zawaly_serca_klasyfikacja.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.linear_model import  LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor\n",
    "from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor, plot_tree\n",
    "from sklearn.model_selection import GridSearchCV, StratifiedKFold, KFold\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from sklearn.metrics import *\n",
    "from sklearn.model_selection import train_test_split, cross_val_score\n",
    "from mlxtend import plotting\n",
    "from sklearn.impute import SimpleImputer\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler, LabelEncoder\n",
    "from sklearn.decomposition import PCA\n",
    "\n",
    "import statsmodels.api as sm\n",
    "\n",
    "from hyperopt import hp, fmin, tpe, STATUS_OK, Trials, space_eval\n",
    "from hyperopt.pyll.base import scope\n",
    "from hyperopt.pyll.stochastic import sample\n",
    "from xgboost import XGBClassifier\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**1. Poznanie danych ze zbioru healtcare, odnosnie ataków serca**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>gender</th>\n",
       "      <th>age</th>\n",
       "      <th>hypertension</th>\n",
       "      <th>heart_disease</th>\n",
       "      <th>ever_married</th>\n",
       "      <th>work_type</th>\n",
       "      <th>Residence_type</th>\n",
       "      <th>avg_glucose_level</th>\n",
       "      <th>bmi</th>\n",
       "      <th>smoking_status</th>\n",
       "      <th>stroke</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9046</td>\n",
       "      <td>Male</td>\n",
       "      <td>67.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Private</td>\n",
       "      <td>Urban</td>\n",
       "      <td>228.69</td>\n",
       "      <td>36.6</td>\n",
       "      <td>formerly smoked</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>51676</td>\n",
       "      <td>Female</td>\n",
       "      <td>61.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Self-employed</td>\n",
       "      <td>Rural</td>\n",
       "      <td>202.21</td>\n",
       "      <td>NaN</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>31112</td>\n",
       "      <td>Male</td>\n",
       "      <td>80.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Private</td>\n",
       "      <td>Rural</td>\n",
       "      <td>105.92</td>\n",
       "      <td>32.5</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>60182</td>\n",
       "      <td>Female</td>\n",
       "      <td>49.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Private</td>\n",
       "      <td>Urban</td>\n",
       "      <td>171.23</td>\n",
       "      <td>34.4</td>\n",
       "      <td>smokes</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1665</td>\n",
       "      <td>Female</td>\n",
       "      <td>79.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Self-employed</td>\n",
       "      <td>Rural</td>\n",
       "      <td>174.12</td>\n",
       "      <td>24.0</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5105</th>\n",
       "      <td>18234</td>\n",
       "      <td>Female</td>\n",
       "      <td>80.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Private</td>\n",
       "      <td>Urban</td>\n",
       "      <td>83.75</td>\n",
       "      <td>NaN</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5106</th>\n",
       "      <td>44873</td>\n",
       "      <td>Female</td>\n",
       "      <td>81.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Self-employed</td>\n",
       "      <td>Urban</td>\n",
       "      <td>125.20</td>\n",
       "      <td>40.0</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5107</th>\n",
       "      <td>19723</td>\n",
       "      <td>Female</td>\n",
       "      <td>35.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Self-employed</td>\n",
       "      <td>Rural</td>\n",
       "      <td>82.99</td>\n",
       "      <td>30.6</td>\n",
       "      <td>never smoked</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5108</th>\n",
       "      <td>37544</td>\n",
       "      <td>Male</td>\n",
       "      <td>51.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Private</td>\n",
       "      <td>Rural</td>\n",
       "      <td>166.29</td>\n",
       "      <td>25.6</td>\n",
       "      <td>formerly smoked</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5109</th>\n",
       "      <td>44679</td>\n",
       "      <td>Female</td>\n",
       "      <td>44.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Govt_job</td>\n",
       "      <td>Urban</td>\n",
       "      <td>85.28</td>\n",
       "      <td>26.2</td>\n",
       "      <td>Unknown</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5110 rows × 12 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         id  gender   age  hypertension  heart_disease ever_married  \\\n",
       "0      9046    Male  67.0             0              1          Yes   \n",
       "1     51676  Female  61.0             0              0          Yes   \n",
       "2     31112    Male  80.0             0              1          Yes   \n",
       "3     60182  Female  49.0             0              0          Yes   \n",
       "4      1665  Female  79.0             1              0          Yes   \n",
       "...     ...     ...   ...           ...            ...          ...   \n",
       "5105  18234  Female  80.0             1              0          Yes   \n",
