummy_strategy\"] if params.get(\"dummy_strategy\") else \"prior\"\n",
        "        model = DummyClassifier(strategy=strategy)\n",
        "\n",
        "    elif name==\"bayes\":\n",
        "        model = GaussianNB()\n",
        "    \n",
        "    elif name==\"knn\":\n",
        "        k = params[\"knn__n_neighbors\"] if params.get(\"knn__n_neighbours\") else 5\n",
        "        model = KNeighborsClassifier(n_neighbors=k)\n",
        "\n",
        "    elif name==\"log_regr\":\n",
        "        model = LogisticRegression()\n",
        "        name = \"log_regr\"\n",
        "  \n",
        "    steps.append((name, model))\n",
        "\n",
        "    # generate and fit the pipe on the whole dataset\n",
        "    pipe = Pipeline(steps=steps)\n",
        "\n",
        "    tic = time.time()\n",
        "    pipe.fit(train_grid, train_labels_grid.ravel())\n",
        "    tac = time.time()\n",
        "    train_time = round((tac-tic) * 1000, 3)\n",
        "\n",
        "    return pipe, train_time\n",
        "\n",
        "def compute_best_models_test_scores(best_estimator, useSelector=True, useScaler=True, useSampler=True, usePca=True):\n",
        "    '''\n",
        "    * receives a list with the best model of each classifier type\n",
        "    * for each of these models:\n",
        "        * uses `reTrainEstimator` to build a new model with the same parameters\n",
        "        * fits it on the while training set, tests it on the test set\n",
        "        * returns some dicts with accuracy, f1, train time, test time, predictions\n",
        "    '''\n",
        "    grid_accuracy = dict()\n",
        "    grid_f1 = dict()\n",
        "    grid_train_time = dict()\n",
        "    grid_test_time = dict()\n",
        "    grid_preds = dict()\n",
        "    for est in best_estimator:\n",
        "        name, model, params = est.values()\n",
        "        trained_model, train_time = reTrainEstimator(name, params, useSelector=useSelector, useScaler=useScaler, useSampler=useSampler, usePca=usePca)\n",
        "        tic = time.time()\n",
        "        preds = trained_model.predict(test_grid)\n",
        "        tac = time.time()\n",
        "        grid_preds[name] = preds\n",
        "        grid_train_time[name] = train_time\n",
        "        grid_test_time[name] = round((tac-tic) * 1000, 3)\n",
        "        grid_accuracy[name] = round(accuracy_score(test_labels_grid, preds), 3)\n",
        "        grid_f1[name] = round(f1_score(test_labels_grid, preds), 3)\n",
        "    return grid_accuracy, grid_f1, grid_train_time, grid_test_time, grid_preds\n",
        "\n",
        "def show_best_models_scores(best_estimators, useSelector=True, useScaler=True, useSampler=True, usePca=True, fig_index=50, title_prefix=\"\", color=\"blue\", logging=False, max_acc=None, min_acc=None, max_avg=None):\n",
        "    '''\n",
        "    * uses `compute_best_models_test_scores` to compute the test scores\n",
        "    * returns:\n",
        "        * a dict with the predictions\n",
        "        * three objects for the estimators with\n",
        "            * the max accuracy till now\n",
        "            * the lowest accuracy till now\n",
        "            * the max accuracy-f1 average till now\n",
        "    '''\n",
        "    grid_accuracy, grid_f1, grid_train_time, grid_test_time, grid_preds = compute_best_models_test_scores(\n",
        "        best_estimators,\n",
        "        useSelector=useSelector,\n",
        "        useScaler=useScaler,\n",
        "        useSampler=useSampler,\n",
        "        usePca=usePca\n",
        "    )\n",
        "    # print(grid_accuracy)\n",
        "    # print(grid_f1)\n",
        "    if logging:\n",
        "        print(\"Best models scores on the testing set:\")\n",
        "        best_estimators_trimmed = [{'name': est['name'], 'model': est['model']} for est in best_estimators]\n",
        "        print_scores_table(best_estimators_trimmed, grid_accuracy, grid_f1, grid_train_time, grid_test_time)\n",
        "    #print('---------------------------')\n",
        "                   \n",
        "    print(f\"{'selector-' if useSelector else ''}{'scaler-' if useScaler else ''}{'sampler-' if useSampler else ''}{'pca-' if usePca else ''}:\")\n",
        "    print('\\ttrain times:', grid_train_time)\n",
        "    print('\\ttest times:', grid_test_time)\n",
        "    plot_scores(grid_accuracy, grid_f1, fig_index=fig_index, title_prefix=title_prefix, color=color)\n",
        "    \n",
        "    #print(\"I am show_best_models_scores, testing accuracy, with curr:\", max_acc)\n",
        "    \n",
        "    local_max_name, local_max = None, None\n",
        "    local_min_name, local_min = None, None\n",
        "    local_max_avg_name, local_max_avg = None, None\n",
        "    if max_acc is not None:\n",
        "        for name in grid_accuracy:\n",
        "            if name=='dummy':\n",
        "                continue\n",
        "            if grid_accuracy[name] > max_acc:\n",
        "                local_max_acc = grid_accuracy[name]\n",
        "                local_max_name = name\n",
        "            if grid_accuracy[name] < min_acc:\n",
        "                local_min_acc = grid_accuracy[name]\n",
        "                local_min_name = name\n",
        "            if (grid_accuracy[name]+grid_f1[name])/2 > max_avg:\n",
        "                local_max_avg = (grid_accuracy[name]+grid_f1[name])/2\n",
        "                local_max_avg_name = name\n",
        "        if local_max_name:\n",
        "            local_max = {\n",
        "                'name': f\"{'selector-' if useSelector else ''}{'scaler-' if useScaler else ''}{'sampler-' if useSampler else ''}{'pca-' if usePca else ''}{local_max_name}\",\n",
        "                'accuracy': grid_accuracy[local_max_name],\n",
        "                'f1': grid_f1[local_max_name],\n",
        "                'predictions': grid_preds[local_max_name]\n",
        "            }\n",
        "        if local_min_name:\n",
        "            local_min = {\n",
        "                'name': f\"{'selector-' if useSelector else ''}{'scaler-' if useScaler else ''}{'sampler-' if useSampler else ''}{'pca-' if usePca else ''}{local_min_name}\",\n",
        "                'accuracy': grid_accuracy[local_min_name],\n",
        "                'f1': grid_f1[local_min_name],\n",
        "                'predictions': grid_preds[local_min_name]  \n",
        "            }\n",
        "        if local_max_avg_name:\n",
        "            local_max_avg = {\n",
        "                'name': f\"{'selector-' if useSelector else ''}{'scaler-' if useScaler else ''}{'sampler-' if useSampler else ''}{'pca-' if usePca else ''}{local_max_avg_name}\",\n",
        "                'accuracy': grid_accuracy[local_max_avg_name],\n",
        "                'f1': grid_f1[local_max_avg_name],\n",
        "                'predictions': grid_preds[local_max_avg_name],\n",
        "                'avg': (grid_accuracy[local_max_avg_name]+grid_f1[local_max_avg_name])/2,\n",
        "            }\n",
        "    return grid_preds, (local_max or None), (local_min or None), (local_max_avg or None)\n"
      ],
      "id": "a22cd245-8aeb-477d-acf8-490ceddfe094",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e41dc2fd-abc0-4e65-9a07-246cf478073c"
      },
      "source": [
        "##### Δοκιμές πολλών αρχιτεκτονικών"
      ],
      "id": "e41dc2fd-abc0-4e65-9a07-246cf478073c"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "862ec3db-8494-4d0a-8baf-b5b0a0d8b924"
      },
      "source": [
        "* Για κάθε αρχιτεκτονική:\n",
        "    * Χτίζω τα διάφορα μοντέλα\n",
        "    * για το καλύτερο μοντέλο για κάθε ταξινομητή\n",
        "        * το εκπαιδεύω στο σύνολο του train set\n",
        "        * σχεδιάζω γραφική με τις επιδόσεις του στο test set\n",
        "        * καταγράφω τους χρόνους του στο train και test set\n",
        "* Από όλα τα παραγόμενα μοντέλα, κρατάω τα βασικά στοιχεία των εξής:\n",
        "    * του μοντέλου που έχει το μέγιστο μέχρι τώρα accuracy (ζητείται από την εκφώνηση)\n",
        "    * του μοντέλου που έχει το ελάχιστο μέχρι τώρα accuracy (ζητείται από την εκφώνηση)\n",
        "    * του μοντέλου που έχει τον μέγιστο μέσο όρο accuracy και f1-score (θα το προτείνω στο τέλος)\n",
        "\n",
        "* Κάθε δυάδα (γραμμή) διαγραμμάτων που ακολουθεί, αφορά και μία διαφορετική αρχιτεκτονική"
      ],
      "id": "862ec3db-8494-4d0a-8baf-b5b0a0d8b924"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "0f228561-8375-44a5-826d-42403e70703d",
        "outputId": "abdc852d-db51-4e62-eecb-01f5a2e00650"
      },
      "source": [
        "max_accuracy = {\n",
        "    'name': None,\n",
        "    'predictions': None,\n",
        "    'accuracy': 0.0,\n",
        "}\n",
        "min_accuracy = {\n",
        "    'name': None,\n",
        "    'predictions': None,\n",
        "    'accuracy': 1.0,\n",
        "}\n",
        "\n",
        "max_average = {\n",
        "    'name': None,\n",
        "    'predictions': None,\n",
        "    'avg': 0.0,\n",
        "}\n",
        "\n",
        "c=0\n",
        "print(\"Generating all models... please wait...\")\n",
        "# for each architecture\n",
        "for useSelector in [True, False]:\n",
        "    for useScaler in [True, False]:\n",
        "        for useSampler in [True, False]:\n",
        "            for usePca in [True, False]:\n",
        "                # find the 4 best estimators\n",
        "                best_estimators = pick_best_models(useSelector=useSelector, useScaler=useScaler, useSampler=useSampler, usePca=usePca)\n",
        "                title_prefix = f\" with {'selector' if useSelector else ''} {'scaler' if useScaler else ''} {'sampler' if useSampler else ''} {'pca' if usePca else ''} - \"\n",
        "                # plot their accuracy and f1 bar charts\n",
        "                predictions,  max_accuracy_object, min_accuracy_object, max_average_object = show_best_models_scores(\n",
        "                    best_estimators,\n",
        "                    useSelector=useSelector,\n",
        "                    useScaler=useScaler,\n",
        "                    useSampler=useSampler,\n",
        "                    usePca=usePca,\n",
        "                    fig_index=100+c,\n",
        "                    title_prefix=title_prefix,\n",
        "                    max_acc=max_accuracy['accuracy'],\n",
        "                    min_acc=min_accuracy['accuracy'],\n",
        "                    max_avg=max_average['avg']\n",
        "                )\n",
        "                # update the max/min-accuracy and max-average estimators\n",
        "                if max_accuracy_object is not None:\n",
        "                    max_accuracy = max_accuracy_object\n",
        "                if min_accuracy_object is not None:\n",
        "                    min_accuracy = min_accuracy_object\n",
        "                if max_average_object is not None:\n",
        "                    max_average = max_average_object\n",
        "                c += 1\n",
        "# print the results\n",
        "print()\n",
        "print(f\"Max accuracy: {max_accuracy['accuracy']}, by {max_accuracy['name']}\")\n",
        "print(f\"Lowest accuracy: {min_accuracy['accuracy']}, by {min_accuracy['name']}\")\n",
        "print(f\"Max average (accuracy, f1): {round(max_average['avg'], 3)}, by {max_average['name']}\")\n",
        "\n",
        "# plot the bar chart for the outof-the-box models, to compare\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, fig_index=100+c, title_prefix=\"Out of the box\", color=\"green\")"
      ],
      "id": "0f228561-8375-44a5-826d-42403e70703d",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Generating all models... please wait...\n",
            "selector-scaler-sampler-pca-:\n",
            "\ttrain times: {'dummy': 2.338, 'bayes': 2.739, 'knn': 2.659, 'log_regr': 4.291}\n",
            "\ttest times: {'dummy': 0.423, 'bayes': 0.542, 'knn': 3.186, 'log_regr': 0.489}\n",
            "selector-scaler-sampler-:\n",
            "\ttrain times: {'dummy': 1.876, 'bayes': 2.208, 'knn': 2.123, 'log_regr': 7.298}\n",
            "\ttest times: {'dummy': 0.314, 'bayes': 0.471, 'knn': 3.434, 'log_regr': 0.428}\n",
            "selector-scaler-pca-:\n",
            "\ttrain times: {'dummy': 1.665, 'bayes': 1.973, 'knn': 2.029, 'log_regr': 3.876}\n",
            "\ttest times: {'dummy': 0.421, 'bayes': 0.621, 'knn': 3.268, 'log_regr': 0.525}\n",
            "selector-scaler-:\n",
            "\ttrain times: {'dummy': 1.254, 'bayes': 1.586, 'knn': 1.69, 'log_regr': 6.693}\n",
            "\ttest times: {'dummy': 0.293, 'bayes': 0.475, 'knn': 3.232, 'log_regr': 0.455}\n",
            "selector-sampler-pca-:\n",
            "\ttrain times: {'dummy': 2.014, 'bayes': 2.156, 'knn': 2.073, 'log_regr': 4.474}\n",
            "\ttest times: {'dummy': 0.335, 'bayes': 0.459, 'knn': 3.196, 'log_regr': 0.394}\n",
            "selector-sampler-:\n",
            "\ttrain times: {'dummy': 1.333, 'bayes': 1.697, 'knn': 1.671, 'log_regr': 6.33}\n",
            "\ttest times: {'dummy': 0.209, 'bayes': 0.355, 'knn': 3.105, 'log_regr': 0.288}\n",
            "selector-pca-:\n",
            "\ttrain times: {'dummy': 1.286, 'bayes': 1.632, 'knn': 1.481, 'log_regr': 3.297}\n",
            "\ttest times: {'dummy': 0.321, 'bayes': 0.449, 'knn': 3.198, 'log_regr': 0.364}\n",
            "selector-:\n",
            "\ttrain times: {'dummy': 0.785, 'bayes': 1.185, 'knn': 1.144, 'log_regr': 6.121}\n",
            "\ttest times: {'dummy': 0.182, 'bayes': 0.358, 'knn': 3.171, 'log_regr': 0.274}\n",
            "scaler-sampler-pca-:\n",
            "\ttrain times: {'dummy': 1.997, 'bayes': 2.416, 'knn': 2.373, 'log_regr': 4.214}\n",
            "\ttest times: {'dummy': 0.376, 'bayes': 0.468, 'knn': 3.203, 'log_regr': 0.418}\n",
            "scaler-sampler-:\n",
            "\ttrain times: {'dummy': 1.565, 'bayes': 1.876, 'knn': 1.827, 'log_regr': 9.019}\n",
            "\ttest times: {'dummy': 0.227, 'bayes': 0.432, 'knn': 3.335, 'log_regr': 0.296}\n",
            "scaler-pca-:\n",
            "\ttrain times: {'dummy': 1.338, 'bayes': 1.688, 'knn': 1.703, 'log_regr': 3.766}\n",
            "\ttest times: {'dummy': 0.335, 'bayes': 0.465, 'knn': 3.203, 'log_regr': 0.411}\n",
            "scaler-:\n",
            "\ttrain times: {'dummy': 0.913, 'bayes': 1.323, 'knn': 1.306, 'log_regr': 7.106}\n",
            "\ttest times: {'dummy': 0.233, 'bayes': 0.373, 'knn': 3.254, 'log_regr': 0.334}\n",
            "sampler-pca-:\n",
            "\ttrain times: {'dummy': 1.593, 'bayes': 1.886, 'knn': 1.85, 'log_regr': 3.497}\n",
            "\ttest times: {'dummy': 0.255, 'bayes': 0.361, 'knn': 3.59, 'log_regr': 0.282}\n",
            "sampler-:\n",
            "\ttrain times: {'dummy': 0.965, 'bayes': 1.344, 'knn': 1.307, 'log_regr': 6.806}\n",
            "\ttest times: {'dummy': 0.095, 'bayes': 0.288, 'knn': 3.381, 'log_regr': 0.161}\n",
            "pca-:\n",
            "\ttrain times: {'dummy': 0.835, 'bayes': 1.307, 'knn': 1.202, 'log_regr': 2.824}\n",
            "\ttest times: {'dummy': 0.229, 'bayes': 0.331, 'knn': 2.943, 'log_regr': 0.254}\n",
            ":\n",
            "\ttrain times: {'dummy': 0.337, 'bayes': 0.785, 'knn': 1.235, 'log_regr': 5.769}\n",
            "\ttest times: {'dummy': 0.106, 'bayes': 0.283, 'knn': 3.08, 'log_regr': 0.161}\n",
            "\n",