       "5106  44873  Female  81.0             0              0          Yes   \n",
       "5107  19723  Female  35.0             0              0          Yes   \n",
       "5108  37544    Male  51.0             0              0          Yes   \n",
       "5109  44679  Female  44.0             0              0          Yes   \n",
       "\n",
       "          work_type Residence_type  avg_glucose_level   bmi   smoking_status  \\\n",
       "0           Private          Urban             228.69  36.6  formerly smoked   \n",
       "1     Self-employed          Rural             202.21   NaN     never smoked   \n",
       "2           Private          Rural             105.92  32.5     never smoked   \n",
       "3           Private          Urban             171.23  34.4           smokes   \n",
       "4     Self-employed          Rural             174.12  24.0     never smoked   \n",
       "...             ...            ...                ...   ...              ...   \n",
       "5105        Private          Urban              83.75   NaN     never smoked   \n",
       "5106  Self-employed          Urban             125.20  40.0     never smoked   \n",
       "5107  Self-employed          Rural              82.99  30.6     never smoked   \n",
       "5108        Private          Rural             166.29  25.6  formerly smoked   \n",
       "5109       Govt_job          Urban              85.28  26.2          Unknown   \n",
       "\n",
       "      stroke  \n",
       "0          1  \n",
       "1          1  \n",
       "2          1  \n",
       "3          1  \n",
       "4          1  \n",
       "...      ...  \n",
       "5105       0  \n",
       "5106       0  \n",
       "5107       0  \n",
       "5108       0  \n",
       "5109       0  \n",
       "\n",
       "[5110 rows x 12 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#ładowanie datasetu o zawałach serca\n",
    "df=pd.read_csv(\"healthcare-dataset-stroke-data.csv\")\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#usuniecie kolumny ID\n",
    "del df[\"id\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 5110 entries, 0 to 5109\n",
      "Data columns (total 11 columns):\n",
      " #   Column             Non-Null Count  Dtype  \n",
      "---  ------             --------------  -----  \n",
      " 0   gender             5110 non-null   object \n",
      " 1   age                5110 non-null   float64\n",
      " 2   hypertension       5110 non-null   int64  \n",
      " 3   heart_disease      5110 non-null   int64  \n",
      " 4   ever_married       5110 non-null   object \n",
      " 5   work_type          5110 non-null   object \n",
      " 6   Residence_type     5110 non-null   object \n",
      " 7   avg_glucose_level  5110 non-null   float64\n",
      " 8   bmi                4909 non-null   float64\n",
      " 9   smoking_status     5110 non-null   object \n",
      " 10  stroke             5110 non-null   int64  \n",
      "dtypes: float64(3), int64(3), object(5)\n",
      "memory usage: 439.3+ KB\n"
     ]
    }
   ],
   "source": [
    "#sprawdzamy typy zmiennych\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "gender                 0\n",
       "age                    0\n",
       "hypertension           0\n",
       "heart_disease          0\n",
       "ever_married           0\n",
       "work_type              0\n",
       "Residence_type         0\n",
       "avg_glucose_level      0\n",
       "bmi                  201\n",
       "smoking_status         0\n",
       "stroke                 0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#sprawdzamy ile jest NA\n",
    "df.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#zamienimy na w kolumnie BMI, przy wykorzystaniu SimpleImputer z pakietu sklearn\n",
    "imputer = SimpleImputer()\n",
    "\n",
    "pom1 = df[\"bmi\"].to_frame()\n",
    "df[\"bmi\"]=imputer.fit_transform(pom1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>hypertension</th>\n",
       "      <th>heart_disease</th>\n",
       "      <th>avg_glucose_level</th>\n",
       "      <th>bmi</th>\n",
       "      <th>stroke</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>5110.000000</td>\n",
       "      <td>5110.000000</td>\n",
       "      <td>5110.000000</td>\n",
       "      <td>5110.000000</td>\n",
       "      <td>5110.000000</td>\n",
       "      <td>5110.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>43.226614</td>\n",
       "      <td>0.097456</td>\n",
       "      <td>0.054012</td>\n",
       "      <td>106.147677</td>\n",
       "      <td>28.893237</td>\n",
       "      <td>0.048728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>22.612647</td>\n",
       "      <td>0.296607</td>\n",
       "      <td>0.226063</td>\n",
       "      <td>45.283560</td>\n",
       "      <td>7.698018</td>\n",
       "      <td>0.215320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.080000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>55.120000</td>\n",
       "      <td>10.300000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>25.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>77.245000</td>\n",
       "      <td>23.800000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>45.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>91.885000</td>\n",
       "      <td>28.400000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>61.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>114.090000</td>\n",
       "      <td>32.800000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>82.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>271.740000</td>\n",
       "      <td>97.600000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               age  hypertension  heart_disease  avg_glucose_level  \\\n",
       "count  5110.000000   5110.000000    5110.000000        5110.000000   \n",
       "mean     43.226614      0.097456       0.054012         106.147677   \n",
       "std      22.612647      0.296607       0.226063          45.283560   \n",
       "min       0.080000      0.000000       0.000000          55.120000   \n",
       "25%      25.000000      0.000000       0.000000          77.245000   \n",
       "50%      45.000000      0.000000       0.000000          91.885000   \n",
       "75%      61.000000      0.000000       0.000000         114.090000   \n",
       "max      82.000000      1.000000       1.000000         271.740000   \n",
       "\n",
       "               bmi       stroke  \n",
       "count  5110.000000  5110.000000  \n",
       "mean     28.893237     0.048728  \n",
       "std       7.698018     0.215320  \n",
       "min      10.300000     0.000000  \n",
       "25%      23.800000     0.000000  \n",
       "50%      28.400000     0.000000  \n",
       "75%      32.800000     0.000000  \n",
       "max      97.600000     1.000000  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#poznajmy trochę nasze dane:\n",
    "df[df.columns[1:]].describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Jak widzimy wartosci wskaznika BMI przy zastosowaniu Imputera zaczeły nam się kształtować ponad ramy dopuszczalnych wskazników.\n",
    "Usuńmy jednak wartosci NA z tej zmiennej."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {},
   "outputs": [],
   "source": [
    "df=pd.read_csv(\"healthcare-dataset-stroke-data.csv\")\n",
    "del df[\"id\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df[~df['bmi'].isnull()] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 303,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# poznajmy inne dane\n",
    "zmienne_ciagle= df[df.columns[1:]].select_dtypes(include = [\"float64\",\"int64\"])\n",
    "zmienne_ciagle[zmienne_ciagle.columns[[1,2]]].boxplot()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Widzimy, że zmienne hypertension i heart_disease to tak na prawdę zmienne kategoryczne"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 304,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "zmienne_ciagle[zmienne_ciagle.columns[[0,3,4]]].boxplot()\n",
    "pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "  W przypadku pozostałych zmiennych ciągłych, mamy duzo wartości odstających dla avg_glucose_level i BMI. \n",
    "  Jednak nie wykluczałbym ich, ponieważ często przy chorobach pacjenci charakteryzują sie wysokimi wynikami."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of gender\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of ever_married\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of work_type\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of Residence_type\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAD4CAYAAAAAczaOAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAOsklEQVR4nO3df6zddX3H8edLQEamKKSFsLZZGavZCpk13HRs/CGGbXRuSTEZSckiXWZSR2qmi/4BZokuS40m/khIBkkNBEiYrE4NNZPN2jmdG4oX1lkKdnTC5NqmvUoycVG2lvf+OJ/Gk8vp/d1Ty+f5SE7O977P93vO9/zz7Pd+7/ecpqqQJPXhVWd6ByRJ42P0JakjRl+SOmL0JakjRl+SOnLumd6BuaxYsaLWrl17pndDks4qjz322PerauXM+c989NeuXcvk5OSZ3g1JOqsk+a9Rc0/vSFJHjL4kdcToS1JHjL4kdcToS1JH5ox+kjVJvpzkqSQHkry7zT+Y5HtJ9rXbW4e2uT3JoSQHk9wwNL86yf722B1JcnreliRplPlcsnkceG9VPZ7ktcBjSfa0xz5RVR8dXjnJemALcCXwC8CXkryhqk4AdwHbgK8DXwA2AQ8vz1uRJM1lziP9qjpSVY+35ReAp4BVs2yyGXiwql6sqmeAQ8DGJJcBF1bVIzX4Puf7gRuX+gYkSfO3oHP6SdYCbwK+0UbvSvKtJPckuajNVgHPDW021War2vLM+ajX2ZZkMsnk9PT0QnZRkjSLeX8iN8lrgM8A76mqHya5C/hLoNr9x4A/Bkadp69Z5i8fVu0EdgJMTEws+n95WXvb3y120yV59sO/d0ZeV5LmMq8j/STnMQj+A1X1WYCqOlpVJ6rqJeCTwMa2+hSwZmjz1cDhNl89Yi5JGpP5XL0T4G7gqar6+ND8sqHV3gY80ZZ3A1uSnJ/kcmAd8GhVHQFeSHJNe85bgIeW6X1IkuZhPqd3rgXeDuxPsq/N3g/cnGQDg1M0zwLvBKiqA0l2AU8yuPJne7tyB+BW4F7gAgZX7XjljiSN0ZzRr6qvMfp8/Bdm2WYHsGPEfBK4aiE7KElaPn4iV5I68jP/ffqSdCa90q4C9Ehfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI3NGP8maJF9O8lSSA0ne3eYXJ9mT5Ol2f9HQNrcnOZTkYJIbhuZXJ9nfHrsjSU7P25IkjTKfI/3jwHur6leBa4DtSdYDtwF7q2odsLf9THtsC3AlsAm4M8k57bnuArYB69pt0zK+F0nSHOaMflUdqarH2/ILwFPAKmAzcF9b7T7gxra8GXiwql6sqmeAQ8DGJJcBF1bVI1VVwP1D20iSxmBB5/STrAXeBHwDuLSqjsDgHwbgkrbaKuC5oc2m2mxVW545H/U625JMJpmcnp5eyC5KkmYx7+gneQ3wGeA9VfXD2VYdMatZ5i8fVu2sqomqmli5cuV8d1GSNId5RT/JeQyC/0BVfbaNj7ZTNrT7Y20+BawZ2nw1cLjNV4+YS5LGZD5X7wS4G3iqqj4+9NBuYGtb3go8NDTfkuT8JJcz+IPto+0U0AtJrmnPecvQNpKkMTh3HutcC7wd2J9kX5u9H/gwsCvJO4DvAjcBVNWBJLuAJxlc+bO9qk607W4F7gUuAB5uN0nSmMwZ/ar6GqPPxwNcf4ptdgA7RswngasWsoOSpOXjJ3IlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNGX5I6YvQlqSNzRj/JPUmOJXliaPbBJN9Lsq/d3jr02O1JDiU5mOSGofnVSfa3x+5IkuV/O5Kk2cznSP9eYNOI+SeqakO7fQEgyXpgC3Bl2+bOJOe09e8CtgHr2m3Uc0qSTqM5o19VXwWen+fzbQYerKoXq+oZ4BCwMcllwIVV9UhVFXA/cOMi91mStEhLOaf/riTfaqd/LmqzVcBzQ+tMtdmqtjxzPlKSbUkmk0xOT08vYRclScMWG/27gCuADcAR4GNtPuo8fc0yH6mqdlbVRFVNrFy5cpG7KEmaaVHRr6qjVXWiql4CPglsbA9NAWuGVl0NHG7z1SPmkqQxWlT02zn6k94GnLyyZzewJcn5SS5n8AfbR6vqCPBCkmvaVTu3AA8tYb8lSYtw7lwrJPkUcB2wIskU8AHguiQbGJyieRZ4J0BVHUiyC3gSOA5sr6oT7aluZXAl0AXAw+0mSRqjOaNfVTePGN89y/o7gB0j5pPAVQvaO0nSsvITuZLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR0x+pLUEaMvSR2ZM/pJ7klyLMkTQ7OLk+xJ8nS7v2josduTHEpyMMkNQ/Ork+xvj92RJMv/diRJs5nPkf69wKYZs9uAvVW1DtjbfibJemALcGXb5s4