            "Max accuracy: 0.691, by selector-scaler-pca-log_regr\n",
            "Lowest accuracy: 0.345, by scaler-sampler-bayes\n",
            "Max average (accuracy, f1): 0.466, by scaler-sampler-pca-knn\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABIEAAADTCAYAAAAWPwMuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAf+0lEQVR4nO3de7gddX3v8feHBJQKXtm1CgmhGLURW9CAtd6oRQUv0FOhBi+ItVJ7ytHW6pHWllKq5ynSlh4rekRFW1ER8WhTjU1p5XK8oAk3aULRGKMJgob7TS6R7/ljZsNis3f22nuvnZXseb+eZz2Zy29mvjOz1l7ffOc3s1JVSJIkSZIkaW7badgBSJIkSZIkafZZBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA6wCKROS7Iwye1J5m2lTSV50raMayJJDk6yadhxDFKSY5N8ddhxzAX9vJ8lSdoemZMNnzmZ1A0WgdRpVfXDqtqtqn4GkOSCJL877Lg0fEk+nmRLkicMO5Z+jX0/S5K0ozAn01jte+Cutjg4+np2O++vklzZ5monDTlUaYdiEUjqiCTzhx3DoM1Wj5ckjwBeCdwCvHY2trGVbc+58yRJkh4wF7/rZ7EX8vFtcXD09Y12+jrgfwJfmqXt9m0unk/NbRaBNCcl+csk/9AO75zkjiSntuO7tlcVHptkUdu1eH6S9wDPA97fXml4f88qD0ny3SQ3Jzk9SfqM46VJ1ia5Lck1Sd7eM++IJJcnuTXJ95Ic2k5/Q5Kr2mXWJ/m9raz/iUk+l2Rzku8neUvPvJOSnJvkrCS3Asf2G1+SxyT5Yrvem9rhvXqWuyDJu5N8vT1W/5LkcUk+2e7PqiSLetpXkre0+3N9klOTjPv3J8lTk5yX5MYkVyf57Z55H0/ywSQrktwB/Ho/52EaXgncDJwMvH5MfI9N8rEkP2qPzRd65k10TjckOaSn3UlJzmqHR9+Db0zyQ+Ar7fTPJrkuyS1JLkrytJ7ld03yt0l+0M7/ajvt/vdz2+5RST6a5Nr2/L57NElL8qQkF7bLX5/kM7N0LCVJHWZOZk42G6rqH6vqy8Btk7VN47QkP2mPyZVJ9mvnjZtTtfMOT7Kmfa9dkOSXeta5Ick7k3wbuKN93/5qex5uTnJFkoN72h/bHvPb2vfHawZ+UKR+VZUvX3PuBbwQuLId/jXge8A3e+Zd0Q4vAgqY345fAPzumHUV8EXg0cBCYDNwaJ9xXAs8rx1+DPCMdvggml4mL6Ipxu4JPLWd9zJgXyDAC4A7e5Y7GNjUDu8EXAKcCOwC/CKwHnhJO/8k4F7gN9u2u04hvsfRFEJ+Dtgd+CzwhZ7lLqC5ArMv8ChgLfAd4BBgPvBPwMfGHMPzgce2x/A7o8eZJhH6ajv8CGAj8IZ2PQcA1wNL2vkfb4/bc9p9evgsvX/+A3gv8HhgC/DMnnlfAj7THq+dgRf0cU43AIf0rOMk4Kwx78F/avd/13b677TH/mHA3wOX9yx/ensO9gTm0bzHH8ZD38+fBz7UrvfngW8Bv9fO+zTwrtHjCDx32J9bX758+fI1916Yk41+75uTTf2985D3wDhtzgJOmqTNS9rz8+j2XP4S8IR23kQ51ZOBO9r3xc40vY7WAbu0y20ALgcWALu2y98AvLQ9Hi9qx0faY3kr8JR22ScATxv2Z9NXd1/2BNJc9Q1gcZLHAc8HPgrsmWQ3mi/xC6e4vr+uqpur6oc0X5z797ncvcCSJI+sqpuq6tJ2+huBM6vqvKq6r6quqar/AqiqL1XV96pxIfBvNFfDxjoQGKmqk6vqnqpaD3wYWNbT5htV9YV2Gz/tN76quqGqPldVd1bVbcB7aI5br4+1cd4CfBn4XlX9e1VtoUlQDhjT/pSqurE9hn8PHD1OPC8HNlTVx6pqS1VdBnwOOKqnzT9X1dfafbprnHXMSJKFNFezPlVVP6YpCB3TznsCcBjw5vZ43dueI9jKOe3TSVV1x+h5qqozq+q2qrqbJnn8lTQ9e3aiKRC9td3Gz6rq62273v14PE0i8often8CnMYD7497gb2BJ1bVXVXlgyAlSbPBnKxhTjY972t71tyc5NLJm4/rXpoC2lOBVNVVVXXtJDnVq4Avte+Le4G/oSn2/FpvbFW1sT2frwVWVNWK9nicB6ymycUA7gP2S7JrVV1bVWumuS/SjFkE0pzU/jFeTfMl+XyaBOPrNFcrppNwXNczfCewW5/LvZLmj/8P0tx68+x2+gKaK2EPkeSwJBe3XW9vbpffY5ymewNP7PlivBn4U5reK6M2Tie+JD+X5ENt19hbgYuAR+fB93v/uGf4p+OMjz1GvbH8AHjiBPv0rDH79BrgF6awT/dL8po88CDBL/e52OuAq6rq8nb8k8Crk+xMc95urKqbxlluwnPap/v3K8m8JH+dpkv6rTRXm6B5H+xB03Nnsm3tTXPl6tqeY/khmh5B0FzRCvCttqvz78wgdkmSxmVOdj9zsqnnZABvqapHt69n9LmtNT3bel5VfQV4P02vn58kOSPJI9l6TvVEmmMDQFXdR7O/e/a06d3/vYGjxhyv59L0OLqDpqj0Zpq87EtJntrvAZAGzYdYaS67kKab8QHAqnb8JTTdfi+aYJkaZABVtQo4oi0gHA+cQ5NsbKTptvsgSR5Gc5XlGJqrK/emeebMePe7bwS+X1WLtxbCNOP7Y+ApwLOq6rok+wOXTRBHvxYAo1c9FgI/GqfNRuDCqnrR1sLud4NV9UmaIs5UHAMsTDKaZM6n6Yr9UprbqR6b5NFVdfOY5cY9p607aLpxj/qFcdr07tergSNounJvoOnefRPN8b8euKvd1hVb2Y+NwN3AHu2VwAdvrOo64E0ASZ4L/HuSi6pq3VbWKUnSdJiTmZNNJyeblqp62jjT3kfTq+jnaY7tO4C/YOKc6kfA00dHkoTmuF3Tu9qe4Y3AJ6rqTRPEtBJYmeZ5Q++m6Sk2Xq8yadbZE0hz2YU0X9xrq+oe2vuKab6kN0+wzI9p7uOesSS7tFc9HtV2I72VpisoNF2h35DkN5LslGTP9orALjT3IW8GtiQ5DHjxBJv4FnBbmofS7dr2HtkvyYEDiG93mitHNyd5LM2X5Ey9I83DDRcAb6V5rs5YXwSenOR1aR4euXOSA9PzIL7Z1F5125cmKd2/fe0HfAo4pqqupelm/YF2X3ZO8vx28YnOKTT3jC9r2y8FjpwklN1pCjg30BSP/tfojPZK1JnA36V5COW8JM9uk1V62l1L0239b5M8so1p3yQvaPf1qDzwYMmbaBKZ+5AkafDMyaYfXydzssm08Tyc5v+z85M8PBP8Qlkb97PaAtsdNIWf+ybJqc4BXta+L3amKcbdTdOLbTxnAa9I8pJ2PQ9PcnCSvZI8Ps3Dxx/RruN2zLk0RBaBNJd9nebe3dErTGtp/uhPdMUJ4H8DR6b59YX3DSCG1wEb0nTffTNNN1qq6ls0D9o7jeahehcCe7f3er+F5ovnJpoeIcvHW3FV/Yzmfu39ge/T9BD5CE2vkRnFR3N/+K7tOi8G/nUK65zIP9M8lO9ymocrf3Rsg3b/X0xzD/2PaLp8n0KThG0Lr6e52ndlVV03+qJ5X7y8Tb5eR3Nv+X8BPwH+sI193HParvfPaYpLNwF/SVNU2pp/oumCfA3N+/biMfPfDlxJczX1RppjNN7f82Nokti17bbPpXkYITTPL/hmkttp3mNvbZ9hIEnSoJmTTTM+upuTTebDNMWxo2l+6OKnNMdwPI9s299Ek1/dAJzazhs3p6qqq2me8/MPNMf+FcAr2iLmQ1TVRppe3H9KUzjcSNPbaKf29Taa43gjzW2Qvz+93ZZmLlUD7WkpSQ+RpIDF3mokSZI0POZkkuwJJEmSJEmS1AEWgSRJkiRJkjrA28EkSZIkSZI6wJ5AkiRJkiRJHWARSJIkSZIkqQPmD2vDe+yxRy1atGhYm5ckSbPskksuub6qRoYdhx7MHEySpLltaznY0IpAixYtYvXq1cPavCRJmmVJfjDsGPRQ5mCSJM1tW8vB+rodLMmhSa5Osi7JCePMPy3J5e3rO0lunknAkiRJkiRJGqxJewIlmQecDrwI2ASsSrK8qtaOtqmqP+pp/z+AA2YhVkmSJEmSJE1TPz2BDgLWVdX6qroHOBs4YivtjwY+PYjgJEmSJEmSNBj9FIH2BDb2jG9qpz1Ekr2BfYCvzDw0SZIkSZIkDcqgfyJ+GXBuVf1svJlJjkuyOsnqzZs3D3jTkiRJGo85mCRJgv6KQNcAC3rG92qnjWcZW7kVrKrOqKqlVbV0ZMRfjJUkSdoWzMEkSRL09xPxq4DFSfahKf4sA149tlGSpwKPAb4x0AinIRl2BKoadgSSJEmSJKnXpD2BqmoLcDywErgKOKeq1iQ5OcnhPU2XAWdX+d9/SZIkSZKk7U0/PYGoqhXAijHTThwzftLgwpIkSZIkSdIgDfrB0JIkSZIkSdoOWQSSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdUBfRaAkhya5Osm6JCdM0Oa3k6xNsibJpwYbpiRJkiRJkmZi/mQNkswDTgdeBGwCViVZXlVre9osBv4EeE5V3ZTk52crYEmSJEmSJE1dPz2BDgLWVdX6qroHOBs4YkybNwGnV9VNAFX1k8GGKUmSJEmSpJnopwi0J7CxZ3xTO63Xk4EnJ/lakouTHDqoACVJkiRJkjRzk94ONoX1LAYOBvYCLkry9Kq6ubdRkuOA4wAWLlw4oE1LkiRpa8zBJEkS9NcT6BpgQc/4Xu20XpuA5VV1b1V9H/gOTVHoQarqjKpaWlVLR0ZGphuzJEmSpsAcTJIkQX9FoFXA4iT7JNkFWAYsH9PmCzS9gEiyB83tYesHGKckSZIkSZJmYNIiUFVtAY4HVgJXAedU1ZokJyc5vG22ErghyVrgfOAdVXXDbAUtSZIkSZKkqenrmUBVtQJYMWbaiT3DBbytfUmSJEmSJGk708/tYJIkSZIkSdrBWQSSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA7oqwiU5NAkVydZl+SEceYfm2Rzksvb1+8OPlRJkiRJkiRN1/zJGiSZB5wOvAjYBKxKsryq1o5p+pmqOn4WYpQkSZIkSdIM9dMT6CBgXVWtr6p7gLOBI2Y3LEmSJEmSJA1SP0WgPYGNPeOb2mljvTLJt5Ocm2TBQKKTJEmSJEnSQAzqwdD/Aiyqql8GzgP+cbxGSY5LsjrJ6s2bNw9o05IkSdoaczBJkgT9FYGuAXp79uzVTrtfVd1QVXe3ox8BnjneiqrqjKpaWlVLR0ZGphOvJEmSpsgcTJIkQX9FoFXA4iT7JNkFWAYs722Q5Ak9o4cDVw0uREmSJEmSJM3UpL8OVlVbkhwPrATmAWdW1ZokJwOrq2o58JYkhwNbgBuBY2cxZkmSJEmSJE3RpEUggKpaAawYM+3EnuE/Af5ksKFJkiRJkiRpUAb1YGhJkiRJkiRtxywCSZIkSZIkdYBFIEmSJEmSpA6wCCRJkiRJktQBFoEkSZIkSZI6wCKQJEmSJElSB1gEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjqgryJQkkOTXJ1kXZITttLulUkqydLBhShJkiRJkqSZmrQIlGQecDpwGLAEODrJknHa7Q68FfjmoIOUJEmSJEnSzPTTE+ggYF1Vra+qe4CzgSPGafdXwCnAXQOMT5IkSZIkSQPQTxFoT2Bjz/imdtr9kjwDWFBVXxpgbJIkSZIkSRqQGT8YOslOwN8Bf9xH2+OSrE6yevPmzTPdtCRJkvpgDiZJkqC/ItA1wIKe8b3aaaN2B/YDLkiyAfhVYPl4D4euqjOqamlVLR0ZGZl+1JIkSeqbOZgkSYL+ikCrgMVJ9kmyC7AMWD46s6puqao9qmpRVS0CLgYOr6rVsxKxJEmSJEmSpmzSIlBVbQGOB1YCVwHnVNWaJCcnOXy2A5QkSZIkSdLMze+nUVWtAFaMmXbiBG0PnnlYkgTJsCNQ1eyu33M8fLN9jiVJkrT9mPGDoSVJkiRJkrT9swgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpA+YPOwBpOpJhRyCAqmFHIEmSJEnqlz2BJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpA/x1MEmSJEnS/fwl3uHzV3g1W/rqCZTk0CRXJ1mX5IRx5r85yZVJLk/y1SRLBh+qJEmSJEmSpmvSIlCSecDpwGHAEuDocYo8n6qqp1fV/sB7gb8beKSSJEmSJEmatn56Ah0ErKuq9VV1D3A2cERvg6q6tWf0EYCd1yRJkiRJkrYj/TwTaE9gY8/4JuBZYxsl+QPgbcAuwAvHW1GS44DjABYuXDjVWCVJkjQN5mCSJAkG+OtgVXV6Ve0LvBP4swnanFFVS6tq6cjIyKA2LUmSpK0wB5MkSdBfT6BrgAU943u10yZyNvDBmQQlSZIkSZJmh78AN3zD+gW4fnoCrQIWJ9knyS7AMmB5b4Mki3tGXwZ8d3AhSpIkSZIkaaYm7QlUVVuSHA+sBOYBZ1bVmiQnA6urajlwfJJDgHuBm4DXz2bQkiRJkiRJmpp+bgejqlYAK8ZMO7Fn+K0DjkuSJEmSJEkDNLAHQ0uSJEmSJGn7ZRFIkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA6wCCRJkiRJktQBFoEkSZIkSZI6wCKQJEmSJElSB1gEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AF9FYGSHJrk6iTrkpwwzvy3JVmb5NtJ/iPJ3oMPVZIkSZIkSdM1aREoyTzgdOAwYAlwdJIlY5pdBiytql8GzgXeO+hAJUmSJEmSNH399AQ6CFhXVeur6h7gbOCI3gZVdX5V3dmOXgzsNdgwJUmSJEmSNBP9FIH2BDb2jG9qp03kjcCXZxKUJEmSJEmSBmv+IFeW5LXAUuAFE8w/DjgOYOHChYPctCRJkiZgDiZJkqC/nkDXAAt6xvdqpz1IkkOAdwGHV9Xd462oqs6oqqVVtXRkZGQ68UqSJGmKzMEkSRL0VwRaBSxOsk+SXYBlwPLeBkkOAD5EUwD6yeDDlCRJkiRJ0kxMWgSqqi3A8cBK4CrgnKpak+TkJIe3zU4FdgM+m+TyJMsnWJ0kSZIkSZKGoK9nAlXVCmDFmGkn9gwfMuC4JEmSJG1nkmFHoKphRyBpR9bP7WCSJEmSJEnawVkEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQO6KsIlOTQJFcnWZfkhHHmPz/JpUm2JDly8GFKkiRJkiRpJiYtAiWZB5wOHAYsAY5OsmRMsx8CxwKfGnSAkiRJkiRJmrn5fbQ5CFhXVesBkpwNHAGsHW1QVRvaeffNQoySJEmSJEmaoX5uB9sT2NgzvqmdNmVJjkuyOsnqzZs3T2cVkiRJmiJzMEmSBNv4wdBVdUZVLa2qpSMjI9ty05IkSZ1lDiZJkqC/ItA1wIKe8b3aaZIkSZIkSdpB9FMEWgUsTrJPkl2AZcDy2Q1LkiRJkiRJgzRpEaiqtgDHAyuBq4BzqmpNkpOTHA6Q5MAkm4CjgA8lWTObQUuSJEmSJGlq+vl1MKpqBbBizLQTe4ZX0dwmJkmSJEmSpO3QNn0wtCRJkiRJkobDIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpA/oqAiU5NMnVSdYlOWGc+Q9L8pl2/jeTLBp0oJIkSZIkSZq+SYtASeYBpwOHAUuAo5MsGdPsjcBNVfUk4DTglEEHKkmSJEmSpOnrpyfQQcC6qlpfVfcAZwNHjGlzBPCP7fC5wG8kyeDClCRJkiRJ0kz0UwTaE9jYM76pnTZum6raAtwCPG4QAUqSJEmSJGnm5m/LjSU5DjiuHb09ydXbcvs7mD2A64cdxHTZD6wvO/Q5Bs9zHzzH3bBDn+dZPsd7z+ra1TdzsL7t0J9n9W2HPs9+N/dlhz7H4Hnug+d46ybMwfopAl0DLOgZ36udNl6bTUnmA48Cbhi7oqo6Azijj212XpLVVbV02HFo9niO5z7PcTd4nrUjMAfrj5/nbvA8z32e47nPczx9/dwOtgpYnGSfJLsAy4DlY9osB17fDh8JfKWqanBhSpIkSZIkaSYm7QlUVVuSHA+sBOYBZ1bVmiQnA6urajnwUeATSdYBN9IUiiRJkiRJkrSd6OuZQFW1AlgxZtqJPcN3AUcNNrTOs8v23Oc5nvs8x93geZbmDj/P3eB5nvs8x3Of53ia4l1bkiRJkiRJc18/zwSSJEmSJEnSDs4i0CxKclKStw87Ds2uJIuS/Oew49C257mXpO2TOVg3+D3cXZ57afosAkmSOiXJ7cOOQZIkqWvMwbYPFoEGLMm7knwnyVeBp7TTLkiytB3eI8mGdvjYJF9Icl6SDUmOT/K2JJcluTjJY3uWPy3J6iRXJTkwyf9N8t0k727bnJzkD3vieE+St27r/e+w+Uk+2Z6fc5P8XJITk6xK8p9Jzkhj3ySXji6UZPHoeJJnJrkwySVJViZ5Qjv9LUnWJvl2krOHtYPauiS/2H5239F+Pv+1/Yy+t6fN7e1n84r2M/74YcaswUsyb4bL9/WDDZIeyhyss8zBOs4cTGAONhUWgQYoyTOBZcD+wEuBA/tYbD/gt9q27wHurKoDgG8Ax/S0u6eqlgL/B/hn4A/aZY9N8jjgzNH2SXZq4zhrALul/jwF+EBV/RJwK/DfgfdX1YFVtR+wK/DyqvoecEuS/dvl3gB8LMnOwD8AR1bVM2nO53vaNicAB1TVLwNv3na7pH4leQrwOeBYYDPN34BXAU8HXpVkQdv0EcDFVfUrwEXAm7Z9tBrV/qfg1PY/CVcmeVU7fackH0jyX+1/EFckOXIr69mQ5JT2PxNHJXlxkm8kuTTJZ5Ps1rZ7abvOS5K8L8kX2+knJflEkq8Bn9gW+y7NNeZgnWYO1mHmYDsmc7Dhsgg0WM8DPl9Vd1bVrcDyPpY5v6puq6rNwC3Av7TTrwQW9bRb3jN9TVVdW1V3A+uBBVW1AbghyQHAi4HLquqGGe+R+rWxqr7WDp8FPBf49STfTHIl8ELgae38jwBvaKvVrwI+RZPA7Aecl+Ry4M+Avdr23wY+meS1wJZtsjeaihGa/xS8pqquaKf9R1XdUlV3AWuBvdvp9wBfbIcv4cGfcW17v0WTLP4KcAhwanv197dozs0S4HXAs/tY1w1V9Qzg32k+v4e046uBtyV5OPAh4LD2PxkjY5Zf0i5z9Iz3Suomc7DuMgfrLnOwHZc52BB1psvTkG3hgYLbw8fMu7tn+L6e8ft48Pm5e5w2Y9t9hKYK/gs0VzG07dQ44x8AllbVxiQn8cC5/xzwF8BXgEuq6oYkT6RJLMf7Q/cy4PnAK4B3JXl6VZmIbD9uAX5Ik3Subaf1fkZ/xgOf0XurqsaZruF4LvDpqvoZ8OMkF9L0CHgu8Nmqug+4Lsn5fazrM+2/v0qTTHwtCcAuNL0Kngqsr6rvt+0+DRzXs/zyqvrpTHdI0kOYg8195mDdZQ624zIHGyJ7Ag3WRcBvJtk1ye40XxgAG4BntsMTdmcbgM8Dh9J8gFbO4nb0UAuTjCYPrwa+2g5f33ZDvP+8t1cmVgIfBD7WTr4aGBldR5Kdkzyt7Va+oKrOB94JPArYbdb3RlNxD/DfgGOSvHrYwWho7mj/DXBeVe3fvpZU1RunsLyk6TEH6y5zsO4yBxOYg02ZRaABqqpLaSqRVwBfBla1s/4G+P0klwF7zOL27wHOB85pq6radq4G/iDJVcBjaJKLDwP/SZNsrBrT/pM0VxD/De4/d0cCpyS5Argc+DVgHnBW2535MuB9VXXz7O+OpqKq7gBeDvwR8Mghh6P+/T+a5wXMSzJCc7X3W8DXgFe296U/Hjh4Cuu8GHhOkicBJHlEkifT/I34xSSL2navGswuSAJzsI4zB+swc7AdljnYEOWBXnHa0bVXLC4Fjqqq7w47Hk0syduBR1XVnw87FqlrktxeVbul6Sv8XuAwmtsH3l1Vn2n/ln6AJvHYSHNl6ZSqOm+C9W2gue3g+nb8hcApwMPaJn9WVcuTvAI4leaK0ypg96p6TXurwu1V9TezssOSZp052I7DHEwaHnOw7YNFoDkiyRKah519vqr+eNjxaGJJPg/sC7xw9A+WpO1Lkt2q6vY0v/zzLeA5VXXdgNYZ4HTgu1V12iDilTQ85mA7DnMwaftnDjb7LAJJkjRGkguAR9M8VPC9VfXxAazzj4DXt+u8DHhTVd050/VKkiTNFeZgs88ikCRJfWivIO8zZvI7q8qHwEqSJM0Sc7DBsggkSZIkSZLUAf46mCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdcD/B5wt9gx3bxaGAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABIEAAADTCAYAAAAWPwMuAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAeJklEQVR4nO3de7gddX3v8feHRIQKosiuVZIQ1Fgb0aIErHdqUcEL9FSswSvWmtpjjlqrR6yWUtQ+RVp5jhrPkSrVekPEo401npRWxHpBExDEhKbGiCYUauQOKiHyPX/MbFhs9k7W3nvtvZLM+/U868ma3/xm5jszWXt/93d+MytVhSRJkiRJkvZsew07AEmSJEmSJM08i0CSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA6wCCRJkiRJktQBFoGkGZZkQZJbk8zZQZ9K8ojZjGsiSY5OsmXYceyJ+vm/IEmSBs98TJIaFoGkGVZVP66q/arqlwBJvpLkD4cd1+4uyUeSbE/ykGHH0q+x/xckSdLsMB8brPb4/aItrI2+ntjOe0eSK9o87bQhhyppDItAkgYmydxZ2s79gBcANwEvnY1t9mx7VvZRkiRpKmYxV1neFtZGX99s2zcC/xP44izFMSHzNuneLAJJU5TkL5O8r31/nyS3JTmznd63vTpyYJKF7fDiuUneBTwVeH97xeT9Pas8Jsn3k9yYZEWS9BnHc5KsT3JLkquTvKln3glJLktyc5IfJDm2bX9lkivbZTYl+aMdrP+hST6bZGuSHyZ5Xc+805Kcn+TjSW4GTp7EIZyOFwA3AqcDrxgT74FJ/j7Jfya5Icnne+ZNdDyuSnJMT7/Tkny8fT96/l6V5MfAl9v2zyS5NslNSb6a5NE9y++b5G+T/Kid/7W27a7/C22/A5J8OMk17bl75+gw9SSPSHJRu/xPk3x6pg6mJEm7K/OxoeZj46qqj1bVl4BbdtY3jbOS/KQ9PlckOaydN24+1c47Psm69jx9Jclv9KzzqiRvSfJd4Lb2nP9Wkm+0/S9PcnRP/5Pb439Le2xfMvCDIu1CLAJJU3cRcHT7/kjgWuBp7fQTgQ1VdX3vAlX1NuDfuPvKyfKe2c9r1/NY4PeBZ/cZx4eBP6qq/YHDuLtIcRTwD8CbgQe0sV3VLvOTdnv3B14JnJXk8WNXnGQv4AvA5cDBwO8Ab0jSG9sJwPntNj7RZ8zT9QrgU8C5wKOSHNEz72PArwCPBn4VOAt2ejz68XTgN7j7vHwJWNRu41Luue9/AxwBPAk4kOZq2J3jrPMjwHbgEcDjgGcBo0PT3wH8M/BAYB7wvknEKklSV5iPNYaRjw3Cs2iOySOBA2iO+XXtvHHzqSSPpMkD3wCMAKuALyTZu2e9JwHPpTkeD6YZlfTOdj1vAj6bZCTN6PL3Ase15+5JwGUztrfSLsAikDR13wQWJXkQzS+vDwMHJ9mPpmBw0STX99dVdWNV/Ri4EDi8z+XuABYnuX9V3VBVl7btrwLOqaoLqurOqrq6qv4doKq+WFU/qMZFNMWGp46z7iOBkao6vaq2VdUm4O+ApT19vllVn2+38fNJ7vOkJVkA/Dbwyar6L+BfgZe38x4CHAe8pj0Wd7T7Bzs4Hn06rapuG93Hqjqnqm6pqtuB04DfbEf27AX8AfD6dhu/rKpvtP169+PBwHOAN7Tr/QlNwWr02N4BHAI8tKp+UVVfm+ShkiSpC8zHGrOaj7Xe246suTHJpTvvPq47gP2BRwGpqiur6pqd5FMvAr7YHtM7aIpF+9IUcO6Krao2t8fipcCqqlrVHp8LgLU0eRg0F+oOS7JvVV1TVeumuC/SbsEikDRF7S+VtTQJxtNokoxvAE9maknHtT3vfwbs1+dyL6D5Jfaj9vahJ7bt84EfjLdAkuOSXJzk+iQ3tssfNE7XQ4CH9vyCvxH4M5orKqM29xnneHG8JHc/TPBLfS72MuDKqhq9SvMJ4MVJ7kOzz9dX1Q3jLDfh8ejTXfuZZE6Sv26HdN/M3Vf0Dmpf+/SxrUOA+wDX9BzbD9KMLILmaleAb7fDnf9gGrFLkrRHMh+7y2znYwCvq6oHtK97jWCaYFvrerb11Kr6MvB+YAXwkyRnJ7k/O86nHgr8aHSiqu6k2f+De/r0Ho9DgBeOOX5PAR5SVbfRFJVeQ5OTfTHJo/o9ANLuyAdlSdNzEfAMmlt51rTTzwaOAr46wTI1yACqag1wQlsEWQ6cR5NwbAYePrZ/kvsCn6UZPfOPVXVHmufmjHfP+2bgh1W1aEchTCP2TzD5IcsvBxYkGU3S5gIPokmcvg0cmOQBVXXjmOXGPR6t22huIRv1a+OF2/P+xTTDro+hKQAdANxAcwx/Cvyi3dblO9iPzcDtwEFVtf1eG6u6Fng1QJKnAP+S5KtVtXEH65QkqYvMx2Y/H5vqth49Ttt7aUYV/SrNcXsz8BdMnE/9J/CY0YkkoTnWV/eutuf9ZuBjVfXqCWJaDaxunzf0TppRVuONyJL2CI4EkqbnIppf3uurahvwFZpnuvywqrZOsMx/AQ8bxMaT7N1evTmgHQ57M3c/e+bDwCuT/E6SvZIc3F7Z2Bu4L7AV2J7kOJr7scfzbeCW9uF6+7YjYA5LcuQg4p+s9qraw2mSusPb12HAJ4GXV9U1NM/q+UCSB6Z5QOTocwEmOh7Q3Pu9tO2/BDhxJ6HsT1PAuY6mePRXozPaq1HnAO9J8xDHOUme2CZ79PS7hmbY998muX8b08OTPL3d1xcmmdd2v4EmmRnvuUKSJHWd+dgupM2n9qH5W3Nukn3SfvHFOH2PTPKEtnh2G03h586d5FPnAc9tj+l9gD+lycu+MUFIHween+TZ7Xr2SXJ0knlJHpzmwd33a9dxK+Zb2sNZBJKm5xs09yCPXmVaT/PLa6KrTgD/CzgxzTdXvXcAMbwMuKq9Lek1wEsAqurbtA8ZpPkq9YuAQ6rqFuB1NL9Ab6AZ1bJyvBVX1S9pHlh4OPBDmlEuH6IZ+TIMr6C5WnZFVV07+qI5ps9LciDN8bgD+HeaBy6+ASY+Hu16/5ymuHQD8Jc0RaUd+QeaYchX05zzi8fMfxNwBc3VyOuBMxj/5+3LaZLA9e22zwce0s47EvhWkltpzs/r22cASJKkezIf27X8HfBzmoczv619/7IJ+t6/7X8DTW51HXBmO2/cfKqqNtA85+d9NMfi+cDz2wLgvVTVZpoR3H9GU3TbTDPaaK/29Uaa0UXX09xC+MdT221p95CqgY6ElCRJkiRJ0i7IkUCSJEmSJEkdYBFIkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA6wCCRJkiRJktQBc4e14YMOOqgWLlw4rM1LkqQZdskll/y0qkaGHYfuyRxMkqQ9245ysKEVgRYuXMjatWuHtXlJkjTDkvxo2DHo3szBJEnas+0oB+vrdrAkxybZkGRjklPGmX9Wksva138kuXE6AUuSJEmSJGmwdjoSKMkcYAXwTGALsCbJyqpaP9qnqv6kp///AB43A7FKkiRJkiRpivoZCXQUsLGqNlXVNuBc4IQd9D8J+NQggpMkSZIkSdJg9FMEOhjY3DO9pW27lySHAIcCX55+aJIkSZIkSRqUQX9F/FLg/Kr65XgzkyxLsjbJ2q1btw5405IkSRqPOZgkSYL+ikBXA/N7pue1beNZyg5uBauqs6tqSVUtGRnxG2MlSZJmgzmYJEmC/r4ifg2wKMmhNMWfpcCLx3ZK8ijggcA3BxrhFCTDjkBVw45AkiRJkiT12ulIoKraDiwHVgNXAudV1bokpyc5vqfrUuDcKv/8lyRJkiRJ2tX0MxKIqloFrBrTduqY6dMGF5YkSZIkSZIGadAPhpYkSZIkSdIuqK+RQJIkzQSf4TZ83sQtSZLUHY4EkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQP6KgIlOTbJhiQbk5wyQZ/fT7I+yboknxxsmJIkSZIkSZqOuTvrkGQOsAJ4JrAFWJNkZVWt7+mzCHgr8OSquiHJr85UwJIkSZIkSZq8fkYCHQVsrKpNVbUNOBc4YUyfVwMrquoGgKr6yWDDlCRJkiRJ0nT0UwQ6GNjcM72lbev1SOCRSb6e5OIkxw4qQEmSJEmSJE3fTm8Hm8R6FgFHA/OAryZ5TFXd2NspyTJgGcCCBQsGtGlJkiTtiDmYJEmC/kYCXQ3M75me17b12gKsrKo7quqHwH/QFIXuoarOrqolVbVkZGRkqjFLkiRpEszBJEkS9FcEWgMsSnJokr2BpcDKMX0+TzMKiCQH0dwetmmAcUqSJEmSJGkadloEqqrtwHJgNXAlcF5VrUtyepLj226rgeuSrAcuBN5cVdfNVNCSJEmSJEmanL6eCVRVq4BVY9pO7XlfwBvblyRJkiRJknYx/dwOJkmSJEmSpN2cRSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkdMHfYAUjSRJJhR6CqYUcgSZIkaVAcCSRJkiRJktQBFoEkSZIkSZI6wCKQJEmSJElSB1gEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkD+ioCJTk2yYYkG5OcMs78k5NsTXJZ+/rDwYcqSZIkSZKkqZq7sw5J5gArgGcCW4A1SVZW1foxXT9dVctnIEZJkiRJkiRNUz8jgY4CNlbVpqraBpwLnDCzYUmSJEmSJGmQ+ikCHQxs7pne0raN9YIk301yfpL5A4lOkiRJkiRJAzGoB0N/AVhYVY8FLgA+Ol6nJMuSrE2yduvWrQPatCRJknbEHEySJEF/RaCrgd6RPfPatrtU1XVVdXs7+SHgiPFWVFVnV9WSqloyMjIylXglSZI0SeZgkiQJ+isCrQEWJTk0yd7AUmBlb4ckD+mZPB64cnAhSpIkSZIkabp2+u1gVbU9yXJgNTAHOKeq1iU5HVhbVSuB1yU5HtgOXA+cPIMxS5IkSZIkaZJ2WgQCqKpVwKoxbaf2vH8r8NbBhiZJkiRJkqRBGdSDoSVJkiRJkrQLswgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeqAvr4dTNrVJMOOQABVw45AkiRJktQvRwJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHdBXESjJsUk2JNmY5JQd9HtBkkqyZHAhSpIkSZIkabp2WgRKMgdYARwHLAZOSrJ4nH77A68HvjXoICVJkiRJkjQ9/YwEOgrYWFWbqmobcC5wwjj93gGcAfxigPFJkiRJkiRpAPopAh0MbO6Z3tK23SXJ44H5VfXFAcYmSZIkSZKkAZn2g6GT7AW8B/jTPvouS7I2ydqtW7dOd9OSJEnqgzmYJEmC/opAVwPze6bntW2j9gcOA76S5Crgt4CV4z0cuqrOrqolVbVkZGRk6lFLkiSpb+ZgkiQJ+isCrQEWJTk0yd7AUmDl6MyquqmqDqqqhVW1ELgYOL6q1s5IxJIkSZIkSZq0nRaBqmo7sBxYDVwJnFdV65KcnuT4mQ5QkiRJkiRJ0ze3n05VtQpYNabt1An6Hj39sCRJkiRJkjRI034wtCRJkiRJknZ9FoEkSZIkSZI6wCKQJEmSJElSB1gEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQP6KgIlOTbJhiQbk5wyzvzXJLkiyWVJvpZk8eBDlSRJkiRJ0lTttAiUZA6wAjgOWAycNE6R55NV9ZiqOhx4N/CegUcqSZIkSZKkKetnJNBRwMaq2lRV24BzgRN6O1TVzT2T9wNqcCFKkiRJkiRpuub20edgYHPP9BbgCWM7JXkt8EZgb+AZ460oyTJgGcCCBQsmG6skSZKmwBxMkiTBAB8MXVUrqurhwFuAt0/Q5+yqWlJVS0ZGRga1aUmSJO2AOZgkSYL+ikBXA/N7pue1bRM5F/jd6QQlSZIkSZKkweqnCLQGWJTk0CR7A0uBlb0dkizqmXwu8P3BhShJkiRJkqTp2ukzgapqe5LlwGpgDnBOVa1LcjqwtqpWAsuTHAPcAdwAvGImg5YkSZIkSdLk9PNgaKpqFbBqTNupPe9fP+C4JEmSJEmSNEADezC0JEmSJEmSdl0WgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOsAikCRJkiRJUgdYBJIkSZIkSeoAi0CSJEmSJEkd0FcRKMmxSTYk2ZjklHHmvzHJ+iTfTfKvSQ4ZfKiSJEmSJEmaqp0WgZLMAVYAxwGLgZOSLB7T7TvAkqp6LHA+8O5BBypJkiRJkqSp62ck0FHAxqraVFXbgHOBE3o7VNWFVfWzdvJiYN5gw5QkSZIkSdJ09FMEOhjY3DO9pW2byKuAL00nKEmSJEmSJA3W3EGuLMlLgSXA0yeYvwxYBrBgwYJBblqSJEkTMAeTJEnQ30igq4H5PdPz2rZ7SHIM8Dbg+Kq6fbwVVdXZVbWkqpaMjIxMJV5JkiRNkjmYJEmC/opAa4BFSQ5NsjewFFjZ2yHJ44AP0hSAfjL4MCVJkiRJkjQdOy0CVdV2YDmwGrgSOK+q1iU5Pcnxbbczgf2AzyS5LMnKCVYnSZIkSZKkIejrmUBVtQpYNabt1J73xww4LkmSJEmSJA1QP7eDSZIkSZIkaTdnEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjrAIpAkSZIkSVIHWASSJEmSJEnqAItAkiRJkiRJHWARSJIkSZIkqQMsAkmSJEmSJHWARSBJkiRJkqQOsAgkSZIkSZLUARaBJEmSJEmSOqCvIlCSY5NsSLIxySnjzH9akkuTbE9y4uDDlCRJkiRJ0nTstAiUZA6wAjgOWAyclGTxmG4/Bk4GPjnoACVJkiRJkjR9c/vocxSwsao2ASQ5FzgBWD/aoaquaufdOQMxSpIkSZIkaZr6KQIdDGzumd4CPGEqG0uyDFgGsGDBgqmsQpIkSZNkDiZJ6pUMOwJVDWe7s/pg6Ko6u6qWVNWSkZGR2dy0JElSZ5mDSZIk6K8IdDUwv2d6XtsmSZIkSZKk3UQ/RaA1wKIkhybZG1gKrJzZsCRJkiRJkjRIOy0CVdV2YDmwGrgSOK+q1iU5PcnxAEmOTLIFeCHwwSTrZjJoSZIkSZIkTU4/D4amqlYBq8a0ndrzfg3NbWKSJEmSJEnaBfVVBJIkSZIkdYPfHDV8w/rmKO35ZvXbwSRJkiRJkjQcFoEkSZIkSZI6wNvBJEmSJPXF24SGz9uEJE2HI4EkSZIkSZI6wCKQJEmSJElSB1gEkiRJkiRJ6gCLQJIkSZIkSR1gEUiSJEmSJKkDLAJJkiRJkiR1gEUgSZIkSZKkDrAIJEmSJEmS1AEWgSRJkiRJkjqgryJQkmOTbEiyMckp48y/b5JPt/O/lWThoAOVJEmSJEnS1O20CJRkDrACOA5YDJyUZPGYbq8CbqiqRwBnAWcMOlBJkiRJkiRNXT8jgY4CNlbVpqraBpwLnDCmzwnAR9v35wO/kySDC1OSJEmSJEnT0U8R6GBgc8/0lrZt3D5VtR24CXjQIAKUJEmSJEnS9M2dzY0lWQYsaydvTbJhNre/mzkI+Omwg5gqx4H1Zbc+x+B57oPnuBt26/M8w+f4kBldu/pmDta33frzrL7t1ufZ38192a3PMXie++A53rEJc7B+ikBXA/N7pue1beP12ZJkLnAAcN3YFVXV2cDZfWyz85Ksraolw45DM8dzvOfzHHeD51m7A3Ow/vh57gbP857Pc7zn8xxPXT+3g60BFiU5NMnewFJg5Zg+K4FXtO9PBL5cVTW4MCVJkiRJkjQdOx0JVFXbkywHVgNzgHOqal2S04G1VbUS+DDwsSQbgetpCkWSJEmSJEnaRfT1TKCqWgWsGtN2as/7XwAvHGxoneeQ7T2f53jP5znuBs+ztOfw89wNnuc9n+d4z+c5nqJ415YkSZIkSdKer59nAkmSJEmSJGk3ZxFoBiU5Lcmbhh2HZlaShUm+N+w4NPs895K0azIH6wZ/D3eX516aOotAkqROSXLrsGOQJEnqGnOwXYNFoAFL8rYk/5Hka8Cvt21fSbKkfX9Qkqva9ycn+XySC5JclWR5kjcm+U6Si5Mc2LP8WUnWJrkyyZFJ/m+S7yd5Z9vn9CRv6InjXUleP9v732Fzk3yiPT/nJ/mVJKcmWZPke0nOTuPhSS4dXSjJotHpJEckuSjJJUlWJ3lI2/66JOuTfDfJucPaQe1Ykoe1n903t5/P/9d+Rt/d0+fW9rN5efsZf/AwY9bgJZkzzeX7+sIGSfdmDtZZ5mAdZw4mMAebDItAA5TkCGApcDjwHODIPhY7DPi9tu+7gJ9V1eOAbwIv7+m3raqWAP8H+Efgte2yJyd5EHDOaP8ke7VxfHwAu6X+/Drwgar6DeBm4L8D76+qI6vqMGBf4HlV9QPgpiSHt8u9Evj7JPcB3gecWFVH0JzPd7V9TgEeV1WPBV4ze7ukfiX5deCzwMnAVpqfAS8CHgO8KMn8tuv9gIur6jeBrwKvnv1oNar9o+DM9o+EK5K8qG3fK8kHkvx7+wfiqiQn7mA9VyU5o/1j4oVJnpXkm0kuTfKZJPu1/Z7TrvOSJO9N8k9t+2lJPpbk68DHZmPfpT2NOVinmYN1mDnY7skcbLgsAg3WU4HPVdXPqupmYGUfy1xYVbdU1VbgJuALbfsVwMKefit72tdV1TVVdTuwCZhfVVcB1yV5HPAs4DtVdd2090j92lxVX2/ffxx4CvDbSb6V5ArgGcCj2/kfAl7ZVqtfBHySJoE5DLggyWXA24F5bf/vAp9I8lJg+6zsjSZjhOaPgpdU1eVt279W1U1V9QtgPXBI274N+Kf2/SXc8zOu2fd7NMnibwLHAGe2V39/j+bcLAZeBjyxj3VdV1WPB/6F5vN7TDu9Fnhjkn2ADwLHtX9kjIxZfnG7zEnT3iupm8zBusscrLvMwXZf5mBD1JkhT0O2nbsLbvuMmXd7z/s7e6bv5J7n5/Zx+ozt9yGaKviv0VzF0OypcaY/ACypqs1JTuPuc/9Z4C+ALwOXVNV1SR5Kk1iO94PuucDTgOcDb0vymKoyEdl13AT8mCbpXN+29X5Gf8ndn9E7qqrGaddwPAX4VFX9EvivJBfRjAh4CvCZqroTuDbJhX2s69Ptv79Fk0x8PQnA3jSjCh4FbKqqH7b9PgUs61l+ZVX9fLo7JOlezMH2fOZg3WUOtvsyBxsiRwIN1leB302yb5L9aX5hAFwFHNG+n3A42wB8DjiW5gO0ega3o3tbkGQ0eXgx8LX2/U/bYYh3nff2ysRq4H8Df982bwBGRteR5D5JHt0OK59fVRcCbwEOAPab8b3RZGwD/hvw8iQvHnYwGprb2n8DXFBVh7evxVX1qkksL2lqzMG6yxysu8zBBOZgk2YRaICq6lKaSuTlwJeANe2svwH+OMl3gINmcPvbgAuB89qqqmbPBuC1Sa4EHkiTXPwd8D2aZGPNmP6foLmC+M9w17k7ETgjyeXAZcCTgDnAx9vhzN8B3ltVN8787mgyquo24HnAnwD3H3I46t+/0TwvYE6SEZqrvd8Gvg68oL0v/cHA0ZNY58XAk5M8AiDJ/ZI8kuZnxMOSLGz7vWgwuyAJzME6zhysw8zBdlvmYEOUu0fFaXfXXrG4FHhhVX1/2PFoYkneBBxQVX8+7Fikrklya1Xtl2as8LuB42huH3hnVX26/Vn6AZrEYzPNlaUzquqCCdZ3Fc1tBz9tp58BnAHct+3y9qpameT5wJk0V5zWAPtX1UvaWxVuraq/mZEdljTjzMF2H+Zg0vCYg+0aLALtIZIspnnY2eeq6k+HHY8mluRzwMOBZ4z+wJK0a0myX1Xdmuabf74NPLmqrh3QOgOsAL5fVWcNIl5Jw2MOtvswB5N2feZgM88ikCRJYyT5CvAAmocKvruqPjKAdf4J8Ip2nd8BXl1VP5vueiVJkvYU5mAzzyKQJEl9aK8gHzqm+S1V5UNgJUmSZog52GBZBJIkSZIkSeoAvx1MkiRJkiSpAywCSZIkSZIkdYBFIEmSJEmSpA6wCCRJkiRJktQBFoEkSZIkSZI64P8D8hkt5iQOu80AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "de97f8f8-5dc6-4d5a-b0a0-783debc5652b"
      },
      "source": [
        "##### Καλύτερος ταξινομητής ως προς την ορθότητα"
      ],
      "id": "de97f8f8-5dc6-4d5a-b0a0-783debc5652b"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "49417660-56b2-46a6-b291-bffeab066770"
      },
      "source": [
        "* Το accuracy και το f1-score του μοντέλου αυτού είναι κρατημένα στο object `max_accuracy`\n",
        "* Εκτυπώνω εκ νέου τις συγκριτικές γραφικές παραστάσεις, και τον ζητούμενο πίνακα markdown για το μοντέλο:"
      ],
      "id": "49417660-56b2-46a6-b291-bffeab066770"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1fb63050-4758-4ef1-a9a2-2dffe86608e1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 526
        },
        "outputId": "41b43c5e-91c3-4a11-90bf-679be4f16f66"
      },
      "source": [
        "plt.title(f\"{max_accuracy['name'].replace('-', '->')}, GridSearch \")\n",
        "plt.bar(['accuracy', 'f1'], [max_accuracy['accuracy'], max_accuracy['f1']])\n",
        "print(\"accuracy:\", max_accuracy['accuracy'])\n",
        "print(\"f1-score:\", max_accuracy['f1'])\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, fig_index=102+c, title_prefix=\"Out of the box\", color=\"green\")"
      ],
      "id": "1fb63050-4758-4ef1-a9a2-2dffe86608e1",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "accuracy: 0.691\n",
            "f1-score: 0.0\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "c2e2d5d0-c7eb-447c-a45a-ad42f3835459"
      },
      "source": [
        "* Ο πίνακας markdown:\n",
        "<table>\n",
        "<thead>\n",
        "  <td>Model</td>\n",
        "  <td>Grid-Search accuracy-score</td>\n",
        "  <td>Grid-Search f1-score</td>\n",
        "  <td>Out of the box accuracy-score</td>\n",
        "  <td>Out of the box f1-score</td>\n",
        "  <td>Grid-Search test-time (ms)</td>\n",
        "  <td>Grid-Search train-time (ms)</td>\n",
        "</thead>\n",
        "    <tbody>\n",
        "        <tr>\n",
        "            <td>Log.regression</td>\n",
        "            <td>0.6191</td>\n",
        "            <td>0.0</td>\n",
        "            <td>0.709</td>\n",
        "            <td>0.0</td>\n",
        "            <td>5.786</td>\n",
        "            <td>0.73</td>\n",
        "        </tr>\n",
        "    </tbody>\n",
        "</table>"
      ],
      "id": "c2e2d5d0-c7eb-447c-a45a-ad42f3835459"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "41b56970-2893-4582-a06a-c958545092e2"
      },
      "source": [
        "###### Παρατηρήσεις"
      ],
      "id": "41b56970-2893-4582-a06a-c958545092e2"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5c6f3812-2a00-47c4-ae35-971a283dc8b9"
      },
      "source": [
        "* Στο μοντέλο με το βέλτιστο accuracy:\n",
        "    * Παρατηρώ  ότι το accuracy-score δεν ξεπερνάει την out-of-the-box επίδοση.\n",
        "    * Το f1-score παραμένει μηδενικό.   "
      ],
      "id": "5c6f3812-2a00-47c4-ae35-971a283dc8b9"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e4cd7f75-6005-4b18-af78-84fabc0f5c83"
      },
      "source": [
        "###### Πίνακας σύγχυσης"
      ],
      "id": "e4cd7f75-6005-4b18-af78-84fabc0f5c83"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "af719eb7-ad2e-421e-ab43-038b93f2b108",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "outputId": "f37ca9f5-625b-45a9-ce2d-0507eb29d16a"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns; sns.set()\n",
        "\n",
        "pred = max_accuracy['predictions']\n",
        "print(f\"{max_accuracy['name']}:\")\n",
        "mat = confusion_matrix(test_labels_grid , pred)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "af719eb7-ad2e-421e-ab43-038b93f2b108",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "selector-scaler-pca-log_regr:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2ab039fb-b205-4e24-9bd8-d05d3282be1a"
      },
      "source": [
        "##### Χειρότερος ταξινομητής ως προς την ορθότητα:"
      ],
      "id": "2ab039fb-b205-4e24-9bd8-d05d3282be1a"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0be31d64-1302-4110-9a3a-633f13aaa6f6"
      },
      "source": [
        "* Το accuracy και το f1-score του μοντέλου αυτού είναι κρατημένα στο object `min_accuracy`\n",
        "* Εκτυπώνω εκ νέου τις συγκριτικές γραφικές παραστάσεις, και τον ζητούμενο πίνακα markdown για το μοντέλο:"
      ],
      "id": "0be31d64-1302-4110-9a3a-633f13aaa6f6"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "778bb368-05fe-43d7-a944-f762e9ab6730",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 532
        },
        "outputId": "44732d27-2304-49ac-ddff-a762230eec88"
      },
      "source": [
        "plt.title(f\"{min_accuracy['name'].replace('-', '->')}, GridSearch \")\n",
        "plt.bar(['accuracy', 'f1'], [min_accuracy['accuracy'], min_accuracy['f1']])\n",
        "print(\"accuracy:\", min_accuracy['accuracy'])\n",
        "print(\"f1-score:\", min_accuracy['f1'])\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, fig_index=102+c, title_prefix=\"Out of the box\", color=\"green\")"
      ],
      "id": "778bb368-05fe-43d7-a944-f762e9ab6730",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "accuracy: 0.345\n",
            "f1-score: 0.182\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "05fd8835-0fbf-43f5-87ab-80dc88644de6"
      },
      "source": [
        "###### Παρατηρήσεις"
      ],
      "id": "05fd8835-0fbf-43f5-87ab-80dc88644de6"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "10eb0ece-cdaf-45ac-bef0-bd52c7f2d169"
      },
      "source": [
        "* Στο μοντέλο με το ελάχιστο accuracy:\n",
        "    * Παρατηρώ  ότι το accuracy-score είναι αρκετά μικρότερο από την out-of-the-box επίδοση.\n",
        "    * Το f1-score ωστόσο έχει αυξηθεί σημαντικά.   "
      ],
      "id": "10eb0ece-cdaf-45ac-bef0-bd52c7f2d169"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "67c5727c-8bb2-4deb-acf3-e1412e13d09f"
      },
      "source": [
        "###### Πίνακας σύγχυσης"
      ],
      "id": "67c5727c-8bb2-4deb-acf3-e1412e13d09f"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4cc1a449-7026-4ebc-a46e-2aac99b74296",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "outputId": "92e8c8b9-8f07-4a00-cd80-a41161c4c823"
      },
      "source": [
        "pred = min_accuracy['predictions']\n",
        "print(f\"{min_accuracy['name']}:\")\n",
        "mat = confusion_matrix(test_labels_grid , pred)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "4cc1a449-7026-4ebc-a46e-2aac99b74296",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "scaler-sampler-bayes:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "66ec8232-527c-4788-8d53-ab5455e2bd35"
      },
      "source": [
        "##### Ταξινομητής που προτείνω"
      ],
      "id": "66ec8232-527c-4788-8d53-ab5455e2bd35"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3bb698cc-d865-4831-99df-760345e6836f"
      },
      "source": [
        "* Προτείνω τον ταξινομητή που έχει τον μεγαλύτερο μέσο όρο των δύο scores\n",
        "* Το accuracy και το f1-score του μοντέλου αυτού είναι κρατημένα στο object `max_average`"
      ],
      "id": "3bb698cc-d865-4831-99df-760345e6836f"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b8a9ecf2-cfab-4ae5-827d-548ae05ae1cc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 532
        },
        "outputId": "08059f9c-8245-4129-f4d2-491af510dc77"
      },
      "source": [
        "plt.title(f\"{max_average['name'].replace('-', '->')}, GridSearch \")\n",
        "plt.bar(['accuracy', 'f1'], [max_average['accuracy'], max_average['f1']])\n",
        "print(\"accuracy:\", max_average['accuracy'])\n",
        "print(\"f1-score:\", max_average['f1'])\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, fig_index=102+c, title_prefix=\"Out of the box\", color=\"green\")"
      ],
      "id": "b8a9ecf2-cfab-4ae5-827d-548ae05ae1cc",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "accuracy: 0.564\n",
            "f1-score: 0.368\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "40def99d-6fa8-4a14-b99d-70f59511cee4"
      },
      "source": [
        "###### Παρατηρήσεις"
      ],
      "id": "40def99d-6fa8-4a14-b99d-70f59511cee4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8e161cab-c70c-4795-b03a-ca39acc6a886"
      },
      "source": [
        "* Παρατηρώ ότι το accuracy score δεν ξεπερνά και πάλι το out-of-the-box μοντέλο.\n",
        "* Το f1-score έχει αυξηθεί σημαντικά, πλησιάζοντας το accuracy-score."
      ],
      "id": "8e161cab-c70c-4795-b03a-ca39acc6a886"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d4e72815-f131-4236-a1dd-28b2020b93a7"
      },
      "source": [
        "###### Πίνακας σύγχυσης"
      ],
      "id": "d4e72815-f131-4236-a1dd-28b2020b93a7"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "16f232fd-b5a9-4f3c-b571-b7891899eb68",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "outputId": "64fa03b9-cc3e-40a6-f086-14a1b6490714"
      },
      "source": [
        "pred = max_average['predictions']\n",
        "print(f\"{max_average['name']}:\")\n",
        "mat = confusion_matrix(test_labels_grid , pred)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "16f232fd-b5a9-4f3c-b571-b7891899eb68",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "scaler-sampler-pca-knn:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_P1cP80bP1Nv"
      },
      "source": [
        "## ΜΕΡΟΣ 2"
      ],
      "id": "_P1cP80bP1Nv"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Y3XAV3LUP8cT"
      },
      "source": [
        "### Dataset: K05 - Εισαγωγή και επισκόπηση"
      ],
      "id": "Y3XAV3LUP8cT"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9K40k_YcQCZG"
      },
      "source": [
        "#### Σύντομη Παρουσίαση"
      ],
      "id": "9K40k_YcQCZG"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "a4dQqkB9QKyh"
      },
      "source": [
        "* Το dataset είναι βασισμένο σε μία εφαρμογή εύρεσης ταξί , της οποίας στόχος είναι το βέλτιστο matching ταξί και πελάτη βάσει διάφορων παραγόντων.\n",
        "* Ο στόχος είναι η ταξινόμηση σε έναν από τους 3 τύπους στη στήλη που ονομάζεται SurgePricingType. \n",
        "* Η ταξινόμηση είναι ένα `Multi-classification` πρόβλημα.\n",
        "* Στα δεδομένα συμπεριλαμβάνονται η απόσταση του ταξιδιού, στοιχεία για το ταξί ,τον προορισμό καθώς και διάφοροι δείκτες για τον πελάτη.\n",
        "* Δουλεύουμε με το αρχείο `sigma_cabs.csv`."