k57Rt7gK2AevabeZzSpJOszmjX1VfBZ6fMd4M3NeW7wNuHJo/WFUvVtUzwCFgY5LLgAur6pGqKuD+oW0kSWOy2HP6l1bVEYB2f0mbrwKeG1pvqs1WteWZc0nSGC33H3JHnaevWeajnyTZlmQyyeT09PSy7Zwk9W6x0T/aTtnQ7o+1+RSwZmi91cDhNl89Yj5SVe2sqomqmli5cuUid1GSNNNio78b2NqWtwIPDc23JDk/yeUM/mD7aDsF9EKSa9pVO7cMbSNJGpNz51ohyaeA64AVSaaADwAfBnYleQfwXeAmgKo6kGQX8CRwHNheVSfaU93K4EqgC4CH202SNEZzRr+qbj7FQ9efYv0dwI4R80ngqgXtnSRpWfmJXEnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqiNGXpI4YfUnqyJKin+TZJPuT7Esy2WYXJ9mT5Ol2f9HQ+rcnOZTkYJIblrrzkqSFWY4j/bdU1Yaqmmg/3wbsrap1wN72M0nWA1uAK4FNwJ1JzlmG15ckzdPpOL2zGbivLd8H3Dg0f7CqXqyqZ4BDwMbT8PqSpFNYavQL+GKSx5Jsa7NLq+oIQLu/pM1XAc8NbTvVZi+TZFuSySST09PTS9xFSdJJ5y5x+2ur6nCSS4A9Sb49y7oZMatRK1bVTmAnwMTExMh1JEkLt6Qj/ao63O6PAZ9jcLrmaJLLANr9sbb6FLBmaPPVwOGlvL4kaWEWHf0kP5/ktSeXgd8BngB2A1vbaluBh9rybmBLkvOTXA6sAx5d7OtLkhZuKad3LgU+l+Tk8/x1Vf19km8Cu5K8A/gucBNAVR1Isgt4EjgObK+qE0vae0nSgiw6+lX1HeCNI+Y/AK4/xTY7gB2LfU1J0tL4iVxJ6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6ojRl6SOGH1J6sjYo59kU5KDSQ4luW3cry9JPRtr9JOcA/wV8LvAeuDmJOvHuQ+S1LNxH+lvBA5V1Xeq6n+BB4HNY94HSerWuWN+vVXAc0M/TwG/PnOlJNuAbe3HHyU5uMjXWwF8f5HbLlo+Mu5XlPRKk48suV+/OGo47uhnxKxeNqjaCexc8oslk1U1sdTnkaRxO139GvfpnSlgzdDPq4HDY94HSerWuKP/TWBdksuTvBrYAuwe8z5IUrfGenqnqo4neRfwD8A5wD1VdeA0vuSSTxFJ0hlyWvqVqpedUpckvUL5iVxJ6ojRl6SOnBXRT7I2yRMzZh9M8r4R696b5A/Gt3eSNLckJ5LsS/JEks8nef0yPveP5rvuWRH9+Uoy7s8dSNJ8/biqNlTVVcDzwPb5bricbTvro5/kn5J8KMlXgHe38W8l+eck/5Hk99t6a9vs8Xb7zTa/rj3H3yb5dpIHkoz6EJkkLZdHGHxDwcmGTbTlFUmebct/lOTTST4PfDHJa5Lsbf3an2RRX2HzSjkyfn1VvRkGp3eAtcCbgSuALyf5ZeAY8NtV9ZMk64BPASc/7fYm4EoGHxT7F+Ba4GvjfAOS+tC+ePJ64O55rP4bwK9V1fPtaP9tVfXDJCuAryfZXQu8BPNsOdI/1Zs6Of+bGfNdVfVSVT0NfAf4FeA84JNJ9gOfZvAtnyc9WlVTVfUSsI/BPxqStJwuSLIP+AFwMbBnHtvsqarn23KADyX5FvAlBr8pXLrQnThbov8D4KIZs4v56ZcR/c+Mx2b+I1HAnwFHgTcyOMJ/9dDjLw4tn+CV8xuQpJ8dP66qDQy+CO3V/PSc/nF+2uKfm7HNcNv+EFgJXN2e5+iI9ed0VkS/qn4EHElyPUCSi4FNnPoUzE1JXpXkCuCXgIPA64Aj7Wj+7Qw+ESxJY1VV/w38KfC+JOcBzwJXt4dnu/LwdcCxqvq/JG/hFN+iOZezIvrNLcCft1+P/hH4i6r6z1OsexD4CvAw8CdV9RPgTmBrkq8Db+Dlvx1I0lhU1b8B/87g+8c+Ctya5F8ZfB38qTwATCSZZHDU/+3FvLZfwyBJHTmbjvQlSUtk9CWpI0Zfkjpi9CWpI0Zfkjpi9CWpI0Zfkjry/2c6mMDM4TT8AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of smoking_status\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of hypertension\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "plot of heart_disease\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "zmienne_kategoryczne = df.select_dtypes(include = \"object\").join(zmienne_ciagle[zmienne_ciagle.columns[[1,2]]])\n",
    "for i in zmienne_kategoryczne.columns:\n",
    "    plt.hist(df[i])\n",
    "    print(\"plot of\", i)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Struktura zmiennej gender \n",
      "Female    0.590141\n",
      "Male      0.409656\n",
      "Other     0.000204\n",
      "Name: gender, dtype: float64 \n",
      "\n",
      "Struktura zmiennej ever_married \n",
      "Yes    0.652679\n",
      "No     0.347321\n",
      "Name: ever_married, dtype: float64 \n",
      "\n",
      "Struktura zmiennej work_type \n",
      "Private          0.572622\n",
      "Self-employed    0.157873\n",
      "children         0.136688\n",
      "Govt_job         0.128336\n",
      "Never_worked     0.004482\n",
      "Name: work_type, dtype: float64 \n",
      "\n",
      "Struktura zmiennej Residence_type \n",
      "Urban    0.507232\n",
      "Rural    0.492768\n",
      "Name: Residence_type, dtype: float64 \n",
      "\n",
      "Struktura zmiennej smoking_status \n",
      "never smoked       0.377266\n",
      "Unknown            0.302098\n",
      "formerly smoked    0.170503\n",
      "smokes             0.150132\n",
      "Name: smoking_status, dtype: float64 \n",
      "\n",
      "Struktura zmiennej hypertension \n",
      "0    0.908128\n",
      "1    0.091872\n",
      "Name: hypertension, dtype: float64 \n",
      "\n",
      "Struktura zmiennej heart_disease \n",
      "0    0.950499\n",
      "1    0.049501\n",
      "Name: heart_disease, dtype: float64 \n",
      "\n"
     ]
    }
   ],
   "source": [
    "for i in zmienne_kategoryczne.columns:\n",
    "    print(f\"Struktura zmiennej {i} \\n{df[i].value_counts()/df.shape[0]} \\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Widzimy, że mamy niemal o połowe wiecej kobiet w zbiorze i niewielki odsetek osób nieokreślonych.\n",
    "Analizując zmienną ever_married możemy dostrzec, że było prawie dwa razy wiecej osób, ktore by kiedykolwiek po slubie.  \n",
    "Najwięcej obserwacji dotyczy osob z sektora prywatnego i 3x mniej, niemal po równo, osób samozatrudnionych, wychowujących dzieci czy pracujących w państwowych instytucjach. Nigdy nie pracujacy stanowią 4 promile.  \n",
    "W przypadku osób z nadciśnieniem (hypertension) mamy 90% oznaczonych jako 0 - brak i niecałe 10% cierpiących na te dolegliwosć.\n",
    "Warto zwrócić uwagę na zmienna objaśnianą, czyli heart_disease mamy tylko w 5% przypadków oznaczonych jako 1, czyli gdzie wystąpił atak serca, należy to uwzględnić przy wyznaczenia setu testowego."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**2. Wstępna obróbka danych do modelu**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# zmiana kolumn kategorycznych na typ category\n",
    "for i in df.select_dtypes(include = 'object').join(df[[\"hypertension\",\"heart_disease\"]]):\n",
    "       df[i]=df[i].astype('category')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# wyznaczenie dummy variables\n",
    "df_dummy = pd.get_dummies(df,drop_first= True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# zdefiniujmy zbiór X i Y bez podzialu na testowy do ewentualnej cross validacji\n",
    "X = df_dummy.iloc[:,0:3].join(df_dummy.iloc[:,4:])\n",
    "Y = df_dummy[\"stroke\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# podzielmy też na zbiór treningowy i testowy\n",
    "X_train, X_test, Y_train,Y_test = train_test_split(X,Y, test_size = 0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**3. Na początku zaczniemy od prostego modelu drzewa decyzyjnego**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  1.0 \n",
      " acc test:  0.9084\n"
     ]
    }
   ],
   "source": [
    "# prosty model drzewa decyzyjnego, nr1\n",
    "model_dd1 = DecisionTreeClassifier(random_state=1).fit(X_train, Y_train)\n",
    "\n",
    "Y_pred_dd1 =  model_dd1.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_dd1.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_dd1.score(X_test, Y_test),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "W zasadzie acc moglaby nam powiedziec, ze model jest bardzo dobry, ale pamietajmy, ze mamy niezbalansowany zbior 5:95"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.94      0.95       935\n",
      "           1       0.15      0.19      0.17        47\n",
      "\n",
      "    accuracy                           0.91       982\n",
      "   macro avg       0.55      0.57      0.56       982\n",
      "weighted avg       0.92      0.91      0.91       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_dd1 = confusion_matrix(Y_test, Y_pred_dd1)\n",
    "raport_dd1 = classification_report(Y_test, Y_pred_dd1)\n",
    "print(raport_dd1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Na podstawie macierzy błędu, widzimy, że ta metoda zapewnia nam niską precyzję dla klasy 1.   \n",
    "Model nie nadaje się do produkcji!"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**4. Spróbujmy dostroić model drzewa decyzyjnego i stworzyc kolejny model dd2**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# stwórzmy najpierw zestaw parametrów do testów\n",
    "params_dd2 = {\"max_depth\": list(range(2,16,3)),\n",
    "              \"max_features\" : list(range(3,16,3)),\n",
    "              \"max_leaf_nodes\" : list(range(2,53,10))}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 150 candidates, totalling 750 fits\n"
     ]
    }
   ],
   "source": [
    "clf = DecisionTreeClassifier(random_state=1) \n",
    "\n",
    "#zdefiniowalem tez wlasny scoring, poniewaz acc byl identyczny dla kilkunastu zestawow parametrow\n",
    "grid_search_dd2 = GridSearchCV(clf, param_grid = params_dd2, cv = 5, verbose = 1, scoring = make_scorer(f1_score, average = \"macro\")).fit(X,Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 8, 'max_features': 6, 'max_leaf_nodes': 52}"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# wyswietle teraz najlepszy model\n",
    "\n",
    "grid_search_dd2.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9684 \n",
      " acc test:  0.943\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.99      0.97       935\n",
      "           1       0.26      0.11      0.15        47\n",
      "\n",
      "    accuracy                           0.94       982\n",
      "   macro avg       0.61      0.55      0.56       982\n",
      "weighted avg       0.92      0.94      0.93       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# wdroze ten model dd2 i przetestuje jego wynik\n",
    "model_dd2 = DecisionTreeClassifier(max_depth=11, max_features=9, max_leaf_nodes=52,random_state=1).fit(X_train, Y_train)\n",
    "\n",
    "Y_pred_dd2 =  model_dd2.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_dd2.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_dd2.score(X_test, Y_test),4))\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_dd2 = confusion_matrix(Y_test, Y_pred_dd2)\n",
    "raport_dd2 = classification_report(Y_test, Y_pred_dd2)\n",
    "print(raport_dd2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Widzimy, że dzięki zmianie parametrów modelu jesteśmy w stanie znacznie poprawić naszą acc, ale nasz model ma niską precyzje dla drugiej klasy i przez to usredniony wynik wyrazony przez F1 score jest niski. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**5.Skonstruujmy prosty model lasów losowych**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  1.0 \n",
      " acc test:  0.9521\n"
     ]
    }
   ],
   "source": [
    "# prosty model drzewa decyzyjnego, nr1\n",
    "model_ll1 = RandomForestClassifier(random_state=1).fit(X_train, Y_train)\n",
    "\n",
    "Y_pred_ll1 =  model_ll1.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll1.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll1.score(X_test, Y_test),4))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Już na wstępie osiągamy lepsze wyniki niż w naszym prostym modelu drzewa decyzyjnego.  \n",
    "Wynik jest zbliżony do zoptymalizowanego modelu drzewa a jego uzyskanie zabiera dużo mniej czasu."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll1 = confusion_matrix(Y_test, Y_pred_ll1)\n",
    "raport_ll1 = classification_report(Y_test, Y_pred_ll1)\n",
    "print(raport_ll1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Podobna sytuacja jak w poprzednim przypadku, model ma wysoką dokładność, ze względu na niezbalansowane dane. Na mniej licznej klasie uzyskuje zerowa precyzje. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**6. Spróbujmy dostroić model lasów losowych i stworzyc kolejny model ll2**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# stwórzmy najpierw zestaw parametrów do testów\n",