      ],
      "id": "a4dQqkB9QKyh"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-hZruS0e2km-"
      },
      "source": [
        "#### Ανέβασμα dataset\n",
        "\n"
      ],
      "id": "-hZruS0e2km-"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "ok": true,
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 72
        },
        "id": "fyF1G8Uh2ajw",
        "outputId": "70174406-aa9c-4b50-d9e0-7a48fae37d8b"
      },
      "source": [
        "from google.colab import files\n",
        "uploaded_files = files.upload()"
      ],
      "id": "fyF1G8Uh2ajw",
      "execution_count": null,
      "outputs": [
        {
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-f0f73860-ef42-4ed5-a616-2d261266fbd9\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-f0f73860-ef42-4ed5-a616-2d261266fbd9\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Saving sigma_cabs.csv to sigma_cabs.csv\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KyX4ml8NeXYP"
      },
      "source": [
        "#### Διάβασμα dataset"
      ],
      "id": "KyX4ml8NeXYP"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_T72gSOBSF0S",
        "outputId": "6f2e7a48-066d-48ed-d072-768e7e185cc6"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "data = pd.read_csv(\"/content/sigma_cabs.csv\",header=0)\n",
        "\n",
        "examples_num = len(data.iloc[:,0])\n",
        "features_num = len(data.iloc[0,:13])\n",
        "\n",
        "print(f\"Dataset\",f\"\\nSamples: {examples_num}\", f\"\\nFeatures: {features_num}\", \"\\n\")\n",
        "print(\"Shape:\", data.shape)"
      ],
      "id": "_T72gSOBSF0S",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Dataset \n",
            "Samples: 131662 \n",
            "Features: 13 \n",
            "\n",
            "Shape: (131662, 14)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_NfXKTK5sDWD"
      },
      "source": [
        "#### Προεπεξεργασία"
      ],
      "id": "_NfXKTK5sDWD"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x18fOD14OsSg"
      },
      "source": [
        "##### Αντικάτασταση τιμών"
      ],
      "id": "x18fOD14OsSg"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3ahW5rg8eAo1"
      },
      "source": [
        "Παρατηρούμε πως έχουμε 5 κατηγορικές μεταβλητές και πως οι υπόλοιπες είναι αριθμητικές."
      ],
      "id": "3ahW5rg8eAo1"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XpTfXb__d5RI",
        "outputId": "f4746c69-0fa0-4f43-8f3d-b042d9bf006b"
      },
      "source": [
        "cats = []\n",
        "nums = []\n",
        "for i in range(len(data.columns)-1):\n",
        "  c = data.columns[i]\n",
        "  if data[c].dtypes == 'object':\n",
        "    print(f\"{c} with index {i} is categorical\")\n",
        "    cats.append(i)\n",
        "  else:\n",
        "    print(f\"{c} with index {i} is numerical\")\n",
        "    nums.append(i)"
      ],
      "id": "XpTfXb__d5RI",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Trip_ID with index 0 is categorical\n",
            "Trip_Distance with index 1 is numerical\n",
            "Type_of_Cab with index 2 is categorical\n",
            "Customer_Since_Months with index 3 is numerical\n",
            "Life_Style_Index with index 4 is numerical\n",
            "Confidence_Life_Style_Index with index 5 is categorical\n",
            "Destination_Type with index 6 is categorical\n",
            "Customer_Rating with index 7 is numerical\n",
            "Cancellation_Last_1Month with index 8 is numerical\n",
            "Var1 with index 9 is numerical\n",
            "Var2 with index 10 is numerical\n",
            "Var3 with index 11 is numerical\n",
            "Gender with index 12 is categorical\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6MDjqC0zd-E0",
        "outputId": "0e952fde-f046-418a-8f7f-e3f37ff4c872"
      },
      "source": [
        "print(\"Categorical columns indexes:\", cats)\n",
        "print(\"Numerical columns indexes:\", nums)"
      ],
      "id": "6MDjqC0zd-E0",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Categorical columns indexes: [0, 2, 5, 6, 12]\n",
            "Numerical columns indexes: [1, 3, 4, 7, 8, 9, 10, 11]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GaZJNOGIyDkq"
      },
      "source": [
        "##### Προεπεξεργασία διαχείρισης απουσιάζουσων τιμών"
      ],
      "id": "GaZJNOGIyDkq"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lvcKYwC2sG4_",
        "outputId": "dc42c961-0ab1-4a75-e259-5c390ffb99b8"
      },
      "source": [
        "import numpy as np\n",
        "\n",
        "data_grid = np.array(data)\n",
        "none_val = np.sum(np.sum(data.isna()))\n",
        "print(\"Number of empty values in train_set:\", none_val)\n",
        "print(\"Percentage of empty values in train_set:\",f\"{round(none_val/(features_num*examples_num)*100,3)}%\")"
      ],
      "id": "lvcKYwC2sG4_",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of empty values in train_set: 137546\n",
            "Percentage of empty values in train_set: 8.036%\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dDz10VkceLGW"
      },
      "source": [
        "Θα αντικαταστήσουμε τις τιμές με αριθμούς στις κατηγορικές μεταβλητές, και το 'NaN' με το np.NaN, ώστε να ορίσουμε μετά τους κατάλληλους imputers. "
      ],
      "id": "dDz10VkceLGW"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TntkOXVNO2gV",
        "outputId": "c5655b2d-2fea-4995-ab25-715dec00b9d4"
      },
      "source": [
        "#to Dataframes\n",
        "data_df = pd.DataFrame(data_grid)\n",
        "\n",
        "#create mappings for features that represent categories in order to represent them with numbers\n",
        "cab_mapping = {'A':0,'B':1,'C':2,'D':3,'E':4}\n",
        "confidence_mapping={'A':0,'B':1,'C':2}\n",
        "dest_mapping={'A':0,'B':1,'C':2,'D':3,'E':4,'F':5,'G':6,'H':7,'I':8,'J':9,'K':10,'L':11,'M':12,'N':13}\n",
        "gender_mapping={'Male':0,'Female':1}\n",
        "surge_mapping={1:0,2:1,3:2}\n",
        "\n",
        "#attach columns to Dataframe\n",
        "data_df.columns = list(data.columns)\n",
        "\n",
        "#delete Trip_ID column as it is not needed\n",
        "data_df = data_df.drop('Trip_ID',1)\n",
        "cats.remove(0)\n",
        "# all indexes on nums and cats lists should: -=1\n",
        "nums = list(map(lambda x: x-1, nums))\n",
        "cats = list(map(lambda x: x-1, cats))\n",
        "\n",
        "#for training set\n",
        "data_df['Type_of_Cab'] = data_df['Type_of_Cab'].map(cab_mapping)\n",
        "data_df['Confidence_Life_Style_Index'] = data_df['Confidence_Life_Style_Index'].map(confidence_mapping)\n",
        "data_df.Destination_Type = data_df.Destination_Type.map(dest_mapping)\n",
        "data_df.Gender = data_df.Gender.map(gender_mapping)\n",
        "data_df.Surge_Pricing_Type = data_df.Surge_Pricing_Type.map(surge_mapping)\n",
        "\n",
        "#then, we are going to replace empty values in the categoric features with a specific value in order to \n",
        "#replace it later with the most frequent method in imputing\n",
        "data_df.replace('NaN',np.NaN,inplace=True)\n",
        "\n",
        "#finally, we get \n",
        "print(data_df)"
      ],
      "id": "TntkOXVNO2gV",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:15: FutureWarning: In a future version of pandas all arguments of DataFrame.drop except for the argument 'labels' will be keyword-only\n",
            "  from ipykernel import kernelapp as app\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "       Trip_Distance  Type_of_Cab  ... Gender Surge_Pricing_Type\n",
            "0               6.77          1.0  ...      1                  1\n",
            "1              29.47          1.0  ...      0                  1\n",
            "2              41.58          NaN  ...      0                  1\n",
            "3              61.56          2.0  ...      0                  2\n",
            "4              54.95          2.0  ...      0                  1\n",
            "...              ...          ...  ...    ...                ...\n",
            "131657         11.72          3.0  ...      0                  2\n",
            "131658         74.81          2.0  ...      0                  1\n",
            "131659         40.17          2.0  ...      1                  1\n",
            "131660         46.88          1.0  ...      0                  1\n",
            "131661         31.96          0.0  ...      0                  0\n",
            "\n",
            "[131662 rows x 13 columns]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S9BHwIpv5cUg"
      },
      "source": [
        "#### Αναλογία test και train sets"
      ],
      "id": "S9BHwIpv5cUg"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wN-nUQoo6sa5"
      },
      "source": [
        "Θα θέσουμε την αναλογία train-test set σε 70-30 με την συναρτηση train_test_split του scikit."
      ],
      "id": "wN-nUQoo6sa5"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_tlzxgq55nsg",
        "outputId": "92c32a1d-460d-445f-b9e8-ded2f2e2e084"
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "labels =  data_df.iloc[:, [12]].values\n",
        "features = data_df.iloc[:, :12].values\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(features, labels, test_size=0.3)\n",
        "\n",
        "print(\"Train features shape:\", X_train.shape)\n",
        "print(\"Test features shape:\", X_test.shape)\n",
        "print(\"Train labels shape:\", Y_train.shape)\n",
        "print(\"Test labels shape:\", Y_test.shape)\n"
      ],
      "id": "_tlzxgq55nsg",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Train features shape: (92163, 12)\n",
            "Test features shape: (39499, 12)\n",
            "Train labels shape: (92163, 1)\n",
            "Test labels shape: (39499, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kIGJF6gxL1Td"
      },
      "source": [
        "##### Εξάλειψη απουσιάζουσων τιμών"
      ],
      "id": "kIGJF6gxL1Td"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BdeBmC1vAigl"
      },
      "source": [
        "* Μελετώντας το dataset παρατηρούμε ότι οι απουσιάζουσες τιμές είναι σε συγκεκριμένες στήλες και όχι σε όλες.\n",
        "* Ορίζουμε δύο ειδών imputers, ώστε να χειριστούμε αλλιώς τις απυσιάζουσες τιμές στις στήλες με κατηγορικά features, και αλλιώς στις στήλες με αριθμητικά features.\n",
        "  * Κατηγορικά features: strategy = most_frequent\n",
        "  * Αριθμητικά features: strategy = mean\n"
      ],
      "id": "BdeBmC1vAigl"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8xWHKkoXMQPc"
      },
      "source": [
        "from sklearn.impute import SimpleImputer\n",
        "from sklearn.compose import ColumnTransformer\n",
        "from sklearn.pipeline import Pipeline\n",
        "\n",
        "imp_numeric = Pipeline(\n",
        "    steps = [\n",
        "      ('imputer', SimpleImputer(strategy='mean')),\n",
        "    ]\n",
        ")\n",
        " \n",
        "imp_categ = Pipeline(\n",
        "    steps = [\n",
        "      ('imputer', SimpleImputer(strategy='most_frequent')),\n",
        "    ]\n",
        ")\n",
        " \n",
        "preprocessor = ColumnTransformer(\n",
        "    transformers = [\n",
        "      ('imputer_numeric', imp_numeric, nums),\n",
        "      ('imputer_categoric', imp_categ, cats),\n",
        "    ]\n",
        ")\n",
        "\n",
        "# fit transform on the training set\n",
        "new_X_train_df = pd.DataFrame(preprocessor.fit_transform(pd.DataFrame(X_train)))\n",
        "# ONLY transform on the testing set\n",
        "new_X_test_df  = pd.DataFrame(preprocessor.transform(pd.DataFrame(X_test)))\n"
      ],
      "id": "8xWHKkoXMQPc",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dyl5UYy3ZjGw"
      },
      "source": [
        "Βλέπουμε ότι αντικαταστήσαμε όλες τις απουσιάζουσες τιμές , αφού πλέον δεν υπάρχουν."
      ],
      "id": "dyl5UYy3ZjGw"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EXBR-FzwRU1P",
        "outputId": "b8b804a5-d013-4b7b-f92e-70f49299a312"
      },
      "source": [
        "none_val = np.sum(np.sum(new_X_train_df).isna()) + np.sum(np.sum(new_X_test_df).isna())\n",
        "print(\"Number of empty values in train and test set features:\", none_val)"
      ],
      "id": "EXBR-FzwRU1P",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Number of empty values in train and test set features: 0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VxQehHS-yc6d"
      },
      "source": [
        "Τα **labels** βρίσκονται στην τελευταία στήλη (SurgePricingType) και όπως φαίνεται δεν έχουν απουσιάζουσες τιμές.\n",
        "\n"
      ],
      "id": "VxQehHS-yc6d"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oRNp82Mlyf07",
        "outputId": "240c294a-adfe-4cb9-d265-c1ec2285421c"
      },
      "source": [
        "print(\"Labels:\", labels.ravel())\n",
        "\n",
        "#Missing values in SurgePricingType\n",
        "missing =  np.sum(np.sum(pd.DataFrame(labels).isna()))\n",
        "print(f\"\\nNumber of missing values in labels: {missing}\")\n"
      ],
      "id": "oRNp82Mlyf07",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Labels: [1 1 1 ... 1 1 0]\n",
            "\n",
            "Number of missing values in labels: 0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "o1yXfv-NByZd"
      },
      "source": [
        "##### Από εδώ και πέρα δουλεύουμε με τα: new_X_train_df, new_X_test_df, Y_train, Y_test"
      ],
      "id": "o1yXfv-NByZd"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "T8NvjKde84_w",
        "outputId": "eaa04ecc-9b31-4a49-a2a7-cc1c6e5a4c4f"
      },
      "source": [
        "print(\"Transformed train shape:\", new_X_train_df.shape)\n",
        "print(\"Transformed test shape:\", new_X_test_df.shape)\n",
        "print(\"Train labels shape:\", Y_train.shape)\n",
        "print(\"Test labels shape:\", Y_test.shape)"
      ],
      "id": "T8NvjKde84_w",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Transformed train shape: (92163, 12)\n",
            "Transformed test shape: (39499, 12)\n",
            "Train labels shape: (92163, 1)\n",
            "Test labels shape: (39499, 1)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HB5n87mD27g0"
      },
      "source": [
        "#### Μετράμε ότι τo dataset μας δεν είναι ισορροπημένο."
      ],
      "id": "HB5n87mD27g0"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 382
        },
        "id": "L6tyFiwq19Ut",
        "outputId": "424aa074-4053-4024-b7b3-72400ae82768"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "#Check if dataset is balanced\n",
        "unique, counts = np.unique(labels, return_counts = True)\n",
        "counts_dict = dict(zip(unique,counts))\n",
        "\n",
        "types=dict()\n",
        "print(\"\\nTypes \\t|Percentage \\t\\t|Total times\")\n",
        "for key in counts_dict.keys() :\n",
        "  types[f\"Type {key}\"]= round(counts_dict[key]/len(labels)*100,3)\n",
        "  print(f\"Type {key} \\t| {round(counts_dict[key]/len(labels)*100,3)}% \\t\\t|{counts_dict[key]}\")\n",
        "\n",
        "plt.figure(21)\n",
        "plt.title(\"Percentages\")\n",
        "plt.xlabel(\"Type\")\n",
        "plt.ylabel(\"Percentage\")\n",
        "barlist = plt.bar(types.keys(), types.values(),color=['blue','red','green'])\n",
        "plt.show()"
      ],
      "id": "L6tyFiwq19Ut",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            "Types \t|Percentage \t\t|Total times\n",
            "Type 0 \t| 20.67% \t\t|27214\n",
            "Type 1 \t| 43.086% \t\t|56728\n",
            "Type 2 \t| 36.244% \t\t|47720\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eXi4HR20gY2Q"
      },
      "source": [
        "#### Συμπεράσματα-Παρατηρήσεις"
      ],
      "id": "eXi4HR20gY2Q"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Jj9J68oXgevq"
      },
      "source": [
        "Παρατηρούμε τα εξής: \n",
        "* Το dataset μας αποτελείται απο 14 στήλες και έχουμε 131.662 examples.\n",
        "* Η πρώτη στήλη είναι το ID του ταξιδιού, οπότε δεν έχει πρακτική σημασία και την αφαιρούμε.\n",
        "* Η τελευταία στήλη είναι αυτή η οποία διαχωρίζει κάθε example σε μία απο τις 3 κατηγορίες.\n",
        "* Καταγράφεται επίσης η απόσταση του ταξιδιού, η κατηγορία του ταξί, ο προορισμός(διαχωρισμένος σε 1 από 14 κατηγορίες).\n",
        "* Έχουμε επικεφαλίδες.\n",
        "* Υπάρχουν 137.546 απουσιάζουσες τιμές στο train set σε κάποια χαρακτηριστικά, τα οποία θα αφαιρέσουμε αργότερα, στο στάδιο της βελτιστοποίησης, για να δούμε την διαφορά που θα επιφέρει η βελτίωση αυτή στην επίδοση των ταξινομητών.\n",
        "* Το dataset μας **δεν είναι ισορροπημένο** αφού βλέπουμε\n",
        "ανόμοια κατανομή στις 3 κατηγορίες."
      ],
      "id": "Jj9J68oXgevq"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yb1_55RSPkO_"
      },
      "source": [
        "### Μετρικές\n"
      ],
      "id": "yb1_55RSPkO_"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "69Q0LtAGPo90"
      },
      "source": [
        "Το συγκεκριμένο πρόβλημα είναι multi-classification πρόβλημα ,γεγονός που πρέπει να λάβουμε υπόψιν κατά την επιλογή των μετρικών. Έτσι θα χρησιμοποιήσουμε:\n",
        "1. Accuracy Score\n",
        "2. F1 Score\n",
        "3. Recall Score"
      ],
      "id": "69Q0LtAGPo90"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7-9PW5T0f6Qw"
      },
      "source": [
        "### Cross-Validation και Train-Test Split"
      ],
      "id": "7-9PW5T0f6Qw"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1XPFMYG2gXxn"
      },
      "source": [
        "* **Train-Test Split**\n",
        "> Το dataset μας το χωρίσαμε σε train και test sets με αναλογία είναι 70-30.\n",
        "\n",
        "* **Cross-Validation**\n",
        "> Θα χρησιμοποιήσουμε cross-validation και συγκεκριμένα το σχήμα k-folds CV με k=10. Στο k-folds το train-set μας χωρίζεται σε 10 μικρότερα μέρη. Θα την υλοποιήσουμε χρησιμοποιώντας την *cross_val_score* της *sklearn.model_selection* ."
      ],
      "id": "1XPFMYG2gXxn"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Tco6thN-n5L0"
      },
      "source": [
        "### Ταξινόμηση"
      ],
      "id": "Tco6thN-n5L0"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OCAQJroin-On"
      },
      "source": [
        "#### Θα δουλέψουμε με 2 ταξινομητές:\n",
        "* Mylti-Layer Perceptron (MLP), και\n",
        "* Support Vector Machines (SVM)"
      ],
      "id": "OCAQJroin-On"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qh95iB7uTe_t"
      },
      "source": [
        "#### Out of the box"
      ],
      "id": "qh95iB7uTe_t"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xlEir4g8TmSt"
      },
      "source": [
        "from sklearn.neural_network import MLPClassifier\n",
        "from sklearn.dummy import DummyClassifier\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import accuracy_score, f1_score, recall_score\n",
        "\n",
        "models = [\n",
        "    {\n",
        "      'name': 'dummy',\n",
        "      'model': DummyClassifier(),\n",
        "    },\n",
        "    {\n",
        "      'name': 'MLP',\n",
        "      'model': MLPClassifier(),\n",
        "    },\n",
        "    {\n",
        "      'name': 'SVM',\n",
        "      'model': SVC(decision_function_shape='ovo')\n",
        "    },\n",
        "]\n",
        "\n",
        "def compute_simple_scores(models, train, train_labels, test, test_labels):\n",
        "    accuracy = dict()\n",
        "    f1 = dict()\n",
        "    recall = dict()\n",
        "    for m in models:\n",
        "      name, model = m.values()\n",
        "      model.fit(train, train_labels.ravel())\n",
        "      preds = model.predict(test)\n",
        "      accuracy[name] = round(accuracy_score(test_labels, preds), 3)\n",
        "      f1[name] = round(f1_score(test_labels, preds, average='macro'), 3)\n",
        "      recall[name] = round(recall_score(test_labels, preds, average='macro'), 3)\n",
        "    return accuracy, f1, recall\n"
      ],
      "id": "xlEir4g8TmSt",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NbBJQysp9fDN"
      },
      "source": [
        "##### Fit σε ολόκληρο το train set και έλεγχος επίδοσης στο test set"
      ],
      "id": "NbBJQysp9fDN"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RaX40iSY9QM9",
        "outputId": "3fdba698-fd8a-4148-cbe3-a7182ad20e3c"
      },
      "source": [
        "out_of_the_box_accuracy, out_of_the_box_f1, out_of_the_box_recall = compute_simple_scores(models, new_X_train_df, Y_train, new_X_test_df, Y_test)\n",
        "\n",
        "print(\"Accuracy:\", out_of_the_box_accuracy)\n",
        "print(\"F1:\", out_of_the_box_f1)\n",
        "print(\"Recall:\", out_of_the_box_recall)"
      ],
      "id": "RaX40iSY9QM9",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracy: {'dummy': 0.428, 'MLP': 0.684, 'SVM': 0.637}\n",
            "F1: {'dummy': 0.2, 'MLP': 0.668, 'SVM': 0.606}\n",
            "Recall: {'dummy': 0.333, 'MLP': 0.655, 'SVM': 0.588}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ORt2uwpK3bbl"
      },
      "source": [
        "##### Πίνακας markdown"
      ],
      "id": "ORt2uwpK3bbl"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SOAiCSNK3c8P"
      },
      "source": [
        "<table>\n",
        "<thead>\n",
        "  <td>Model</td>\n",
        "  <td>Accuracy-score</td>\n",
        "  <td>F1-score</td>\n",
        "  <td>Recall-score</td>\n",
        "</thead>\n",
        "<tbody>\n",
        "<tr>\n",
        "  <td>Dummy</td>\n",
        "  <td>0.428</td>\n",
        "  <td>0.2</td>\n",
        "  <td>0.333</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>MLP</td>\n",
        "  <td>0.684</td>\n",
        "  <td>0.668</td>\n",
        "  <td>0.655</td>\n",
        "</tr>\n",
        "<tr>\n",
        "  <td>SVM</td>\n",
        "  <td>0.637</td>\n",
        "  <td>0.606</td>\n",
        "  <td>0.588</td>\n",
        "</tr>\n",
        "\n",
        "</tbody>\n",
        "</table>"
      ],
      "id": "SOAiCSNK3c8P"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sdAwUvDi3iUU"
      },
      "source": [
        "##### Bar plots"
      ],
      "id": "sdAwUvDi3iUU"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 229
        },
        "id": "KiOmWDxC3j3G",
        "outputId": "213716c0-2c82-498e-e579-3d4ee5ab6696"
      },
      "source": [
        "# attach SVMs results\n",
        "# out_of_the_box_accuracy['SVM']=0.64\n",
        "# out_of_the_box_f1['SVM']=0.612\n",
        "# out_of_the_box_recall['SVM']=0.594\n",
        "\n",
        "def plot_scores(accuracy, f1, recall, fig_index=1, title_prefix=\"\", color=\"blue\"):\n",
        "    fig, axs = plt.subplots(1, 3, figsize=(20, 3), sharey=True)\n",
        "    \n",
        "    axs[0].bar(accuracy.keys(), accuracy.values(), color=color)\n",
        "    axs[0].set_title(f\"{title_prefix} Accuracies\")\n",
        "\n",
        "    axs[1].bar(f1.keys(), f1.values(), color=color)\n",
        "    axs[1].set_title(f\"{title_prefix} F1-scores\")\n",
        "\n",
        "    axs[2].bar(recall.keys(), recall.values(), color=color)\n",
        "    axs[2].set_title(f\"{title_prefix} Recall-scores\")\n",
        "\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, out_of_the_box_recall)"
      ],
      "id": "KiOmWDxC3j3G",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 3 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "inG-TR603rCx"
      },
      "source": [
        "##### Παρατηρήσεις"
      ],
      "id": "inG-TR603rCx"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cdcF6RUr3t72"
      },
      "source": [
        "* Και ο MLP classifier και τα SVMs παρουσιάζουν πολύ καλύτερες επιδόσεις από τον dummy classifier σε όλες τις μετρικές.\n",
        "* Ο MLP classifier είναι ελαφρώς καλύτερος από τα SVMs σε όλες τις μετρικές."