    "params_ll2 = {\"max_depth\": list(range(2,36,3)),\n",
    "              \"max_features\" : list(range(3,16,3)),\n",
    "              \"max_leaf_nodes\" : list(range(2,58,5)),\n",
    "              \"n_estimators\" : list(range(50,200,10))}\n",
    "\n",
    "r=  0   # policzymy ile jest mozliwych kombinacji parametrow\n",
    "for j in params_ll2.values():\n",
    "    s=len(j)\n",
    "    r += s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomizedSearchCV(cv=5, estimator=RandomForestClassifier(random_state=1),\n",
       "                   n_iter=15.0,\n",
       "                   param_distributions={'max_depth': [2, 5, 8, 11, 14, 17, 20,\n",
       "                                                      23, 26, 29, 32, 35],\n",
       "                                        'max_features': [3, 6, 9, 12, 15],\n",
       "                                        'max_leaf_nodes': [2, 7, 12, 17, 22, 27,\n",
       "                                                           32, 37, 42, 47, 52,\n",
       "                                                           57],\n",
       "                                        'n_estimators': [50, 60, 70, 80, 90,\n",
       "                                                         100, 110, 120, 130,\n",
       "                                                         140, 150, 160, 170,\n",
       "                                                         180, 190]},\n",
       "                   random_state=1,\n",
       "                   scoring=make_scorer(f1_score, average=macro))"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# wylosujmy najlepsze kombinacje parametrów dla tego modelu.\n",
    "clf1 = RandomForestClassifier(random_state=1) \n",
    "\n",
    "random  = RandomizedSearchCV(clf1, param_distributions = params_ll2, cv = 5, n_iter = round(r/3,0), scoring = make_scorer(f1_score, average = \"macro\"), random_state = 1)\n",
    "\n",
    "random.fit(X,Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'n_estimators': 70, 'max_leaf_nodes': 52, 'max_features': 12, 'max_depth': 17}\n",
      "{'n_estimators': 100, 'max_leaf_nodes': 2, 'max_features': 3, 'max_depth': 5}\n",
      "{'n_estimators': 180, 'max_leaf_nodes': 22, 'max_features': 12, 'max_depth': 26}\n",
      "{'n_estimators': 50, 'max_leaf_nodes': 7, 'max_features': 6, 'max_depth': 11}\n",
      "{'n_estimators': 60, 'max_leaf_nodes': 7, 'max_features': 12, 'max_depth': 17}\n",
      "{'n_estimators': 140, 'max_leaf_nodes': 47, 'max_features': 3, 'max_depth': 2}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 27, 'max_features': 12, 'max_depth': 14}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 27, 'max_features': 12, 'max_depth': 14}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 27, 'max_features': 12, 'max_depth': 14}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 27, 'max_features': 12, 'max_depth': 14}\n"
     ]
    }
   ],
   "source": [
    "# otrzymamy liste 10 najlepszych zestawow parametrow z posrod wylosowanych\n",
    "ranks_ll2 = random.cv_results_['rank_test_score']\n",
    "ranks_ll2.sort()\n",
    "n = 10\n",
    "\n",
    "for i in range(n):\n",
    "    print(random.cv_results_['params'][ranks_ll2[i]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# dodajmy do modelu wylosowane najlepsze parametry\n",
    "new_params_ll2 = {\"max_depth\": list(range(10,21,2)),\n",
    "              \"max_features\" : [10,12],\n",
    "              \"max_leaf_nodes\" : list(range(20,31,2)),\n",
    "              \"n_estimators\" : list(range(70,100,10))}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 216 candidates, totalling 1080 fits\n"
     ]
    }
   ],
   "source": [
    "# wybierzmy teraz najlepszy zestaw parametrow z okrojonej listy parametrow\n",
    "grid_search_ll2 = GridSearchCV(clf1, param_grid = new_params_ll2, cv = 5, verbose = 1, scoring = make_scorer(f1_score, average = \"macro\")).fit(X,Y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 10, 'max_features': 12, 'max_leaf_nodes': 24, 'n_estimators': 90}"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# mamy proponowany najlepszy zestaw parametrow\n",
    "grid_search_ll2.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.959 \n",
      " acc test:  0.9521\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll2 =  RandomForestClassifier(max_depth= 14, max_features = 9,max_leaf_nodes= 22, n_estimators = 70, random_state=1).fit(X_train, Y_train)\n",
    "\n",
    "Y_pred_ll2 =  model_ll2.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll2.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll2.score(X_test, Y_test),4))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nasz model uzysał najwyzsza dokładność, ze wszystkich dotychczasowych modeli, ale to jeszcze nie znaczy, że jest dokładny."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll2 = confusion_matrix(Y_test, Y_pred_ll2)\n",
    "raport_ll2 = classification_report(Y_test, Y_pred_ll2)\n",
    "print(raport_ll2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Niestety i ta metoda nie daje nam dobrego rozwiązania właściwe klasyfikujacego klase \"1\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**7. Spróbujmy popracować trochę z naszym zbiorem i dostroić wagi pod niezbalansowany zbiór**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "#zdefiniujmy najpierw zakres wag jakie chcemy próbować w modelu\n",
    "weights = np.linspace(0.1,0.90,50)\n",
    "\n",
    "param_grid_weight = {'class_weight': [{0:x, 1:1.0-x} for x in weights]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 50 candidates, totalling 250 fits\n"
     ]
    }
   ],
   "source": [
    "# wybierzmy teraz najlepszy zestaw wag dla modelu\n",
    "grid_search_ll3 = GridSearchCV(model_ll2, param_grid = param_grid_weight, cv = 5, verbose = 1, scoring = make_scorer(f1_score, average = \"macro\")).fit(X,Y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'class_weight': {0: 0.1326530612244898, 1: 0.8673469387755102}}"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#w ten sposob uzyskalismy najlepszy model przy ponizszych wagach\n",
    "grid_search_ll3.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9343 \n",
      " acc test:  0.9134\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll3 =  RandomForestClassifier(max_depth= 14, max_features = 9,max_leaf_nodes= 22, n_estimators = 70,\n",
    "                                   class_weight= {0: 0.1163265306122449, 1: 0.883673469387755},random_state=1).fit(X_train, Y_train)\n",
    "\n",
    "Y_pred_ll3 =  model_ll3.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll3.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll3.score(X_test, Y_test),4))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dokladnosc naszego modelu mocno spadła, ale sprawdźmy czy poprawiła się inna metryka. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.94      0.95       935\n",
      "           1       0.25      0.40      0.31        47\n",
      "\n",
      "    accuracy                           0.91       982\n",
      "   macro avg       0.61      0.67      0.63       982\n",
      "weighted avg       0.93      0.91      0.92       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll3 = confusion_matrix(Y_test, Y_pred_ll3)\n",
    "raport_ll3 = classification_report(Y_test, Y_pred_ll3)\n",
    "print(raport_ll3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dzięki odpowiedniemu wyważeniu parametrów dla niezbalansowanej klasy poprawiliśmy precyzje i czulość, nie mniej jednak wskaźnik f1 dla klasy 1 jest nadal na niskim poziomie."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sprobujmy przypisac ze decyzja o przypisaniu do klasy 1 jest podejmowana gdy prawdopodobienstwo wynosi >0.3, >0,4, >0.6 lub >0.8\n",
    "decyzje = pd.DataFrame( model_ll3.predict_proba(X_test)[::,1], columns= [\"a\"])\n",
    "\n",
    "lista_03 = []\n",
    "lista_04 = []\n",
    "lista_06 = []\n",
    "lista_08 = []\n",
    "for i in decyzje[\"a\"]:\n",
    "    if i >0.6:\n",
    "        lista_06.append(1)\n",
    "    else:\n",
    "        lista_06.append(0)\n",
    "    if i >0.8:\n",
    "        lista_08.append(1)\n",
    "    else:\n",
    "        lista_08.append(0)\n",
    "    if i >0.4:\n",
    "        lista_04.append(1)\n",
    "    else:\n",
    "        lista_04.append(0)\n",
    "    if i >0.3:\n",
    "        lista_03.append(1)\n",
    "    else:\n",
    "        lista_03.append(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.99      0.97       935\n",
      "           1       0.33      0.13      0.18        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.65      0.56      0.58       982\n",
      "weighted avg       0.93      0.95      0.93       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# przeanalizujmy jak w tych wypadkach wyglada nasza macierz pomylek\n",
    "\n",
    "matrix_ll3_06 = confusion_matrix(Y_test, lista_06)\n",
    "raport_ll3_06 = classification_report(Y_test, lista_06)\n",
    "print(raport_ll3_06)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "matrix_ll3_08 = confusion_matrix(Y_test, lista_08)\n",
    "raport_ll3_08 = classification_report(Y_test, lista_08)\n",
    "print(raport_ll3_08)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.88      0.93       935\n",
      "           1       0.21      0.64      0.32        47\n",
      "\n",
      "    accuracy                           0.87       982\n",
      "   macro avg       0.60      0.76      0.62       982\n",
      "weighted avg       0.94      0.87      0.90       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "matrix_ll3_04 = confusion_matrix(Y_test, lista_04)\n",
    "raport_ll3_04 = classification_report(Y_test, lista_04)\n",
    "print(raport_ll3_04)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.81      0.89       935\n",
      "           1       0.16      0.70      0.26        47\n",
      "\n",
      "    accuracy                           0.81       982\n",
      "   macro avg       0.57      0.76      0.58       982\n",
      "weighted avg       0.94      0.81      0.86       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "matrix_ll3_03 = confusion_matrix(Y_test, lista_03)\n",
    "raport_ll3_03 = classification_report(Y_test, lista_03)\n",
    "print(raport_ll3_03)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Poprzez zmiane wartosci od ktorej przypisujemy dana prognoze do jednej z klas, nie widzimy nadal znaczacej poprawy wskaznika f1.\n",
    "Widzimy jednak, ze dzięki temu (wskaznik recall) poprawil sie i teraz jak już klasyfikujemy kogos jako chorego to z 70% dokladnoscia. Niestety klasyfikujemy tylko 16% wszystkich przypadkow."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**8. Wyprobujmy nasz najlepszy dotad model na danych przeskalowanych + PCA**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "#przeskalowanie danych dla calego zbioru treningowego\n",