      ],
      "id": "cdcF6RUr3t72"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "psEvUSy9C_0P"
      },
      "source": [
        "#### Βελτιστοποίηση SVM"
      ],
      "id": "psEvUSy9C_0P"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t_gGCwZ3_egF"
      },
      "source": [
        "##### Με pipeline"
      ],
      "id": "t_gGCwZ3_egF"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-f0A6uNIAJgz"
      },
      "source": [
        "* Χρησιμοποιούμε:\n",
        "  * pipeline για να εισαγάγουμε την προεπεξεργασία\n",
        "  * την βιβλιοθήκη optuna για την βελτιστοποίηση των υπερπαραμέτρων των μετασχηματιστών και των εκτιμητών\n",
        "* Όσον αφορά τα εύρη που μελετάμε στις υπερπαραμέτρους:\n",
        "  * PCA:\n",
        "    * δεδομένου ότι έχουμε 12 features, δοκιμάζουμε να τα αναλύσουμε σε έως 10 \n",
        "      components\n",
        "  * SVM:\n",
        "    * δοκιμάζουμε τις πιο συνήθεις επιλογές για τις υπερπαραμέτρους kernel, c, \n",
        "      gamma και degree"
      ],
      "id": "-f0A6uNIAJgz"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lt-QMxrwSJPB"
      },
      "source": [
        "# will be used to store the scores of the best models\n",
        "best_accuracies = dict()\n",
        "best_f1s = dict()\n",
        "best_recalls = dict()"
      ],
      "id": "lt-QMxrwSJPB",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "J79ZmMoWDFsM",
        "outputId": "4a629510-e1f6-47cb-841e-771a626e90e6"
      },
      "source": [
        "import optuna\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import f1_score\n",
        "\n",
        "predictions = []\n",
        "\n",
        "def objective_svm(trial):\n",
        "    \n",
        "    steps = []\n",
        "    \n",
        "    scaler = StandardScaler()\n",
        "    steps.append(('scaler', scaler))\n",
        "\n",
        "    #ros = RandomOverSampler()\n",
        "    #steps.append(('ros', ros))\n",
        "    \n",
        "    n_components = trial.suggest_int('n_components', 2, 10, step=4)\n",
        "    pca = PCA(n_components=n_components)\n",
        "    steps.append(('pca', pca))\n",
        "    \n",
        "    kernel=trial.suggest_categorical('kernel',['rbf','poly','linear','sigmoid'])\n",
        "    c=trial.suggest_float(\"C\",0.1,3.0,log=True)\n",
        "    gamma=trial.suggest_categorical('gamma',['auto','scale'])\n",
        "    degree=trial.suggest_int(\"degree\",1,3,log=True)\n",
        "    \n",
        "    svm = SVC(decision_function_shape='ovo',degree=degree,gamma=gamma,C=c,kernel=kernel)\n",
        "    steps.append(('svm', svm))\n",
        "    \n",
        "    pipe = Pipeline(steps=steps)\n",
        "    \n",
        "    pipe.fit(new_X_train_df, Y_train.ravel())\n",
        "  \n",
        "    preds = pipe.predict(new_X_test_df)\n",
        "\n",
        "    predictions.append({\n",
        "        'n_components': n_components,\n",
        "        'kernel':kernel,\n",
        "        'gamma':gamma,\n",
        "        'C':c,\n",
        "        'degree':degree,\n",
        "        'preds': preds\n",
        "    })\n",
        "    \n",
        "    score = f1_score(Y_test, preds, average='macro')\n",
        "    \n",
        "    return score\n",
        "\n",
        "\n",
        "study = optuna.create_study(direction='maximize')\n",
        "study.optimize(objective_svm, n_trials=4)"
      ],
      "id": "J79ZmMoWDFsM",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "\u001b[32m[I 2021-12-02 21:02:27,591]\u001b[0m A new study created in memory with name: no-name-d0466d31-2fea-4340-aec8-00aa4617a03e\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 21:12:36,361]\u001b[0m Trial 0 finished with value: 0.4310638160275016 and parameters: {'n_components': 10, 'kernel': 'poly', 'C': 1.204552851257661, 'gamma': 'scale', 'degree': 2}. Best is trial 0 with value: 0.4310638160275016.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 21:24:40,814]\u001b[0m Trial 1 finished with value: 0.6600886858013236 and parameters: {'n_components': 10, 'kernel': 'rbf', 'C': 0.20796091832380836, 'gamma': 'scale', 'degree': 1}. Best is trial 1 with value: 0.6600886858013236.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 21:32:22,065]\u001b[0m Trial 2 finished with value: 0.5508205656652657 and parameters: {'n_components': 6, 'kernel': 'linear', 'C': 0.5103197999781499, 'gamma': 'scale', 'degree': 1}. Best is trial 1 with value: 0.6600886858013236.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 21:44:35,113]\u001b[0m Trial 3 finished with value: 0.5625411524902337 and parameters: {'n_components': 6, 'kernel': 'rbf', 'C': 0.32060168355070784, 'gamma': 'auto', 'degree': 1}. Best is trial 1 with value: 0.6600886858013236.\u001b[0m\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3x2Qvs5VWDuk"
      },
      "source": [
        "Τελικά οι καλυτέροι υπερπαράμετροι για την ταξινόμηση με SVM είναι:"
      ],
      "id": "3x2Qvs5VWDuk"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "x_ofCXc6WDfj",
        "outputId": "ff2e6e43-9030-4b25-f453-af684d8922a1"
      },
      "source": [
        "study.best_params"
      ],
      "id": "x_ofCXc6WDfj",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'C': 0.20796091832380836,\n",
              " 'degree': 1,\n",
              " 'gamma': 'scale',\n",
              " 'kernel': 'rbf',\n",
              " 'n_components': 10}"
            ]
          },
          "metadata": {},
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZgA9jJFYXBM5"
      },
      "source": [
        "Στα παρακάτω διαγράμματα βλέπουμε ένα διάγραμμα για κάθε υπερπαράμετρο που δοκιμάστηκε , και με πιο σκούρο χρώμα τις υπερπαραμέτρους που οδήγησαν στη βέλτιστη επίδοση."
      ],
      "id": "ZgA9jJFYXBM5"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 562
        },
        "id": "lpbcVEa0WUoS",
        "outputId": "724850ee-31c5-4edc-bbed-9ab10b7e6189"
      },
      "source": [
        "optuna.visualization.plot_slice(study)"
      ],
      "id": "lpbcVEa0WUoS",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"1d17b60e-9ed1-4066-8468-2ee4de77c531\" class=\"plotly-graph-div\" style=\"height:525px; width:1500px;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"1d17b60e-9ed1-4066-8468-2ee4de77c531\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '1d17b60e-9ed1-4066-8468-2ee4de77c531',\n",
              "                        [{\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": true}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [1.204552851257661, 0.20796091832380836, 0.5103197999781499, 0.32060168355070784], \"xaxis\": \"x\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [2, 1, 1, 1], \"xaxis\": \"x2\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y2\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"scale\", \"scale\", \"scale\", \"auto\"], \"xaxis\": \"x3\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y3\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"poly\", \"rbf\", \"linear\", \"rbf\"], \"xaxis\": \"x4\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y4\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [10, 10, 6, 6], \"xaxis\": \"x5\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y5\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Slice Plot\"}, \"width\": 1500, \"xaxis\": {\"anchor\": \"y\", \"domain\": [0.0, 0.16799999999999998], \"title\": {\"text\": \"C\"}, \"type\": \"log\"}, \"xaxis2\": {\"anchor\": \"y2\", \"domain\": [0.208, 0.376], \"title\": {\"text\": \"degree\"}}, \"xaxis3\": {\"anchor\": \"y3\", \"domain\": [0.416, 0.584], \"title\": {\"text\": \"gamma\"}}, \"xaxis4\": {\"anchor\": \"y4\", \"domain\": [0.624, 0.792], \"title\": {\"text\": \"kernel\"}}, \"xaxis5\": {\"anchor\": \"y5\", \"domain\": [0.832, 1.0], \"title\": {\"text\": \"n_components\"}}, \"yaxis\": {\"anchor\": \"x\", \"domain\": [0.0, 1.0], \"title\": {\"text\": \"Objective Value\"}}, \"yaxis2\": {\"anchor\": \"x2\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis3\": {\"anchor\": \"x3\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis4\": {\"anchor\": \"x4\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis5\": {\"anchor\": \"x5\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('1d17b60e-9ed1-4066-8468-2ee4de77c531');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1wleb2khXYR9"
      },
      "source": [
        "Εδώ βλέπουμε το διάγραμμα που συνδυάζει τις τιμές των υπερπαραμέτρων και την τιμή για το f1_score. Με το πιο σκούρο χρώμα φαίνεται η ευθεία που αντιστοιχεί στη βέλτιστη τιμή του f1."
      ],
      "id": "1wleb2khXYR9"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "rq7-JjftXU8Z",
        "outputId": "435bda27-94ef-4e3d-83a1-f4e18adf11ff"
      },
      "source": [
        "optuna.visualization.plot_parallel_coordinate(study)"
      ],
      "id": "rq7-JjftXU8Z",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"818b5255-30cf-432b-a1c4-5f00a81502a4\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"818b5255-30cf-432b-a1c4-5f00a81502a4\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '818b5255-30cf-432b-a1c4-5f00a81502a4',\n",
              "                        [{\"dimensions\": [{\"label\": \"Objective Value\", \"range\": [0.4310638160275016, 0.6600886858013236], \"values\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337]}, {\"label\": \"C\", \"range\": [-0.682018273455798, 0.08082585996409282], \"ticktext\": [\"0.208\", \"1\", \"1.2\"], \"tickvals\": [-0.682018273455798, 0, 0.08082585996409282], \"values\": [0.08082585996409282, -0.682018273455798, -0.29215758108099066, -0.49403420139902093]}, {\"label\": \"degree\", \"range\": [1, 2], \"values\": [2, 1, 1, 1]}, {\"label\": \"gamma\", \"range\": [0, 1], \"ticktext\": [\"scale\", \"auto\"], \"tickvals\": [0, 1], \"values\": [0, 0, 0, 1]}, {\"label\": \"kernel\", \"range\": [0, 2], \"ticktext\": [\"poly\", \"rbf\", \"linear\"], \"tickvals\": [0, 1, 2], \"values\": [0, 1, 2, 1]}, {\"label\": \"n_components\", \"range\": [6, 10], \"values\": [10, 10, 6, 6]}], \"labelangle\": 30, \"labelside\": \"bottom\", \"line\": {\"color\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"colorbar\": {\"title\": {\"text\": \"Objective Value\"}}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"reversescale\": false, \"showscale\": true}, \"type\": \"parcoords\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Parallel Coordinate Plot\"}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('818b5255-30cf-432b-a1c4-5f00a81502a4');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aK2dCKuDHC1F"
      },
      "source": [
        "###### Επιδόσεις καλύτερου μοντέλου SVM στο test set"
      ],
      "id": "aK2dCKuDHC1F"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JUW68tw6Y-Yv"
      },
      "source": [
        "###### Εντοπίζω τις επιδόσεις"
      ],
      "id": "JUW68tw6Y-Yv"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "39Rwhiu7Y-Yw"
      },
      "source": [
        "* Χρησιμοποιούμε το dictionary `study.best_params` για να εντοπίσουμε τις προβλέψεις του καλύτερου μοντέλου SVM από το σύνολο των προβλέψεων `predictions` και να το αξιολογήσουμε στο test set, αυτή τη φορά υπολογίζοντας όλες τις μετρικές."
      ],
      "id": "39Rwhiu7Y-Yw"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hUlaLfFUHCGp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "1cec3f09-9fa8-43d3-d10c-1dfe02faa0e4"
      },
      "source": [
        "# pretty print with sep='\\n'\n",
        "for pred in predictions:\n",
        "  print(pred)"
      ],
      "id": "hUlaLfFUHCGp",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'n_components': 10, 'kernel': 'poly', 'gamma': 'scale', 'C': 1.204552851257661, 'degree': 2, 'preds': array([2, 1, 2, ..., 1, 1, 1])}\n",
            "{'n_components': 10, 'kernel': 'rbf', 'gamma': 'scale', 'C': 0.20796091832380836, 'degree': 1, 'preds': array([2, 1, 0, ..., 1, 1, 1])}\n",
            "{'n_components': 6, 'kernel': 'linear', 'gamma': 'scale', 'C': 0.5103197999781499, 'degree': 1, 'preds': array([2, 2, 0, ..., 1, 2, 1])}\n",
            "{'n_components': 6, 'kernel': 'rbf', 'gamma': 'auto', 'C': 0.32060168355070784, 'degree': 1, 'preds': array([2, 2, 0, ..., 1, 2, 1])}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7jH_G3zr3_VG",
        "outputId": "49b73404-cf8e-4685-c60d-20705cf92f5c"
      },
      "source": [
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.metrics import recall_score\n",
        "\n",
        "best_predictions = None\n",
        "best_n_components = study.best_params['n_components']\n",
        "best_C = study.best_params['C']\n",
        "best_degree = study.best_params['degree']\n",
        "best_gamma = study.best_params['gamma']\n",
        "best_kernel = study.best_params['kernel']\n",
        "\n",
        "for pred_object in predictions:\n",
        "  if pred_object['n_components']==best_n_components and pred_object['gamma']==best_gamma and pred_object['C']==best_C and pred_object['kernel'] == best_kernel and pred_object['degree']==best_degree:\n",
        "    best_predictions = pred_object['preds']\n",
        "    break\n",
        "\n",
        "best_accuracies['SVM_pipe'] = round(accuracy_score(Y_test, best_predictions), 3)\n",
        "best_f1s['SVM_pipe'] = round(f1_score(Y_test, best_predictions, average='macro'), 3)\n",
        "best_recalls['SVM_pipe'] = round(recall_score(Y_test, best_predictions, average='macro'), 3) \n",
        "\n",
        "print('SVM with pipe best scores:')\n",
        "print(\"Accuracy:\", best_accuracies['SVM_pipe'])\n",
        "print(\"F1:\", best_f1s['SVM_pipe'])\n",
        "print(\"Recall:\", best_recalls['SVM_pipe'])\n",
        "\n"
      ],
      "id": "7jH_G3zr3_VG",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "SVM with pipe best scores:\n",
            "Accuracy: 0.679\n",
            "F1: 0.66\n",
            "Recall: 0.645\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "riNSPXALZGzu"
      },
      "source": [
        "###### Σχεδιάζω τον πίνακα σύγχυσης"
      ],
      "id": "riNSPXALZGzu"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Yf0OF2oYZISW",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "outputId": "2e4e3fc6-7738-4478-e332-2a4e1e1f198d"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns; sns.set()\n",
        "print(f\"SVM:\")\n",
        "mat = confusion_matrix(Y_test , best_predictions)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "Yf0OF2oYZISW",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "SVM:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "05119467-e1d6-4c35-bf68-e5cf88cb12d0"
      },
      "source": [
        "##### Με αναζήτηση πλέγματος με διασταυρούμενη επικύρωση"
      ],
      "id": "05119467-e1d6-4c35-bf68-e5cf88cb12d0"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fzjmj0-1sOYG"
      },
      "source": [
        "###### Δοκιμάσαμε με μεγάλο αριθμό από trials (10), αλλά η διαδικασία αργούσε υπερβολικά (10 ώρες χωρίς να έχει τερματίσει), με την παρακάτω έξοδο:"
      ],
      "id": "Fzjmj0-1sOYG"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cce1f6a2-d8a1-4bf3-9e39-02a3377a4db6",
        "outputId": "3650a2ec-4654-46de-b16e-757c667f835b"
      },
      "source": [
        "import optuna\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.model_selection import cross_val_score\n",
        "from sklearn.svm import SVC\n",
        "    \n",
        "param_distributions = {\n",
        "  \"kernel\":optuna.distributions.CategoricalDistribution(['rbf','poly','linear','sigmoid']),\n",
        "  \"C\":optuna.distributions.LogUniformDistribution(0.1,3),\n",
        "  \"gamma\":optuna.distributions.CategoricalDistribution(['auto','scale']),\n",
        "  \"degree\":optuna.distributions.LogUniformDistribution(1,3)\n",
        "}\n",
        "\n",
        "svm = SVC()\n",
        "optuna_search = optuna.integration.OptunaSearchCV(svm, param_distributions, cv=10, n_trials = 10, n_jobs = -1, refit=True)\n",
        "optuna_search.fit(new_X_train_df, Y_train.ravel())\n",
        "preds_cv = optuna_search.predict(new_X_test_df)"
      ],
      "id": "cce1f6a2-d8a1-4bf3-9e39-02a3377a4db6",
      "execution_count": null,
      "outputs": [
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/tmp/ipykernel_19269/2695495994.py:18: ExperimentalWarning: OptunaSearchCV is experimental (supported from v0.17.0). The interface can change in the future.\n",
            "  optuna_search = optuna.integration.OptunaSearchCV(svm, param_distributions, cv=10, n_trials = 10, n_jobs = -1, refit=True)\n",
            "\u001b[32m[I 2021-12-03 19:46:05,741]\u001b[0m A new study created in memory with name: no-name-26f010a7-9f19-4548-89e6-e49cd21034c3\u001b[0m\n",
            "/home/andreas/neural/lib/python3.9/site-packages/optuna/study/study.py:393: FutureWarning: `n_jobs` argument has been deprecated in v2.7.0. This feature will be removed in v4.0.0. See https://github.com/optuna/optuna/releases/tag/v2.7.0.\n",
            "  warnings.warn(\n",
            "\u001b[32m[I 2021-12-03 20:11:10,025]\u001b[0m Trial 1 finished with value: 0.4310732079649018 and parameters: {'kernel': 'sigmoid', 'C': 0.520455325772685, 'gamma': 'auto', 'degree': 1.5540801174820085}. Best is trial 1 with value: 0.4310732079649018.\u001b[0m\n",
            "\u001b[32m[I 2021-12-03 20:11:51,207]\u001b[0m Trial 8 finished with value: 0.4310732079649018 and parameters: {'kernel': 'sigmoid', 'C': 0.22224984422102514, 'gamma': 'auto', 'degree': 2.1982349555524863}. Best is trial 1 with value: 0.4310732079649018.\u001b[0m\n",
            "\u001b[32m[I 2021-12-03 20:21:12,980]\u001b[0m Trial 9 finished with value: 0.5650749167638602 and parameters: {'kernel': 'poly', 'C': 0.7359220993976273, 'gamma': 'scale', 'degree': 1.1135796969874636}. Best is trial 9 with value: 0.5650749167638602.\u001b[0m\n",
            "\u001b[32m[I 2021-12-03 20:29:25,501]\u001b[0m Trial 4 finished with value: 0.33049062915333083 and parameters: {'kernel': 'sigmoid', 'C': 0.17536240182524263, 'gamma': 'scale', 'degree': 1.0445830172394617}. Best is trial 9 with value: 0.5650749167638602.\u001b[0m\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2Augqrw3sc0h"
      },
      "source": [
        "###### Οπότε, στη συνέχεια δοκιμάσαμε με μικρότερο αριθμό από trials (4):"
      ],
      "id": "2Augqrw3sc0h"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lv1xWihqBI7P",
        "outputId": "76d35f01-8dc2-437c-9ca6-bea6b208072e"
      },
      "source": [
        "import optuna\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.model_selection import cross_val_score\n",
        "from sklearn.svm import SVC\n",
        "    \n",
        "param_distributions = {\n",
        "  \"kernel\":optuna.distributions.CategoricalDistribution(['rbf','poly','linear','sigmoid']),\n",
        "  \"C\":optuna.distributions.LogUniformDistribution(0.1,3),\n",
        "  \"gamma\":optuna.distributions.CategoricalDistribution(['auto','scale']),\n",
        "  \"degree\":optuna.distributions.LogUniformDistribution(1,3)\n",
        "}\n",
        "\n",
        "svm = SVC()\n",
        "optuna_search = optuna.integration.OptunaSearchCV(svm, param_distributions, cv=10, n_trials = 4, refit=True)\n",
        "optuna_search.fit(new_X_train_df, Y_train.ravel())\n",
        "preds_cv = optuna_search.predict(new_X_test_df)"
      ],
      "id": "lv1xWihqBI7P",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:18: ExperimentalWarning:\n",
            "\n",
            "OptunaSearchCV is experimental (supported from v0.17.0). The interface can change in the future.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 22:25:14,232]\u001b[0m A new study created in memory with name: no-name-85d872f1-3b9b-4ee6-8f83-d3f8639647bf\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 23:47:38,162]\u001b[0m Trial 0 finished with value: 0.4099364707556538 and parameters: {'kernel': 'sigmoid', 'C': 1.6612257445958554, 'gamma': 'scale', 'degree': 1.2032199776299406}. Best is trial 0 with value: 0.4099364707556538.\u001b[0m\n",
            "\u001b[32m[I 2021-12-03 01:12:25,940]\u001b[0m Trial 1 finished with value: 0.4099364719329017 and parameters: {'kernel': 'sigmoid', 'C': 2.2782807397614846, 'gamma': 'scale', 'degree': 1.0968076766273904}. Best is trial 1 with value: 0.4099364719329017.\u001b[0m\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9ogC3ChERQda"
      },
      "source": [
        "Τελικά οι καλυτέροι υπερπαράμετροι για την ταξινόμηση με SVM, με cross_validation είναι αυτές του trial 9:"
      ],
      "id": "9ogC3ChERQda"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pz9leJs7RQdb"
      },
      "source": [
        "optuna_search.study_.best_params"
      ],
      "id": "pz9leJs7RQdb",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zASqbGq5qheM"
      },
      "source": [
        "best_params_ = {\n",
        "    'kernel': 'sigmoid',\n",
        "    'C': 2.2782807397614846,\n",
        "    'gamma': 'scale',\n",
        "    'degree': 1.0968076766273904\n",
        "}"
      ],
      "id": "zASqbGq5qheM",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fls-x6qzRQdc"
      },
      "source": [
        "Στα παρακάτω διαγράμματα βλέπουμε ένα διάγραμμα για κάθε υπερπαράμετρο που δοκιμάστηκε , και με πιο σκούρο χρώμα τις υπερπαραμέτρους που οδήγησαν στη βέλτιστη επίδοση."
      ],
      "id": "fls-x6qzRQdc"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 562
        },
        "id": "QwZVYOfaRQdd",
        "outputId": "724850ee-31c5-4edc-bbed-9ab10b7e6189"
      },
      "source": [
        "optuna.visualization.plot_slice(optuna_search.study_)"
      ],
      "id": "QwZVYOfaRQdd",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"1d17b60e-9ed1-4066-8468-2ee4de77c531\" class=\"plotly-graph-div\" style=\"height:525px; width:1500px;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"1d17b60e-9ed1-4066-8468-2ee4de77c531\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '1d17b60e-9ed1-4066-8468-2ee4de77c531',\n",
              "                        [{\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": true}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [1.204552851257661, 0.20796091832380836, 0.5103197999781499, 0.32060168355070784], \"xaxis\": \"x\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [2, 1, 1, 1], \"xaxis\": \"x2\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y2\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"scale\", \"scale\", \"scale\", \"auto\"], \"xaxis\": \"x3\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y3\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"poly\", \"rbf\", \"linear\", \"rbf\"], \"xaxis\": \"x4\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y4\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [10, 10, 6, 6], \"xaxis\": \"x5\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y5\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Slice Plot\"}, \"width\": 1500, \"xaxis\": {\"anchor\": \"y\", \"domain\": [0.0, 0.16799999999999998], \"title\": {\"text\": \"C\"}, \"type\": \"log\"}, \"xaxis2\": {\"anchor\": \"y2\", \"domain\": [0.208, 0.376], \"title\": {\"text\": \"degree\"}}, \"xaxis3\": {\"anchor\": \"y3\", \"domain\": [0.416, 0.584], \"title\": {\"text\": \"gamma\"}}, \"xaxis4\": {\"anchor\": \"y4\", \"domain\": [0.624, 0.792], \"title\": {\"text\": \"kernel\"}}, \"xaxis5\": {\"anchor\": \"y5\", \"domain\": [0.832, 1.0], \"title\": {\"text\": \"n_components\"}}, \"yaxis\": {\"anchor\": \"x\", \"domain\": [0.0, 1.0], \"title\": {\"text\": \"Objective Value\"}}, \"yaxis2\": {\"anchor\": \"x2\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis3\": {\"anchor\": \"x3\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis4\": {\"anchor\": \"x4\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis5\": {\"anchor\": \"x5\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('1d17b60e-9ed1-4066-8468-2ee4de77c531');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "h2jCjeP_RQdd"
      },
      "source": [
        "Εδώ βλέπουμε το διάγραμμα που συνδυάζει τις τιμές των υπερπαραμέτρων και την τιμή για το f1_score. Με το πιο σκούρο χρώμα φαίνεται η ευθεία που αντιστοιχεί στη βέλτιστη τιμή του f1."
      ],
      "id": "h2jCjeP_RQdd"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "bImUjPPARQdd",
        "outputId": "435bda27-94ef-4e3d-83a1-f4e18adf11ff"
      },
      "source": [
        "optuna.visualization.plot_parallel_coordinate(optuna_search.study_)"
      ],
      "id": "bImUjPPARQdd",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"818b5255-30cf-432b-a1c4-5f00a81502a4\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"818b5255-30cf-432b-a1c4-5f00a81502a4\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '818b5255-30cf-432b-a1c4-5f00a81502a4',\n",
              "                        [{\"dimensions\": [{\"label\": \"Objective Value\", \"range\": [0.4310638160275016, 0.6600886858013236], \"values\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337]}, {\"label\": \"C\", \"range\": [-0.682018273455798, 0.08082585996409282], \"ticktext\": [\"0.208\", \"1\", \"1.2\"], \"tickvals\": [-0.682018273455798, 0, 0.08082585996409282], \"values\": [0.08082585996409282, -0.682018273455798, -0.29215758108099066, -0.49403420139902093]}, {\"label\": \"degree\", \"range\": [1, 2], \"values\": [2, 1, 1, 1]}, {\"label\": \"gamma\", \"range\": [0, 1], \"ticktext\": [\"scale\", \"auto\"], \"tickvals\": [0, 1], \"values\": [0, 0, 0, 1]}, {\"label\": \"kernel\", \"range\": [0, 2], \"ticktext\": [\"poly\", \"rbf\", \"linear\"], \"tickvals\": [0, 1, 2], \"values\": [0, 1, 2, 1]}, {\"label\": \"n_components\", \"range\": [6, 10], \"values\": [10, 10, 6, 6]}], \"labelangle\": 30, \"labelside\": \"bottom\", \"line\": {\"color\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"colorbar\": {\"title\": {\"text\": \"Objective Value\"}}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"reversescale\": false, \"showscale\": true}, \"type\": \"parcoords\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Parallel Coordinate Plot\"}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('818b5255-30cf-432b-a1c4-5f00a81502a4');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "E_vKyg9WRQde"
      },
      "source": [
        "###### Επιδόσεις καλύτερου μοντέλου SVM στο test set"
      ],
      "id": "E_vKyg9WRQde"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LBCOx4x3RQde"
      },
      "source": [
        "* Εντοπίζω τις επιδόσεις"
      ],
      "id": "LBCOx4x3RQde"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dbcFX8Q1UnJp"
      },
      "source": [
        "best_accuracies['SVM_cv'] = round(accuracy_score(Y_test, preds_cv), 3)\n",
        "best_f1s['SVM_cv'] = round(f1_score(Y_test, preds_cv, average='macro'), 3)\n",
        "best_recalls['SVM_cv'] = round(recall_score(Y_test, preds_cv, average='macro'), 3) \n",
        "\n",
        "print('SVM with cross validation best scores:')\n",
        "print(\"Accuracy:\", best_accuracies['SVM_cv'])\n",
        "print(\"F1:\", best_f1s['SVM_cv'])\n",
        "print(\"Recall:\", best_recalls['SVM_cv'])"
      ],
      "id": "dbcFX8Q1UnJp",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5wi05ivLRQdh"
      },
      "source": [
        "* Σχεδιάζω τον πίνακα σύγχυσης"
      ],
      "id": "5wi05ivLRQdh"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "id": "DDPxtPjbRQdh",
        "outputId": "2e4e3fc6-7738-4478-e332-2a4e1e1f198d"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns; sns.set()\n",
        "print(f\"SVM:\")\n",
        "mat = confusion_matrix(Y_test , best_predictions)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "DDPxtPjbRQdh",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "SVM:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DfwFYlLH9ZCR"
      },
      "source": [
        "#### Συμπεράσματα SVM"
      ],
      "id": "DfwFYlLH9ZCR"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dQCoT-w798rD"
      },
      "source": [
        "Γενικά, παρατηρήσαμε πως η ταξίνομηση με τα Support Vector Machines ή SVMs όπως λέγονται μπόρει να είναι μια επίπονη διαδικασία σε μεγάλα datasets όπως αυτό που εργαζόμαστε , με την βελτιστοποίηση να διαρκεί κάποιες ώρες (ανάλογα και τον αριθμό n_trials που χρησιμοποιούμε στην Optuna)."
      ],
      "id": "dQCoT-w798rD"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "32gKUi8X_cq5"
      },
      "source": [
        "##### Σύγκριση out-of-the-box με pipelines"
      ],
      "id": "32gKUi8X_cq5"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LBRT1OH9_HG-"
      },
      "source": [
        "Βλέπουμε πως η βελτιστοποίηση λειτούργησε και η υλοποίηση με pipelines μας δίνει ελαφρώς καλύτερα αποτελέσματα από την οut of the box υλοποίηση που είχαμε:\n",
        "\n",
        "\n"
      ],
      "id": "LBRT1OH9_HG-"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YHlkwyb5_mMN",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 229
        },
        "outputId": "ca91dcf9-b5e8-47fc-d66c-d0fb65e7b2dc"
      },
      "source": [
        "plot_scores(accuracies, f1_scores, recalls)"
      ],
      "id": "YHlkwyb5_mMN",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZPWv3U8eHlAE"
      },
      "source": [
        "##### Σύγκριση out-of-the-box με cross_validation\n",
        "\n"
      ],
      "id": "ZPWv3U8eHlAE"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9juB_l1FIP_k"
      },
      "source": [
        "Για το cross validation χρησιμοποίησαμε τη συνάρτηση OptunaSearchCV όπου παρατηρήσαμε και πάλι αργούς χρόνους για τα αποτελέσματα, γι'αυτό και ο χαμηλός αριθμός στo attribute n_trials."
      ],
      "id": "9juB_l1FIP_k"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xVLMjZDKH6R8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 229
        },
        "outputId": "7e0b60ff-c2ac-4179-d8cb-e901ba5d2297"
      },
      "source": [
        "plot_scores(accuracies, f1_scores, recalls)"
      ],
      "id": "xVLMjZDKH6R8",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Sdaf2lIGCPIU"
      },
      "source": [
        "#### Βελτιστοποίηση MLP"
      ],
      "id": "Sdaf2lIGCPIU"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lPE6dV54AomI"
      },
      "source": [
        "##### Με pipeline"
      ],
      "id": "lPE6dV54AomI"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "96RZI4tiAomJ"
      },
      "source": [
        "* Χρησιμοποιούμε:\n",
        "  * pipeline για να εισαγάγουμε την προεπεξεργασία\n",
        "  * την βιβλιοθήκη optuna για την βελτιστοποίηση των υπερπαραμέτρων των μετασχηματιστών και των εκτιμητών\n",
        "* Δοκιμάσαμε διάφορες αρχιτεκτονικές, χρησιμοποιώντας κάθε φορά διαφορετικό συνδυασμό μετασχηματιστών, και καταλήξαμε στην αρχιτεκτονική που φαίνεται στο παρακάτω κελί.\n",
        "* Ομοίως με προηγουμένως, δοκιμάζουμε διάφορες τιμές για τις υπερπαραμέτρους\n",
        "* Εδώ, όσον αφορά τις υπερπαραμέτρους του MLP, δοκιμάζουμε έως 2 hidden layers, και αριθμό units έως 150 στο κάθε στρώμα, ώστε το δίκτυο να μην γίνει υπερβολικά σύνθετο και καταλήξουμε σε overfitting."
      ],
      "id": "96RZI4tiAomJ"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NJvH25aU-HEW",
        "outputId": "40c3fb8e-207a-421f-9419-cec73c763f29"
      },
      "source": [
        "import optuna\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.neural_network import MLPClassifier\n",
        "from sklearn.metrics import f1_score\n",
        "\n",
        "# we keep all the predictions in a list, to restore the predictions of the best model later and count all of its metrics\n",
        "predictions = []\n",
        "\n",
        "def objective_mlp(trial):\n",
        "    \n",
        "    steps = []\n",
        "    \n",
        "    scaler = StandardScaler()\n",
        "    steps.append(('scaler', scaler))\n",
        "\n",
        "    #ros = RandomOverSampler()\n",
        "    #steps.append(('ros', ros))\n",
        "    \n",
        "    \n",
        "    n_components = trial.suggest_int('n_components', 2, 10, step=4)\n",
        "    pca = PCA(n_components=n_components)\n",
        "    steps.append(('pca', pca))\n",
        "    \n",
        "    n_layers = trial.suggest_int('n_layers', 1, 3, step=1)\n",
        "    layers = []\n",
        "    for i in range(n_layers):\n",
        "        layers.append(trial.suggest_int(f\"n_units_{i}\", 10, 150, step=10))\n",
        "    \n",
        "    mlp = MLPClassifier(hidden_layer_sizes=tuple(layers))\n",
        "    steps.append(('mlp', mlp))\n",
        "    \n",
        "    pipe = Pipeline(steps=steps)\n",
        "    \n",
        "    pipe.fit(new_X_train_df, Y_train.ravel())\n",
        "    \n",
        "    \n",
        "    preds = pipe.predict(new_X_test_df)\n",
        "    \n",
        "    predictions.append({\n",
        "        'n_components': n_components,\n",
        "        'hidden_layer_sizes': tuple(layers),\n",
        "        'preds': preds\n",
        "    })\n",
        "\n",
        "    score = f1_score(Y_test, preds, average='macro')\n",
        "    \n",
        "    return score\n",
        "\n",
        "\n",
        "study = optuna.create_study(direction='maximize')\n",
        "study.optimize(objective_mlp, n_trials=10)\n",
        "\n"
      ],
      "id": "NJvH25aU-HEW",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "\u001b[32m[I 2021-12-02 19:05:40,192]\u001b[0m A new study created in memory with name: no-name-91505291-ecbf-4df6-a5e1-9107cec7f390\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 19:05:51,118]\u001b[0m Trial 0 finished with value: 0.2940945758327918 and parameters: {'n_components': 2, 'n_layers': 1, 'n_units_0': 100}. Best is trial 0 with value: 0.2940945758327918.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:09:16,323]\u001b[0m Trial 1 finished with value: 0.6670789336792677 and parameters: {'n_components': 10, 'n_layers': 2, 'n_units_0': 70, 'n_units_1': 60}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:13:31,592]\u001b[0m Trial 2 finished with value: 0.6075977643393383 and parameters: {'n_components': 6, 'n_layers': 2, 'n_units_0': 70, 'n_units_1': 70}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:18:14,903]\u001b[0m Trial 3 finished with value: 0.6603022478036772 and parameters: {'n_components': 10, 'n_layers': 2, 'n_units_0': 60, 'n_units_1': 130}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 19:18:41,617]\u001b[0m Trial 4 finished with value: 0.2983449968667105 and parameters: {'n_components': 2, 'n_layers': 3, 'n_units_0': 10, 'n_units_1': 70, 'n_units_2': 140}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 19:21:42,011]\u001b[0m Trial 5 finished with value: 0.6660543026272889 and parameters: {'n_components': 10, 'n_layers': 3, 'n_units_0': 50, 'n_units_1': 10, 'n_units_2': 120}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:22:54,170]\u001b[0m Trial 6 finished with value: 0.6052030703575476 and parameters: {'n_components': 6, 'n_layers': 1, 'n_units_0': 60}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:26:57,660]\u001b[0m Trial 7 finished with value: 0.6099982431811632 and parameters: {'n_components': 6, 'n_layers': 2, 'n_units_0': 100, 'n_units_1': 50}. Best is trial 1 with value: 0.6670789336792677.\u001b[0m\n",
            "\u001b[32m[I 2021-12-02 19:28:40,665]\u001b[0m Trial 8 finished with value: 0.6675251389899245 and parameters: {'n_components': 10, 'n_layers': 3, 'n_units_0': 50, 'n_units_1': 20, 'n_units_2': 20}. Best is trial 8 with value: 0.6675251389899245.\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "\u001b[32m[I 2021-12-02 19:39:30,300]\u001b[0m Trial 9 finished with value: 0.6011041702983896 and parameters: {'n_components': 6, 'n_layers': 3, 'n_units_0': 90, 'n_units_1': 140, 'n_units_2': 30}. Best is trial 8 with value: 0.6675251389899245.\u001b[0m\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qZnnkll7EqB2"
      },
      "source": [
        "Τελικά οι καλυτέροι υπερπαράμετροι για την ταξινόμηση με MLP είναι:"
      ],
      "id": "qZnnkll7EqB2"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yecDv5eWEqB3",
        "outputId": "0b308aa0-30b9-4328-8c21-b2122a3aa3e1"
      },
      "source": [
        "study.best_params"
      ],
      "id": "yecDv5eWEqB3",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'n_components': 10,\n",
              " 'n_layers': 3,\n",
              " 'n_units_0': 50,\n",
              " 'n_units_1': 20,\n",
              " 'n_units_2': 20}"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G_bnZ4v_Fdce"
      },
      "source": [
        "Στα παρακάτω διαγράμματα βλέπουμε ένα διάγραμμα για κάθε υπερπαράμετρο που δοκιμάστηκε , και με πιο σκούρο χρώμα τις υπερπαραμέτρους που οδήγησαν στη βέλτιστη επίδοση."
      ],
      "id": "G_bnZ4v_Fdce"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 562
        },
        "id": "T4iVJVMdFdcf",
        "outputId": "417f7d60-8ec1-455e-b49e-a47591f5d6b2"
      },
      "source": [
        "optuna.visualization.plot_slice(study)"
      ],
      "id": "T4iVJVMdFdcf",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"c44dddad-3a73-4a9a-962b-ca62c159fc1a\" class=\"plotly-graph-div\" style=\"height:525px; width:1500px;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"c44dddad-3a73-4a9a-962b-ca62c159fc1a\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        'c44dddad-3a73-4a9a-962b-ca62c159fc1a',\n",
              "                        [{\"marker\": {\"color\": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": true}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [2, 10, 6, 10, 2, 10, 6, 6, 10, 6], \"xaxis\": \"x\", \"y\": [0.2940945758327918, 0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6052030703575476, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896], \"yaxis\": \"y\"}, {\"marker\": {\"color\": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [1, 2, 2, 2, 3, 3, 1, 2, 3, 3], \"xaxis\": \"x2\", \"y\": [0.2940945758327918, 0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6052030703575476, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896], \"yaxis\": \"y2\"}, {\"marker\": {\"color\": [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [100, 70, 70, 60, 10, 50, 60, 100, 50, 90], \"xaxis\": \"x3\", \"y\": [0.2940945758327918, 0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6052030703575476, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896], \"yaxis\": \"y3\"}, {\"marker\": {\"color\": [1, 2, 3, 4, 5, 7, 8, 9], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [60, 70, 130, 70, 10, 50, 20, 140], \"xaxis\": \"x4\", \"y\": [0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896], \"yaxis\": \"y4\"}, {\"marker\": {\"color\": [4, 5, 8, 9], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [140, 120, 20, 30], \"xaxis\": \"x5\", \"y\": [0.2983449968667105, 0.6660543026272889, 0.6675251389899245, 0.6011041702983896], \"yaxis\": \"y5\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Slice Plot\"}, \"width\": 1500, \"xaxis\": {\"anchor\": \"y\", \"domain\": [0.0, 0.16799999999999998], \"title\": {\"text\": \"n_components\"}}, \"xaxis2\": {\"anchor\": \"y2\", \"domain\": [0.208, 0.376], \"title\": {\"text\": \"n_layers\"}}, \"xaxis3\": {\"anchor\": \"y3\", \"domain\": [0.416, 0.584], \"title\": {\"text\": \"n_units_0\"}}, \"xaxis4\": {\"anchor\": \"y4\", \"domain\": [0.624, 0.792], \"title\": {\"text\": \"n_units_1\"}}, \"xaxis5\": {\"anchor\": \"y5\", \"domain\": [0.832, 1.0], \"title\": {\"text\": \"n_units_2\"}}, \"yaxis\": {\"anchor\": \"x\", \"domain\": [0.0, 1.0], \"title\": {\"text\": \"Objective Value\"}}, \"yaxis2\": {\"anchor\": \"x2\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis3\": {\"anchor\": \"x3\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis4\": {\"anchor\": \"x4\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis5\": {\"anchor\": \"x5\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('c44dddad-3a73-4a9a-962b-ca62c159fc1a');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HopAdg0EFdcf"
      },
      "source": [
        "Εδώ βλέπουμε το διάγραμμα που συνδυάζει τις τιμές των υπερπαραμέτρων και την τιμή για το f1_score. Με το πιο σκούρο χρώμα φαίνεται η ευθεία που αντιστοιχεί στη βέλτιστη τιμή του f1."
      ],
      "id": "HopAdg0EFdcf"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "fRmbUPqDFdcg",
        "outputId": "43f0d5c6-e79a-4c58-fb7a-8bbd1a642cfc"
      },
      "source": [
        "optuna.visualization.plot_parallel_coordinate(study)"
      ],
      "id": "fRmbUPqDFdcg",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"bab2c1c1-ff91-473a-a7f2-425e45dc63ce\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"bab2c1c1-ff91-473a-a7f2-425e45dc63ce\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        'bab2c1c1-ff91-473a-a7f2-425e45dc63ce',\n",
              "                        [{\"dimensions\": [{\"label\": \"Objective Value\", \"range\": [0.2940945758327918, 0.6675251389899245], \"values\": [0.2940945758327918, 0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6052030703575476, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896]}, {\"label\": \"n_components\", \"range\": [2, 10], \"values\": [2, 10, 6, 10, 2, 10, 6, 6, 10, 6]}, {\"label\": \"n_layers\", \"range\": [1, 3], \"values\": [1, 2, 2, 2, 3, 3, 1, 2, 3, 3]}, {\"label\": \"n_units_0\", \"range\": [10, 100], \"values\": [100, 70, 70, 60, 10, 50, 60, 100, 50, 90]}, {\"label\": \"n_units_1\", \"range\": [10, 140], \"values\": [60, 70, 130, 70, 10, 50, 20, 140]}, {\"label\": \"n_units_2\", \"range\": [20, 140], \"values\": [140, 120, 20, 30]}], \"labelangle\": 30, \"labelside\": \"bottom\", \"line\": {\"color\": [0.2940945758327918, 0.6670789336792677, 0.6075977643393383, 0.6603022478036772, 0.2983449968667105, 0.6660543026272889, 0.6052030703575476, 0.6099982431811632, 0.6675251389899245, 0.6011041702983896], \"colorbar\": {\"title\": {\"text\": \"Objective Value\"}}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"reversescale\": false, \"showscale\": true}, \"type\": \"parcoords\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Parallel Coordinate Plot\"}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('bab2c1c1-ff91-473a-a7f2-425e45dc63ce');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eJ8i_LXdIG7B"
      },
      "source": [
        "###### Επιδόσεις καλύτερου μοντέλου MLP στο test set"
      ],
      "id": "eJ8i_LXdIG7B"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Sdy8jkWHPQEG"
      },
      "source": [
        "###### Εντοπίζω τις επιδόσεις"
      ],
      "id": "Sdy8jkWHPQEG"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "koxIxbytIG7B"
      },
      "source": [
        "* Χρησιμοποιούμε το dictionary `study.best_params` για να εντοπίσουμε τις προβλέψεις του καλύτερου μοντέλου SVM από το σύνολο των προβλέψεων `predictions` και να το αξιολογήσουμε στο test set, αυτή τη φορά υπολογίζοντας όλες τις μετρικές."
      ],
      "id": "koxIxbytIG7B"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oVWW7C9aLwKq",
        "outputId": "9f9a8cad-15ce-421b-d473-b13c376ebc31"
      },
      "source": [
        "# pretty print with sep='\\n'\n",
        "for pred in predictions:\n",
        "  print(pred)"
      ],
      "id": "oVWW7C9aLwKq",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{'n_components': 2, 'hidden_layer_sizes': (100,), 'preds': array([1, 1, 2, ..., 1, 2, 1])}\n",
            "{'n_components': 10, 'hidden_layer_sizes': (70, 60), 'preds': array([2, 1, 0, ..., 1, 2, 1])}\n",
            "{'n_components': 6, 'hidden_layer_sizes': (70, 70), 'preds': array([2, 2, 0, ..., 1, 1, 1])}\n",
            "{'n_components': 10, 'hidden_layer_sizes': (60, 130), 'preds': array([2, 1, 0, ..., 1, 1, 1])}\n",
            "{'n_components': 2, 'hidden_layer_sizes': (10, 70, 140), 'preds': array([1, 1, 2, ..., 1, 2, 1])}\n",
            "{'n_components': 10, 'hidden_layer_sizes': (50, 10, 120), 'preds': array([2, 1, 0, ..., 1, 1, 1])}\n",
            "{'n_components': 6, 'hidden_layer_sizes': (60,), 'preds': array([2, 2, 0, ..., 1, 2, 1])}\n",
            "{'n_components': 6, 'hidden_layer_sizes': (100, 50), 'preds': array([2, 2, 0, ..., 1, 2, 1])}\n",
            "{'n_components': 10, 'hidden_layer_sizes': (50, 20, 20), 'preds': array([2, 1, 0, ..., 1, 1, 1])}\n",
            "{'n_components': 6, 'hidden_layer_sizes': (90, 140, 30), 'preds': array([2, 2, 0, ..., 1, 2, 1])}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wRaul4-bK5yI",
        "outputId": "6760fd8b-0fc3-4fa7-d152-7c62acae43f2"
      },
      "source": [
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.metrics import recall_score\n",
        "\n",
        "best_predictions = None\n",
        "best_n_components = study.best_params['n_components']\n",
        "\n",
        "best_units_tuple = None\n",
        "best_n_units_0 = study.best_params['n_units_0']\n",
        "best_n_units_1 = study.best_params.get('n_units_1')\n",
        "best_n_units_2 = study.best_params.get('n_units_2')\n",
        "if best_n_units_1:\n",
        "  if best_n_units_2:\n",
        "    best_units_tuple = (best_n_units_0, best_n_units_1, best_n_units_2)\n",
        "  else:\n",
        "     best_units_tuple = (best_n_units_0, best_n_units_1)\n",
        "else:\n",
        "     best_units_tuple = (best_n_units_0)\n",
        "\n",
        "for pred_object in predictions:\n",
        "  if pred_object['n_components']==best_n_components and pred_object['hidden_layer_sizes']==best_units_tuple:\n",
        "    best_predictions = pred_object['preds']\n",
        "    break\n",
        "\n",
        "best_accuracies['MLP'] = round(accuracy_score(Y_test, best_predictions), 3)\n",
        "best_f1s['MLP'] = round(f1_score(Y_test, best_predictions, average='macro'), 3)\n",
        "best_recalls['MLP'] = round(recall_score(Y_test, best_predictions, average='macro'), 3) \n",
        "\n",
        "print('MLP best scores:')\n",
        "print(\"Accuracy:\", best_accuracies['MLP'])\n",
        "print(\"F1:\", best_f1s['MLP'])\n",
        "print(\"Recall:\", best_recalls['MLP'])\n",
        "\n"
      ],
      "id": "wRaul4-bK5yI",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MLP best scores:\n",
            "Accuracy: 0.683\n",
            "F1: 0.668\n",
            "Recall: 0.651\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AJSU7zW9T-AS"
      },
      "source": [
        "###### Σχεδιάζω τον πίνακα σύγχυσης"
      ],
      "id": "AJSU7zW9T-AS"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "id": "wImTqKTFUCPG",
        "outputId": "89b5e17b-54a6-4896-9da3-c08969194683"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns; sns.set()\n",
        "print(f\"MLP:\")\n",
        "mat = confusion_matrix(Y_test , best_predictions)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "wImTqKTFUCPG",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MLP:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQwAAAEMCAYAAAAxjIiTAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8/fFQqAAAACXBIWXMAAAsTAAALEwEAmpwYAAAoDElEQVR4nO3dd1xV9R/H8de9l71BWSLinkWSK3OG2zTNclVGzqQUNWc4KNyaO/fKVZZpliIhaJpbM1NT3JspAqJsuPf3B79ukaJH83Kv+nk+Hj6694zv+Xxv8Oac7z1DpdPpdAghhAJqYxcghHh6SGAIIRSTwBBCKCaBIYRQTAJDCKGYBIYQQjEzYxfwqILKdjV2CSZtQexeY5cgnnJ5OTFFzpM9DCGEYhIYQgjFJDCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMQkMIYRiEhhCCMUkMIQQiklgCCEUk8AQQigmgSGEUEwCQwihmASGEEIxCQwhhGISGEIIxSQwhBCKSWAIIRSTwBBCKCaBIYRQTAJDCKGYBIYQQjEJDCGEYhIYQgjFJDCEEIpJYAghFJPAEEIoJoEhhFBMAuNfXMt6MOPsGnrMGgBApfo1GPXzdKacWMHkY8vovXgoju7O+uXNLMx4Z1p/pp1cyYQji3mt9+v6eWX9KvHRmtFM/mM5k44upef8ITi4OhV3lwyqS5c3OHliF7dTznM2eh8NG9TF3Nycb9cv4cK5g+TlxNCkcf1C6wwK6su5M/tJTjrDtStHmTH9MzQajZF6YFj3+3x8fEqTlxNDavI5/b/RwYP167z9dnv27P6RtNQL7IjcYLzi78PM2AWYms6hvbh2/KL+ffz5Gyx4fxJpiSmYWZjR9pOudJnQh6V9pwPQZnBnXMt6ENJgAA6uTgxcP474CzeI3n0ca0db9n+zgzOBM8nPy6dzaC/e/SKQhQGTjdW9J6p5s0ZMnjiad94N5PCRY3h6uuvn7dt3mLlzl7L+m8X3rLdl63a+WvUtt2+n4ezsxHfrlzBwQG9mz1lSnOUbXFGfj7l5wa9dCddq5Ofn37NeSnIqc+Yto2qVCrzWtEGx1vwwEhj/8HL7V8lMy+Dy7+co6eMBwJ2k24WW0Wm1uJb10L+v+1Zj1g1bSGZaOplp6ez/Zgd1325K9O7jRO/6o9C6v66KIOjbEIP3o7iEjBvGhImzOHT4dwBiY+P18+bOWwZAfr72nvUuXbqqf61SqdBqtVSoUNawxRpBUZ+Pj0/pB663Y+ceAHr17G7YAh9DsR2SpKSkEB0dTXR0NCkpKcW1WcWs7KxpO6QzP0xYfc8851IlmHJiBV+cWYN/33ZELf4JAGsHWxzdXYiJ/vsXIDb6Kp6V7v8DUbFeNeLP3zBMB4qZWq2mVi1fXF1LcOb0Xq5c+o05sydgZWWlaP1u3TqSnHSGxPg/8fWtztJlaw1ccfFS8vlcunCIK5d+Y9nSmZQo4fyA1kyHwQPj2rVrBAQE0LJlS4YNG8awYcNo2bIlAQEBXLlyxdCbV6zt0C4c/O4XUuOT75mXEnuLUb69CPbrQ9iMb0m8GAuApW3B//zMOxn6ZTPvZGBpd+8vTamqZWgV9BabJz0bvxju7q5YWFjQqdPrNPXvRK06LfCr+QKjgwcpWn/9+s24lKxK1eoNWbJ0DQkJNw1ccfF60OeTlJRMvVfaUL5iPeq+0hp7ezvWrPrS2CUrYvDAGDFiBG+99RaHDh0iLCyMsLAwDh06RKdOnRg5cqShN6+IV3UfqjR4kV+Whz1wuYzb6RzauJs+S4ah1qjJTs8CCvZO/mJlZ0323axC65X0caf/V5+y6fOvuHTkzJPvgBFkZhb0cf6ClcTHJ3LrVgqz5iyhdWv/R2rnwoXLnDp9li/nTTJEmUbzoM8nPT2Do7+fID8/n8TEJIIGjaZly6bY2dkaueqHM/gYRmpqKm+88UahaWq1mg4dOrBw4UJDb16RSq/UwKW0K5/vXwCApY0VKo0aj0qlmd5uVKFlNRoNDq5OWNlZk3E7ndsJyXhV8+Hs3pMAeFXzIe4fhx3OXiX5eN0YIuZt5MgPe4qvUwaWmnqb69dj0el0+mn/fP0ozMzMKF++7BOqzDQ8yufz13S12vS/tDR4hU5OTmzduvWeD+6nn37CwcHB0JtXZN/XUYQ2CWJq2xFMbTuCvesiOb3zdxa+PxHfVnVxK++JSqXCzsWeN8e8z/U/L5NxOx2Aw5t+pdXATlg72OJWoRT1uzfj8Pe7AHB0d2bg12PZsyqCfeuijNhDw1i1+ls+/qgnrq4lcHJyZHBQX7ZtK+inhYUFlpaW/39trn8NBYN5rq4lAKhWrRIjRwzgl517i78DBlbU51O3jh+VK1dApVLh4uLM7Fnj2bVrP2lpd4CC4LC0tMTMzKzQa1Ng8CqmTJlCSEgIoaGhuLsXfK2UkJBA1apVmTJliqE3r0huVg65WTn69zkZWeRm53I3+Q5OHi68OaYHdiUcyE7P4vzB0yz78Av9suGzNtBlQh8+3/clOVk57Fj0E9G7jwNQv1szSvp40GZwZ9oM7qxfZ3iNgOLrnAFNmDibEiVciD61h6ysbDZ8v4VJk+cCcPrPXylb1huA8G3fAFChUj2uXr3Bq6/WYXzoSOzsbLl58xYbN25l3GfTjdYPQynq8+nYsTUTQkfh5laStLQ7RO3Yw7s9PtKv9967b7Ni+Sz9+/Q7l1i1+jt69xlijG4UotI97n7kI0pOTiYuLg4AT09PXFxcHqudoLJdn2RZz5wFsc/eX2pRvPJyYoqcV2z7OS4uLo8dEkII02D6oyxCCJMhgSGEUEwCQwihmASGEEIxCQwhhGISGEIIxSQwhBCKSWAIIRSTwBBCKCaBIYRQTAJDCKGYBIYQQjEJDCGEYhIYQgjFJDCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMQkMIYRiEhhCCMUkMIQQihX5XBKtVquogafheZBCiCejyMCoXr06KpWqyBV1Oh0qlYro6GiDFCaEMD1FBsaOHTuKsw4hxFOgyMDw8vK6Z5pWqyUpKQk3NzeDFiWEME2KBiDS0tIYOnQovr6+tGzZEijYA5k1a9ZD1hRCPEsUBUZISAh2dnbs3LkTc3NzAPz8/AgPDzdocUII06Lo6e0HDhxgz549mJub6wdCXVxcuHXrlkGLE0KYFkWBYW9vT0pKSqGxi9jYWFxdXQ1WWFHO5N8u9m0+TTJj9xi7BJP3ee0xxi7hqaXokKRz584EBQVx8OBBtFotx44dY+TIkXTr1s3Q9QkhTIiiPYy+fftiaWlJaGgoeXl5BAcH07VrVwICAgxdnxDChCgKDJVKRUBAgASEEM85RYEBBQOfYWFhJCYm4ubmxuuvv079+vUNWZsQwsQoGsNYsWIFn3zyCY6OjjRp0gQnJyeGDh3KihUrDF2fEMKEKNrDWLlyJatWraJy5cr6aR06dKBnz5706tXLYMUJIUyL4ktNfXx8Cr339vZ+4MVpQohnT5GBodVq9f8GDhxIcHAwV65cISsri8uXLzN27FiCgoKKs1YhhJEpurxdp9MBEBYWVmja1q1b6dy5czGUKYQwBXJ5uxBCsUe6vF0I8XxTfB7Gjh07OHLkCCkpKfpDFIBp06YZpDAhhOlR9C3Jl19+SUhICFqtlp9//hknJyf27t2Lg4ODoesTQpgQRYGxceNGVqxYQXBwMObm5gQHB7No0SJu3Lhh6PqEECZE8R23/jppy9zcnNzcXHx9fTly5IhBixNCmBZFYxhlypTh/PnzVKpUiUqVKvHNN9/g4OCAo6OjoesTQpgQRYExePBgUlNTARg6dCjDhg0jIyODkJAQQ9YmhDAxigKjSZMm+tcvvfQSkZGRBitICGG6igyM69evK2rA29v7iRUjhDBtRQZGixYtUKlUhc65+Dd58pkQz5ciA+PMmTPFWYcQ4ikgT1IWQigmgSGEUEwCQwihmASGEEIxCQwhhGJFfkvSpEkTRffs3LVr15OsRwhhwooMjOnTp+tfnzx5ks2bN9OjRw9KlSpFbGwsa9eupWPHjsVRoxDCRBQZGHXr1tW/Dg0NZfny5bi7u+unNW7cmD59+shjBoR4jigaw0hMTMTGxqbQNBsbGxISEgxSlBDCNCm6+Mzf35/AwEACAwPx8PAgLi6OxYsX4+/vb+j6itXIOSOo2aAmVjaWpNxM4buF3/Pz+p8BqNmgJgMnfIyrlytnjp3li09mkBiTCEDjdo14s/ebVKhRnrN/nGN4lxH6Nl+oW4OJqycU2o61rTWh/cazN3xf8XXuMXz9/U9s3hbF+UuXadu8KRPHDAXg4uWrfDp+Bjdi4wCoXqUinw7uT4VyBc+uycnJYfLsxez4dT95eXn4+VZn3PCBuLuWfGC7/7ZwxTrmL1/L0tmTqF/Hrxh6/GSUKOvBgIgpnNp2mO+HLKDJRx1o/HEH/Xy1Ro3GwowptQLJSLmDtaMtb0zsRYUGL6DTwYVfT/DTmBVk380EwKl0STpN/5DSNStyO/YWW8d9xcV9fxqlb4oC4/PPP2fevHmEhITon63aunVrBgwYYOj6itX6+d8yc/gscnNy8a5QmunfTePiqQsk3EgkZMlYZo6YzcGog3wwLIDRCz5lUIchANxJvcsPy3/Au4I3NRvULNTmn4dP0aHqm/r3vq/4ErryM37b9Vtxdu2xuJYswYcfdGPfoaNkZ+cUmj5r4mhKebih1Wr5ZtNWhoVM4YfVCwFYu+FHjp+KZtPqBdjb2vLZtLlMmrmQOZPHPrDdf7p2I5btv+zFtYSL4Tv6hLUP/YCY45f073cv+JHdC37Uv/cf/BZl61YlI+UOAM2HdcHawZYZjQaDCt5ZOAT/wW8RPmEtAF3mDuT67+dZ/cE0Kr9Wk24LBzGr6SdkJN8p1n6BwkMSS0tLhg0bRlRUFCdOnCAqKophw4ZhZWVl6PqK1dVzV8nNyQVApyv45+njScM2Dbhy7ip7wvaQm53LmplrKF+9PN4VSgNwbO8xft26h1sJyQ/dRovOzdmzbS9ZmdkG7cuT0KJpA5o1fhUnx8L3bnWwt8PL0/3/FyeCRq3m+o04/fwbsfE0qFuLki7OWFpa0LpZYy5evvrQdv9p4swFDAnsibm54vtUm4QX29cnMy2DS/tPFblMzU6NOLbxV/17Z29XTm//jey7mWTfyeR0xBHcKhf8bJUo50GpGmXZMet78rJzOf3zERLOXKdGm7pFNW9Qis/D2LdvH8HBwfTv3x8o+ObkwIEDBivMWAZO/Jifzm1mxe5lJCcmc3jnEXwq+3Ap+u+/GFmZ2cRdjcOnss8DWrqXlbUljdo2JHJD1JMu2yjqt3qbWv5vMGnWQvq+31U/vVO7Vhw7eZrEm7fIzMoibPsvNHyltuJ2I3buwcLcnMavGueX4nFZ2lnTbMjb+j2D+ylbtyq2JRw4FX5YP+3Q6kiqNHsZKwdbrBxsqdGmLud2/QGAW+XSJF9PJCc9S798fPQ13CqVNlg/HkRRfK9Zs4bVq1fTuXNnIiIiALCysmLixInUr1/foAUWt3mj5zN/7EKq1arGS/V9yc3JxdrWmtRbqYWWS09Lx9rO5v6NFKFBm4akJadx4uCJJ1ix8RyI+J6MzCx+Co/C08NNP93H2wsPt5L4d3wPjUZNpfJlGT13iqI209MzmLP4K5bOnmSosg2m2dDOHP1uF2nxRe9p+r3ViFPhh8jJ+HsPM/bPy5iZmxH8x2IALu07xeE1BTepsrSxIvtORqE2su5k4ODhbIAePJyiPYxVq1axcuVK+vXrh1pdsEr58uW5fPnyf9p4+/bt/9P6hqLVajl15BQlPUrSvkc7MtMzsbX/17dE9jZk3s0oooX7a/F2cyI3Pht7F3+xsbaiS8e2BI//glspqQBMmDGfnNxc9oV/x5GoH2jepAH9h45V1N6CFeto38ofL0/3hy9sQjyq+1ChwQvsX76tyGXMrSyo0bYexzbuKTS92/xBJF2OY3yN3kx4oTfJ1xJ4e9bHAGRnZGFpZ11oeUs7a7LvZmEMivYw0tPT8fT0BNCf/ZmXl4e5uflD171w4UKR81JSUpRs3mg0Zho8fTy5eu4qLd5urp9uZW2pn66Uq2dJXqrvy5xP5xqiVKPSanVkZWWTeDOJEs5OnD1/iaAPA3B0sAfgnbff4Mtla0hJvY2z04NvHH3wtz9IuJnE+h/CAEhJvc3QsZPo/V5ner/XxeB9eVzlXqmGc+mSDNs/DwALGyvUGjVulbxY0G40ANVb1SHzdjqXD5wutK5HdR+2jFtJ7v/HtQ6v20HfDQX3y008dwPnMm5Y2FrpD0s8q5Xh+E/7i6trhSgKjDp16rBkyRICAwP101avXk29evUeum67du3w8vK67527/rqxsClwKuFIzQY1ORh1iJysHPwa+fFah6ZMGjCF6KPR9B3dh4ZtGnBo52HeHfwul6Mvc/1iwXNZ1Go1GnMNGjM1arUKc0tztPla8vPy9e03e6sZp4+eJu5qXFElmJy8vHzy8/PJz9eSr9WSnZ2DRqPh8O/HcXZyoHKFcmRmZTF3yWoc7O0o71MGgBeqVean8B3U8fPFysqS9Zu24layhD4simrXzEzD8rmTycvL09fQtc8gRgzsR6NHGAMxht++3snJLX+P6TXs+zpOpV35acwK/TS/txrxx7/2LgBiTlykVtfXiJj8NQB1uvuTcOYaALcuxxN/+ir+gzoRNWMDlZq+hHu1MpwKnG3YDhVBUWCMGTOG/v37s2HDBtLT02nVqhW2trYsXrz4oet6eXnx9ddfFzpL9C//vLmwsel00K7H6wRNGohKrSIxJpGFny3iYORBAEI/nMCA8R8xcu4Izhw7y6SP/z4mb/ZWM4bP/PtcgrALW9i+IZIvPpmhn9bireZsWPR98XXoCVi86hsWrlinf781YieBvd6lYjkfJs9aSPzNJKwsLXihWhUWzZyApaUFAMMG9GHyrIW83rU3uXl5VCzvo/9K9UHtftz7vXu+OdGo1TjY22FjU3i33NTkZuWQm/X3V8Q5GdnkZefqv/q0d3em3Ks1+GnsynvW/WH4El7/LIARB78ElYobxy+ycegi/fxvB87jrS/6M/r4UlJjklgfOMcoX6kCqHQPumnnP+h0Ok6ePElMTAyenp74+vrqxzMeZOrUqbRo0YKXX375nnkTJkxgzJgxj1RwS+/Wj7T88ybs2AJjl2DyPq/9aD9zz5sJV74ucp6iQc/AwEBUKhW+vr60adOGmjVrolarFZ24NXLkyPuGBfDIYSGEMC5FgXHo0KH7Tj98+PB9pwshnk0PHMOYM2cOALm5ufrXf7l+/TqlSpUyXGVCCJPzwMCIj48HCsYv/nr9F09PTwYOHGi4yoQQJueBgTF58mQA/Pz86NLFdL8DF0IUD0VjGBYWFvc82OjMmTNs3rzZEDUJIUyUosCYM2eO/kzPv3h4eNwzriGEeLYpCoy7d+9iZ2dXaJq9vT1paWkGKUoIYZoUBUaFChX0V6n+JTIykgoVKhikKCGEaVJ0aviwYcPo168f4eHheHt7c+3aNQ4cOMCSJUsMXZ8QwoQo2sOoXbs2W7Zs4cUXXyQzMxNfX1+2bt1KrVq1DF2fEMKEKL7/mZeXF/369TNkLUIIE1dkYIwdO5bx48cDMHz48CKfgjZt2jTDVCaEMDlFBkbp0n/fM9DH59HuXSmEeDYVGRgffvih/vWz9jgBIcTjKTIwlN4R/Fm7CbAQomhFBsbo0aMLvU9MLHjKl5OTk/7Weu7u7uzYscNw1QkhTEqRgbFz507960WLFpGamsqgQYOwtrYmMzOTuXPn4uTkVBw1CiFMhKLzML766iuGDh2KtXXBfRWtra355JNPWLny3vsTCiGeXYoCw8bGhhMnCj985+TJk/oAEUI8HxSduBUUFESfPn3w9/fHw8OD+Ph4fvnlF8aNG2fo+oQQJkRRYHTs2JEXXniBiIgIEhMTKVeuHIGBgVSsWNHQ9QkhTIjiU8MrVqxI+fLlSUpKws3N7eErCCGeOYrGMNLS0hg6dCi+vr60bNkSgB07djBr1iyDFieEMC2KAiMkJAQ7Ozt27typf56qn58f4eHhBi1OCGFaFB2SHDhwgD179mBubq6/CM3FxYVbt24ZtDghhGlRtIdhb29/z5PWY2NjcXV1NUhRQgjTpCgwOnfuTFBQEAcPHkSr1XLs2DFGjhxJt27dDF2fEMKEKDok6du3L5aWloSGhpKXl0dwcDBdu3YlICDA0PUJIUzIQwMjPz+f4OBgxo8fLwEhxHPuoYckGo2Gffv2FXnHLSHE80PRGEZAQADz5s0jJyfH0PUIIUyYSqfT6R62UJMmTUhKSkKtVuPi4lJob2PXrl2GrO8edjblinV7T5vqTmWMXYLJC/czdgWmrUTY7iLnKRr0nD59+hMrRgjx9FIUGHXr1jV0HUKIp4CiwMjJyWHhwoWEhYWRmJiIm5sbbdu2JTAwEEtLS0PXKIQwEYoC47PPPuPy5cuMHj0aLy8vYmJiWLx4MQkJCUyePNnQNQohTISiwNixYweRkZE4ODgABZe6v/TSS/orV4UQzwdFX6uWLFmSzMzMQtOys7PlWhIhnjOK9jA6dOhAnz596NGjB+7u7sTHx7Nu3To6dOhQ6Pkl8owSIZ5tis7D8Pf3f3hDKlWxPKNEzsN4MDkP4+HkPIwH+8/nYfzzGSVCiOeXojEMIYQACQwhxCOQwBBCKCaBIYRQTAJDCKGYBIYQQjEJDCGEYhIYQgjFJDCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMQkMIYRiEhhCCMUkMIQQiklgCCEUk8AQQigmgSGEUEwCQwihmARGEapUqUDYtnXExB3n+MlfaP/G309569TpdY7+Hklcwkl+O7qddu1b6OdVr16ZzT+u4uq1o9zNuGyM0g3C3MKcMTNG8OPhb/nlXDhrI5dR/7V69yzXe0gAh2N3U6dRLf00V4+STF85kchTW9jy2wY69Xij0DpqtZr+I3oT9vtGfjkXzprty7BzsDN4nwxB7eaB/WdTcf52K85rN2HbfxCoNQCY+frhOGcpzhu24bT8Gyxbt79vG7aDRlIibDdqTy/9NIfJs3H5YTsu34fj8n04TovXFEt//k3RYwaeNxqNhvXfLWX5snW0b9eDRo3q8d33y2hQvx3pGRksWzGTrl36Ebl9N61av8aatfOpUa0RN2/eIjc3j02bwli6dC3ffrfE2F15YjQaDQmxifTvNIj4mAQaNHuFSYs/4x3/nsTdiAfAy6cUzdo15WZ8UqF1P583hvOnLzCq7zjKVy7Lgg2zuXrxOkf3HwOg37Ce+NZ+gd7tPyI+JoHyVcqRk51T7H18Emw/GoL2dgop73VCZWeHw4QZWL3ekaxtm7EfM4GMFYvI/nkLmkpVcZw8i7yzp8m/fFG/vln1F9F4lrpv2+kL55C9Pay4unJfsodxH1WqVMDT040v5y1Hq9Wye/cBDh44Svd33sTLy5PbqWlEbi942EvEz7+QkZ5BuXIFDxA6f/4Sq1d9R/Tpc8bswhOXlZnF0hlfEXcjHp1Ox96oA8Rei6Oqb2X9MiMmDebLiYvIzc3VT7O2saZ2Az9WzllDfl4+509fZGfYbtp3awuAvaMd3fq+zcTh04mPSQDg0tnLT21gqD08ydnzC+TmoEtJJvfoITQ+ZVHZO6C2tSP7l+0A5J8/Q/71a2jKlP3Hyhps+w8ifdEc4xSvgASGQiqViurVK/P70ROcPXuRtq83R61W0659C7JzcvjzzzPGLrFYuZR0pkz50lw6dwWAZu2akpOTy/6dhwotp1L99V/VP6apqFC14Al2FauWJz8vn2btmhD+xya+37OWtz/oWBxdMIiszRuwbNwMLC1RlyiJea165B49jC41hexdUVg2bwNqNWZVa6B2cyfv1An9ulYdO5P753Hyr1y6b9s2H/TD+esfcZj+JWYv1iymHhVm8MBISUlh9OjR9OrVi3Xr1hWaN3DgQENv/rGcO3eJmzdvMXjIh5iZmeHfrBENG9XF2toarVbL119vYsXK2SSnnmXFyjkEDRxNRkbmwxt+RmjMNITOH0PYhgiuXriGja01gaP6MnPsvHuWzUjP5I/DJ+g9JAALSwuqvFiJ19o2xsraEgC3Um7YO9pTprw3HV/pxqh+4+g7tCd1G9cu7m49EXmnTqDxKYvLhm04r95I3oWz5BzYA0D27h3YdA/AZXMkDtPmkrF6GdqkmwCoS7pi1aY9mWtX3LfdjK8Wk9K7Gynvv032z1twGDcZtcf9D10MyeCBERISgqOjI926dSMqKooBAwaQl5cHwPXr1w29+ceSl5dHt64f0rr1a1y8fJigoD5s2riNmNg4mr7WgPETRtGmdXecHSvTulU35i+Ywou+1YxddrFQqVR8Pm80uTl5TB89G4C+Q3sSvnG7fizj38Z9PIFS3h5s+W0DIyd/ws+bIkmMLfhFyc7MBmDZzFVkZ+VwIfoSkT/upIH/vQOqJk+lwj50Gjn7fyW5U2uSu7VHbWePTc/+qEuXwX7kOO7OnERyh+akBn6A9VvdMa/zCgA2/QaS8c1qdBnp920672w0ZGZCXi7ZOyLIjT6Jxf/XLU4GD4wrV64wYsQIWrZsyYoVK3B1deXDDz8kOzvb0Jv+T079eYbWrbrh4/0yHTsEULacN0d/O46vb3X27zvMsd9PotPp+P3oCY4c+YPXXmto7JKLxZiZI3Ep6cKovmPJz8sHoE6jl+nauxPhf2wi/I9NuJdyY9Kiz3j/4+4AxMck8EnAp7R6sQO92gXi5OLIqT+iATgfXTDgp+PvZ4IreD64SVLZO6Bx8yBryw+Ql4vuThpZkeGY166HmU858mNukPv7EdDp0MZcJ+fIASxqFQSj+UsvY9urP85rN+G8dhMAjjMWYNGk+f03pgNQ3X+eARk8MP45AKZSqQgJCaFy5cr069fPpEOjxgtVsbS0wNraiqBBffHwcGPtmo38fvQEr75aR79H4ftSdV59tU6hMQxLSwssLMz/8drCKH140kZN+YSyFcswNOBTsrP+HpT8uMsndH+tJ++16MN7LfqQlHCLKSNnsGHlZgDKVvTBxtYaM3MzWndqQb0mtfl68XcAxFyN5djB4/QM6oG5hTllK/rQooM/e6MOGKOL/4ku7Tb58bFYvt4B1BpUtnZYNWtN/pVL5F08j6aUF2a+BY+OV3uUwqJuffL+P16R2u89Ugf2JnVgH1IH9gHgzuefknPgV1S2dpi/XAfMLUCtwaJpc8xf8CXn6KEiazEUg3+t6u3tzZEjR6hTp45+2siRI5k5cyZLly419OYfW/fubxLwQVfMzc3Yv/8Ib7TrQU5ODnv3HmLSpDmsXbcAN7eSJCUl88X0BezcUXCcWqaMF6fP7NW3cyvlLFev3qBGtUbG6soT4eHlTqf3O5CdlU348U366ZNHzCDih6hCy+bn55N2+w6Z/x/XeaVpHXoO6oGVtSVn/zzPoHdGkJp8W7/8mI9CGTNjBJGnfiIlKZXF05ZzZO/vxdOxJ+zOxLHY9h2A9dvvQH4+uSeOkb70S3SpKdydPQ3bD4PQuHmgzbhLzq4osiO2AqC7nXpPW9q025CTA1bW2PTog6Z0GXTafPJvXOPO+DFoY28Uc+9ApTPw/l9qaioqlQpHR8d75l24cIGKFSs+Unt2NuWeVGnPpOpOZYxdgskL9zN2BaatRNjuIucZfA/DycmpyHmPGhZCCOOS8zCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMQkMIYRiEhhCCMUkMIQQiklgCCEUk8AQQigmgSGEUEwCQwihmASGEEIxCQwhhGISGEIIxSQwhBCKSWAIIRSTwBBCKCaBIYRQTAJDCKGYBIYQQjEJDCGEYhIYQgjFJDCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMZVOp9MZuwghxNNB9jCEEIpJYAghFJPAEEIoJoEhhFBMAkMIoZgEhhBCMQkMIYRiEhhCCMUkMIQQiklgPKbLly/TtWtXWrVqRdeuXbly5YqxSzIpU6dOxd/fnypVqnDu3Dljl2NyUlJS6Nu3L61ataJ9+/YMGDCA5ORkY5f1UBIYjykkJIR33nmHiIgI3nnnHcaNG2fskkxKs2bNWLduHV5eXsYuxSSpVCr69OlDREQEW7Zswdvbmy+++MLYZT2UBMZjuHXrFqdPn6Zdu3YAtGvXjtOnTz8VfyGKS+3atfH09DR2GSbLycmJevXq6d/XrFmT2NhYI1akjATGY4iLi8Pd3R2NRgOARqPBzc2NuLg4I1cmnkZarZZvvvkGf39/Y5fyUBIYQhjZ+PHjsbGx4b333jN2KQ9lZuwCnkaenp4kJCSQn5+PRqMhPz+fxMRE2QUXj2zq1KlcvXqVRYsWoVab/t9v06/QBJUoUYJq1aqxdetWALZu3Uq1atVwcXExcmXiaTJz5kz+/PNP5s+fj4WFhbHLUURuoPOYLl68yKhRo0hLS8PBwYGpU6dSvnx5Y5dlMiZMmMD27dtJSkrC2dkZJycnwsLCjF2WyTh//jzt2rWjbNmyWFlZAVC6dGnmz59v5MoeTAJDCKGYHJIIIRSTwBBCKCaBIYRQTAJDCKGYBIYQQjEJDGFQhw4donHjxoqW3bRpE927d3+s7fyXdYVyEhjPGX9/f/bv32/sMsRTSgJDFJKXl2fsEoQJk8B4jgwfPpzY2Fj69++Pn58fS5cu5caNG1SpUoUNGzbQtGlTAgIC7nsY8c89E61Wy5IlS2jevDn16tVj0KBBpKamKqrhr/X8/Pxo27YtkZGRhebrdDpCQ0OpVasWrVu35sCBA/p5d+7cITg4mIYNG9KoUSNmzZpFfn7+f/tQxCORwHiOTJ8+nVKlSrFo0SKOHTtG37599fOOHDnCtm3bWL58+UPbWbNmDVFRUaxdu5Y9e/bg6OhIaGioohq8vb1Zt24dR48eZcCAAQwfPpzExET9/BMnTlCmTBkOHjxIUFAQAwYM0IfRqFGjMDMzY/v27WzevJl9+/axYcOGR/sQxH8igSEAGDhwIDY2NvrrGh5k/fr1DBkyBA8PDywsLBgwYAARERGKDmfatGmDu7s7arWatm3b4uPjw4kTJ/TzXVxcCAgIwNzcnLZt21KuXDl27dpFUlISu3fvJjg4GBsbG0qUKMEHH3wg16cUM7m8XQDg4eGheNnY2Fg+/vjjQpdjq9Vqbt26hbu7+wPX3bx5MytXriQmJgaAjIwMUlJS9PPd3d1RqVT696VKlSIxMZHY2Fjy8vJo2LChfp5Wq5VbChQzCQwBUOiX1NramqysLP37/Pz8Qrcf9PDwYNKkSdSqVeuRthETE8OYMWP46quv8PPzQ6PR0KFDh0LLJCQkoNPp9PXExcXh7++v35s5ePAgZmbyY2ssckjynClZsiTXr19/4DLlypUjOzubXbt2kZuby8KFC8nJydHP7969O7Nnz9bvJSQnJxMVFfXQbWdmZqJSqfT3Ddm4cSPnz58vtExycjKrV68mNzeX8PBwLl68SJMmTXBzc6NBgwZMmTKFu3fvotVquXbtGocPH37Uj0D8BxIYz5l+/fqxcOFCateuXeQAp729PSEhIYwZM4bGjRtjbW1d6JDl/fffx9/fn169euHn50eXLl0KjUMUpWLFivTq1Ytu3brx6quvcu7cOV5++eVCy/j6+nL16lVeeeUVZs+ezdy5c3F2dgZg2rRp5Obm0rZtW+rUqUNQUBA3b978D5+GeFRyPwwhhGKyhyGEUEwCQwihmASGEEIxCQwhhGISGEIIxSQwhBCKSWAIIRSTwBBCKCaBIYRQ7H/wdEK7X+kN3AAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qA1VtHl43VIx"
      },
      "source": [
        "##### Με αναζήτηση πλέγματος με διασταυρούμενη επικύρωση"
      ],
      "id": "qA1VtHl43VIx"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7VWhev-GJRDs"
      },
      "source": [
        "**Παρατήρηση:** Eπειδή παρατηρήσαμε μεγάλους χρόνους εκτέλεσης επιλέξαμε συγκεκριμένους συνδυασμούς για την εξερεύνηση της παραμέτρου *hidden_layer_sizes* - οι οποίοι ανήκουν στους βέλτιστους συνδυασμούς για το ερώτημα με τα pipelines."
      ],
      "id": "7VWhev-GJRDs"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5WaSGGki3VI1",
        "outputId": "d5fbd568-5a22-427e-c8cf-76669c977a67"
      },
      "source": [
        "import optuna\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "from sklearn.preprocessing import StandardScaler\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.svm import SVC\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.model_selection import cross_val_score\n",
        "from sklearn.neural_network import MLPClassifier\n",
        "\n",
        "#using best combinations found in pipeline optimisation\n",
        "layers = [(70, 60),(70, 70),(10,70,140),(50, 10, 120),(50, 20, 20),(90, 140, 30)]\n",
        "\n",
        "param_distributions = {\n",
        "  \"hidden_layer_sizes\":optuna.distributions.CategoricalDistribution(layers)\n",
        "}\n",
        "\n",
        "mlp = MLPClassifier()\n",
        "optuna_search = optuna.integration.OptunaSearchCV(mlp, param_distributions, cv=10, n_trials = 4, refit=True)\n",
        "optuna_search.fit(new_X_train_df, Y_train.ravel())\n",
        "preds_mlp_cv = optuna_search.predict(new_X_test_df)"
      ],
      "id": "5WaSGGki3VI1",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (70, 60) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (70, 70) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (10, 70, 140) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (50, 10, 120) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (50, 20, 20) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/optuna/distributions.py:427: UserWarning:\n",
            "\n",
            "Choices for a categorical distribution should be a tuple of None, bool, int, float and str for persistent storage but contains (90, 140, 30) which is of type tuple.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/ipykernel_launcher.py:18: ExperimentalWarning:\n",
            "\n",
            "OptunaSearchCV is experimental (supported from v0.17.0). The interface can change in the future.\n",
            "\n",
            "\u001b[32m[I 2021-12-03 16:22:56,105]\u001b[0m A new study created in memory with name: no-name-20c59649-c8d3-4f99-b85c-a2f080824285\u001b[0m\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n",
            "/usr/local/lib/python3.7/dist-packages/sklearn/neural_network/_multilayer_perceptron.py:696: ConvergenceWarning:\n",
            "\n",
            "Stochastic Optimizer: Maximum iterations (200) reached and the optimization hasn't converged yet.\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pnxejFzt3VI2"
      },
      "source": [
        "Τελικά οι καλυτέροι υπερπαράμετροι για την ταξινόμηση με MLP με cross_validation είναι:"
      ],
      "id": "pnxejFzt3VI2"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 165
        },
        "id": "J1BOE98hE-ZZ",
        "outputId": "6b891e69-565f-4896-d080-d4951779f6a0"
      },
      "source": [
        "optuna_search.study_.best_params"
      ],
      "id": "J1BOE98hE-ZZ",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-1-e78e9cbbbb38>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0moptuna_search\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstudy_\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbest_params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m: name 'optuna_search' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-X3J5ZUVE-Za"
      },
      "source": [
        "Στα παρακάτω διαγράμματα βλέπουμε ένα διάγραμμα για κάθε υπερπαράμετρο που δοκιμάστηκε , και με πιο σκούρο χρώμα τις υπερπαραμέτρους που οδήγησαν στη βέλτιστη επίδοση."
      ],
      "id": "-X3J5ZUVE-Za"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 562
        },
        "id": "ksHqxnqjE-Za",
        "outputId": "724850ee-31c5-4edc-bbed-9ab10b7e6189"
      },
      "source": [
        "optuna.visualization.plot_slice(optuna_search.study_)"
      ],
      "id": "ksHqxnqjE-Za",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"1d17b60e-9ed1-4066-8468-2ee4de77c531\" class=\"plotly-graph-div\" style=\"height:525px; width:1500px;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"1d17b60e-9ed1-4066-8468-2ee4de77c531\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '1d17b60e-9ed1-4066-8468-2ee4de77c531',\n",
              "                        [{\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": true}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [1.204552851257661, 0.20796091832380836, 0.5103197999781499, 0.32060168355070784], \"xaxis\": \"x\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [2, 1, 1, 1], \"xaxis\": \"x2\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y2\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"scale\", \"scale\", \"scale\", \"auto\"], \"xaxis\": \"x3\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y3\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [\"poly\", \"rbf\", \"linear\", \"rbf\"], \"xaxis\": \"x4\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y4\"}, {\"marker\": {\"color\": [0, 1, 2, 3], \"colorbar\": {\"title\": {\"text\": \"#Trials\"}, \"x\": 1.0, \"xpad\": 40}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"line\": {\"color\": \"Grey\", \"width\": 0.5}, \"showscale\": false}, \"mode\": \"markers\", \"showlegend\": false, \"type\": \"scatter\", \"x\": [10, 10, 6, 6], \"xaxis\": \"x5\", \"y\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"yaxis\": \"y5\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Slice Plot\"}, \"width\": 1500, \"xaxis\": {\"anchor\": \"y\", \"domain\": [0.0, 0.16799999999999998], \"title\": {\"text\": \"C\"}, \"type\": \"log\"}, \"xaxis2\": {\"anchor\": \"y2\", \"domain\": [0.208, 0.376], \"title\": {\"text\": \"degree\"}}, \"xaxis3\": {\"anchor\": \"y3\", \"domain\": [0.416, 0.584], \"title\": {\"text\": \"gamma\"}}, \"xaxis4\": {\"anchor\": \"y4\", \"domain\": [0.624, 0.792], \"title\": {\"text\": \"kernel\"}}, \"xaxis5\": {\"anchor\": \"y5\", \"domain\": [0.832, 1.0], \"title\": {\"text\": \"n_components\"}}, \"yaxis\": {\"anchor\": \"x\", \"domain\": [0.0, 1.0], \"title\": {\"text\": \"Objective Value\"}}, \"yaxis2\": {\"anchor\": \"x2\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis3\": {\"anchor\": \"x3\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis4\": {\"anchor\": \"x4\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}, \"yaxis5\": {\"anchor\": \"x5\", \"domain\": [0.0, 1.0], \"matches\": \"y\", \"showticklabels\": false}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('1d17b60e-9ed1-4066-8468-2ee4de77c531');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yzmey4VdE-Za"
      },
      "source": [
        "Εδώ βλέπουμε το διάγραμμα που συνδυάζει τις τιμές των υπερπαραμέτρων και την τιμή για το f1_score. Με το πιο σκούρο χρώμα φαίνεται η ευθεία που αντιστοιχεί στη βέλτιστη τιμή του f1."
      ],
      "id": "yzmey4VdE-Za"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 542
        },
        "id": "v9a8BgGTE-Za",
        "outputId": "435bda27-94ef-4e3d-83a1-f4e18adf11ff"
      },
      "source": [
        "optuna.visualization.plot_parallel_coordinate(optuna_search.study_)"
      ],
      "id": "v9a8BgGTE-Za",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<html>\n",
              "<head><meta charset=\"utf-8\" /></head>\n",
              "<body>\n",
              "    <div>\n",
              "            <script src=\"https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.5/MathJax.js?config=TeX-AMS-MML_SVG\"></script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script>\n",
              "                <script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script>\n",
              "        <script src=\"https://cdn.plot.ly/plotly-latest.min.js\"></script>    \n",
              "            <div id=\"818b5255-30cf-432b-a1c4-5f00a81502a4\" class=\"plotly-graph-div\" style=\"height:525px; width:100%;\"></div>\n",
              "            <script type=\"text/javascript\">\n",
              "                \n",
              "                    window.PLOTLYENV=window.PLOTLYENV || {};\n",
              "                    \n",
              "                if (document.getElementById(\"818b5255-30cf-432b-a1c4-5f00a81502a4\")) {\n",
              "                    Plotly.newPlot(\n",
              "                        '818b5255-30cf-432b-a1c4-5f00a81502a4',\n",
              "                        [{\"dimensions\": [{\"label\": \"Objective Value\", \"range\": [0.4310638160275016, 0.6600886858013236], \"values\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337]}, {\"label\": \"C\", \"range\": [-0.682018273455798, 0.08082585996409282], \"ticktext\": [\"0.208\", \"1\", \"1.2\"], \"tickvals\": [-0.682018273455798, 0, 0.08082585996409282], \"values\": [0.08082585996409282, -0.682018273455798, -0.29215758108099066, -0.49403420139902093]}, {\"label\": \"degree\", \"range\": [1, 2], \"values\": [2, 1, 1, 1]}, {\"label\": \"gamma\", \"range\": [0, 1], \"ticktext\": [\"scale\", \"auto\"], \"tickvals\": [0, 1], \"values\": [0, 0, 0, 1]}, {\"label\": \"kernel\", \"range\": [0, 2], \"ticktext\": [\"poly\", \"rbf\", \"linear\"], \"tickvals\": [0, 1, 2], \"values\": [0, 1, 2, 1]}, {\"label\": \"n_components\", \"range\": [6, 10], \"values\": [10, 10, 6, 6]}], \"labelangle\": 30, \"labelside\": \"bottom\", \"line\": {\"color\": [0.4310638160275016, 0.6600886858013236, 0.5508205656652657, 0.5625411524902337], \"colorbar\": {\"title\": {\"text\": \"Objective Value\"}}, \"colorscale\": [[0.0, \"rgb(247,251,255)\"], [0.125, \"rgb(222,235,247)\"], [0.25, \"rgb(198,219,239)\"], [0.375, \"rgb(158,202,225)\"], [0.5, \"rgb(107,174,214)\"], [0.625, \"rgb(66,146,198)\"], [0.75, \"rgb(33,113,181)\"], [0.875, \"rgb(8,81,156)\"], [1.0, \"rgb(8,48,107)\"]], \"reversescale\": false, \"showscale\": true}, \"type\": \"parcoords\"}],\n",
              "                        {\"template\": {\"data\": {\"bar\": [{\"error_x\": {\"color\": \"#2a3f5f\"}, \"error_y\": {\"color\": \"#2a3f5f\"}, \"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"bar\"}], \"barpolar\": [{\"marker\": {\"line\": {\"color\": \"#E5ECF6\", \"width\": 0.5}}, \"type\": \"barpolar\"}], \"carpet\": [{\"aaxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"baxis\": {\"endlinecolor\": \"#2a3f5f\", \"gridcolor\": \"white\", \"linecolor\": \"white\", \"minorgridcolor\": \"white\", \"startlinecolor\": \"#2a3f5f\"}, \"type\": \"carpet\"}], \"choropleth\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"choropleth\"}], \"contour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"contour\"}], \"contourcarpet\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"contourcarpet\"}], \"heatmap\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmap\"}], \"heatmapgl\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"heatmapgl\"}], \"histogram\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"histogram\"}], \"histogram2d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2d\"}], \"histogram2dcontour\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"histogram2dcontour\"}], \"mesh3d\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"type\": \"mesh3d\"}], \"parcoords\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"parcoords\"}], \"pie\": [{\"automargin\": true, \"type\": \"pie\"}], \"scatter\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter\"}], \"scatter3d\": [{\"line\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatter3d\"}], \"scattercarpet\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattercarpet\"}], \"scattergeo\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergeo\"}], \"scattergl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattergl\"}], \"scattermapbox\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scattermapbox\"}], \"scatterpolar\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolar\"}], \"scatterpolargl\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterpolargl\"}], \"scatterternary\": [{\"marker\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"type\": \"scatterternary\"}], \"surface\": [{\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}, \"colorscale\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"type\": \"surface\"}], \"table\": [{\"cells\": {\"fill\": {\"color\": \"#EBF0F8\"}, \"line\": {\"color\": \"white\"}}, \"header\": {\"fill\": {\"color\": \"#C8D4E3\"}, \"line\": {\"color\": \"white\"}}, \"type\": \"table\"}]}, \"layout\": {\"annotationdefaults\": {\"arrowcolor\": \"#2a3f5f\", \"arrowhead\": 0, \"arrowwidth\": 1}, \"coloraxis\": {\"colorbar\": {\"outlinewidth\": 0, \"ticks\": \"\"}}, \"colorscale\": {\"diverging\": [[0, \"#8e0152\"], [0.1, \"#c51b7d\"], [0.2, \"#de77ae\"], [0.3, \"#f1b6da\"], [0.4, \"#fde0ef\"], [0.5, \"#f7f7f7\"], [0.6, \"#e6f5d0\"], [0.7, \"#b8e186\"], [0.8, \"#7fbc41\"], [0.9, \"#4d9221\"], [1, \"#276419\"]], \"sequential\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]], \"sequentialminus\": [[0.0, \"#0d0887\"], [0.1111111111111111, \"#46039f\"], [0.2222222222222222, \"#7201a8\"], [0.3333333333333333, \"#9c179e\"], [0.4444444444444444, \"#bd3786\"], [0.5555555555555556, \"#d8576b\"], [0.6666666666666666, \"#ed7953\"], [0.7777777777777778, \"#fb9f3a\"], [0.8888888888888888, \"#fdca26\"], [1.0, \"#f0f921\"]]}, \"colorway\": [\"#636efa\", \"#EF553B\", \"#00cc96\", \"#ab63fa\", \"#FFA15A\", \"#19d3f3\", \"#FF6692\", \"#B6E880\", \"#FF97FF\", \"#FECB52\"], \"font\": {\"color\": \"#2a3f5f\"}, \"geo\": {\"bgcolor\": \"white\", \"lakecolor\": \"white\", \"landcolor\": \"#E5ECF6\", \"showlakes\": true, \"showland\": true, \"subunitcolor\": \"white\"}, \"hoverlabel\": {\"align\": \"left\"}, \"hovermode\": \"closest\", \"mapbox\": {\"style\": \"light\"}, \"paper_bgcolor\": \"white\", \"plot_bgcolor\": \"#E5ECF6\", \"polar\": {\"angularaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"radialaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"scene\": {\"xaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"yaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}, \"zaxis\": {\"backgroundcolor\": \"#E5ECF6\", \"gridcolor\": \"white\", \"gridwidth\": 2, \"linecolor\": \"white\", \"showbackground\": true, \"ticks\": \"\", \"zerolinecolor\": \"white\"}}, \"shapedefaults\": {\"line\": {\"color\": \"#2a3f5f\"}}, \"ternary\": {\"aaxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"baxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}, \"bgcolor\": \"#E5ECF6\", \"caxis\": {\"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\"}}, \"title\": {\"x\": 0.05}, \"xaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}, \"yaxis\": {\"automargin\": true, \"gridcolor\": \"white\", \"linecolor\": \"white\", \"ticks\": \"\", \"title\": {\"standoff\": 15}, \"zerolinecolor\": \"white\", \"zerolinewidth\": 2}}}, \"title\": {\"text\": \"Parallel Coordinate Plot\"}},\n",
              "                        {\"responsive\": true}\n",
              "                    ).then(function(){\n",
              "                            \n",
              "var gd = document.getElementById('818b5255-30cf-432b-a1c4-5f00a81502a4');\n",
              "var x = new MutationObserver(function (mutations, observer) {{\n",
              "        var display = window.getComputedStyle(gd).display;\n",
              "        if (!display || display === 'none') {{\n",
              "            console.log([gd, 'removed!']);\n",
              "            Plotly.purge(gd);\n",
              "            observer.disconnect();\n",
              "        }}\n",
              "}});\n",
              "\n",
              "// Listen for the removal of the full notebook cells\n",
              "var notebookContainer = gd.closest('#notebook-container');\n",
              "if (notebookContainer) {{\n",
              "    x.observe(notebookContainer, {childList: true});\n",
              "}}\n",
              "\n",
              "// Listen for the clearing of the current output cell\n",
              "var outputEl = gd.closest('.output');\n",
              "if (outputEl) {{\n",
              "    x.observe(outputEl, {childList: true});\n",
              "}}\n",
              "\n",
              "                        })\n",
              "                };\n",
              "                \n",
              "            </script>\n",
              "        </div>\n",
              "</body>\n",
              "</html>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "abHR-1JdE-Za"
      },
      "source": [
        "###### Επιδόσεις καλύτερου μοντέλου SVM στο test set"
      ],
      "id": "abHR-1JdE-Za"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ezPO2MBWE-Za"
      },
      "source": [
        "###### Εντοπίζω τις επιδόσεις"
      ],
      "id": "ezPO2MBWE-Za"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wCfwv42hE-Za"
      },
      "source": [
        "best_accuracies['MLP_cv'] = round(accuracy_score(Y_test, preds_cv), 3)\n",
        "best_f1s['MLP_cv'] = round(f1_score(Y_test, preds_cv, average='macro'), 3)\n",
        "best_recalls['MLP_cv'] = round(recall_score(Y_test, preds_cv, average='macro'), 3) \n",
        "\n",
        "print('SVM with cross validation best scores:')\n",
        "print(\"Accuracy:\", best_accuracies['MLP_cv'])\n",
        "print(\"F1:\", best_f1s['MLP_cv'])\n",
        "print(\"Recall:\", best_recalls['MLP_cv'])"
      ],
      "id": "wCfwv42hE-Za",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7Y3xU_i5E-Zb"
      },
      "source": [
        "###### Σχεδιάζω τον πίνακα σύγχυσης"
      ],
      "id": "7Y3xU_i5E-Zb"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "id": "IOqi9hf0E-Zb",
        "outputId": "2e4e3fc6-7738-4478-e332-2a4e1e1f198d"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "import seaborn as sns; sns.set()\n",
        "print(f\"SVM:\")\n",
        "mat = confusion_matrix(Y_test , preds_mlp_cv)\n",
        "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False)\n",
        "plt.xlabel('true label')\n",
        "plt.ylabel('predicted label');"
      ],
      "id": "IOqi9hf0E-Zb",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "SVM:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5OD4WfDV3VI3"
      },
      "source": [
        "#### Συμπεράσματα MLP"
      ],
      "id": "5OD4WfDV3VI3"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MlSPs5X73VI4"
      },
      "source": [
        "##### Με pipeline"
      ],
      "id": "MlSPs5X73VI4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mg0bJWplPUKk"
      },
      "source": [
        "###### Σχεδιάζουμε τις συγκριτικές γραφικές για τις μελέτες με το optuna"
      ],
      "id": "mg0bJWplPUKk"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 247
        },
        "id": "m1Fxb3AaPZTs",
        "outputId": "73a661f0-a3ea-49de-deba-5798709ef047"
      },
      "source": [
        "print(\"Out of the box:\")\n",
        "plot_scores(out_of_the_box_accuracy, out_of_the_box_f1, out_of_the_box_recall)"
      ],
      "id": "m1Fxb3AaPZTs",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Out of the box:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 3 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uZ4YVmEQYdOk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "84979def-ae22-4c14-a4f3-8dfd60fcd0da"
      },
      "source": [
        "print(\"Accuracies:\",best_accuracies)\n",
        "print(\"F1 scores\",best_f1s)\n",
        "print(\"Recall scores\",best_recalls)"
      ],
      "id": "uZ4YVmEQYdOk",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Accuracies: {'SVM_pipe': 0.679, 'MLP': 0.683}\n",
            "F1 scores {'SVM_pipe': 0.66, 'MLP': 0.668}\n",
            "Recall scores {'SVM_pipe': 0.645, 'MLP': 0.651}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 248
        },
        "id": "6-_QbaRAPiV8",
        "outputId": "b580d97d-79d2-4e93-e9f8-1d7d161aaa12"
      },
      "source": [
        "print(\"After:\")\n",
        "plot_scores(best_accuracies, best_f1s, best_recalls)"
      ],
      "id": "6-_QbaRAPiV8",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "After:\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x216 with 2 Axes>"
            ]
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5CMdS2NzIzqd"
      },
      "source": [
        "### Βέλτιστος ταξινομητής"
      ],
      "id": "5CMdS2NzIzqd"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3KPqLHNSI9v_"
      },
      "source": [
        "Γενικά παρατηρούμε πως ο MLP παρουσιάζει καλύτερες επίδοσεις από τα SVMs με μικρές όμως διαφορές."
      ],
      "id": "3KPqLHNSI9v_"
    }
  ]
}