    "scaler = StandardScaler().fit(X_train)\n",
    "X_train_scaled = scaler.transform(X_train)\n",
    "X_test_scaled = scaler.transform(X_test)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x3121cf40>]"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# wybranie odpowiedniej ilosci komponentow PCA\n",
    "\n",
    "## najpierw max kolumn, a potem sprawdzic ile jest nam rzeczywiscie potrzebnych \n",
    "PCA_results = PCA(n_components = 16)\n",
    "\n",
    "PCA_results.fit(X_train_scaled)\n",
    "X_train_scaled_PCA = PCA_results.transform(X_train_scaled)\n",
    "X_test_scaled_PCA = PCA_results.transform(X_test_scaled) \n",
    "\n",
    "plt.plot(PCA_results.explained_variance_ratio_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.18650245, 0.29209765, 0.38188196, 0.45606811, 0.52755008,\n",
       "       0.59169153, 0.65431283, 0.71500802, 0.77273208, 0.82782356,\n",
       "       0.8788745 , 0.92227164, 0.95422597, 0.97252306, 0.9891024 ,\n",
       "       1.        ])"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.cumsum(PCA_results.explained_variance_ratio_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Jak widzimy 10 zmiennych wyjasnia zmiennosc modelu w prawie 83%, ale skorzystajmy z 11, bo wariancja jeszcze sie calkiem mocna poprawia po dodaniu jednej zmiennej i mamy wynik 88%."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "#ostateczna wersja PCA\n",
    "PCA_results = PCA(n_components = 11)\n",
    "\n",
    "PCA_results.fit(X_train_scaled)\n",
    "X_train_scaled_PCA = PCA_results.transform(X_train_scaled)\n",
    "X_test_scaled_PCA = PCA_results.transform(X_test_scaled) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sprawdzmy teraz wynik modelu na zmiennych po standaryzacji i PCA, wykorzystamy najlepszy wytrenowany dotad model ll3\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9562 \n",
      " acc test:  0.9338\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll4 =  RandomForestClassifier(max_depth= 14, max_features = 9,max_leaf_nodes= 22, n_estimators = 70,\n",
    "                                   class_weight= {0: 0.1163265306122449, 1: 0.883673469387755}, random_state= 1).fit(X_train_scaled_PCA, Y_train)\n",
    "\n",
    "Y_pred_ll4 =  model_ll4.predict(X_test_scaled_PCA)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll4.score(X_train_scaled_PCA, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll4.score(X_test_scaled_PCA, Y_test),4))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.97      0.97       935\n",
      "           1       0.25      0.19      0.22        47\n",
      "\n",
      "    accuracy                           0.93       982\n",
      "   macro avg       0.60      0.58      0.59       982\n",
      "weighted avg       0.93      0.93      0.93       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll4 = confusion_matrix(Y_test, Y_pred_ll4)\n",
    "raport_ll4 = classification_report(Y_test, Y_pred_ll4)\n",
    "print(raport_ll4)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Poprawila nam sie accuracy calkowita i dla klasy 0, jednak dla klasy 1 precyzja i czulosc wciaz sa niskie."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**9. Ze wzgledu na skalowanie i PCA model moze wymagac innych parametrow w takim razie poszukajmy bardziej optymalnego zestawu parametrow**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "# stwórzmy najpierw zestaw parametrów do testów\n",
    "params_ll5 = {\"max_depth\": list(range(2,36,3)),\n",
    "              \"max_features\" : list(range(3,12,2)),\n",
    "              \"max_leaf_nodes\" : list(range(2,58,5)),\n",
    "              \"n_estimators\" : list(range(50,200,10))}\n",
    "\n",
    "r=  0   # policzymy ile jest mozliwych kombinacji parametrow\n",
    "for j in params_ll2.values():\n",
    "    s=len(j)\n",
    "    r += s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomizedSearchCV(cv=5, estimator=RandomForestClassifier(random_state=1),\n",
       "                   n_iter=15.0,\n",
       "                   param_distributions={'max_depth': [2, 5, 8, 11, 14, 17, 20,\n",
       "                                                      23, 26, 29, 32, 35],\n",
       "                                        'max_features': [3, 5, 7, 9, 11],\n",
       "                                        'max_leaf_nodes': [2, 7, 12, 17, 22, 27,\n",
       "                                                           32, 37, 42, 47, 52,\n",
       "                                                           57],\n",
       "                                        'n_estimators': [50, 60, 70, 80, 90,\n",
       "                                                         100, 110, 120, 130,\n",
       "                                                         140, 150, 160, 170,\n",
       "                                                         180, 190]},\n",
       "                   scoring=make_scorer(f1_score, average=macro))"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# wylosujmy najlepsze kombinacje parametrów dla tego modelu.\n",
    "clf5 = RandomForestClassifier(random_state=1) \n",
    "\n",
    "random  = RandomizedSearchCV(clf5, param_distributions = params_ll5, cv = 5, n_iter = round(r/3,0), scoring = make_scorer(f1_score, average = \"macro\"))\n",
    "\n",
    "random.fit(X_train_scaled_PCA, Y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'n_estimators': 100, 'max_leaf_nodes': 47, 'max_features': 7, 'max_depth': 2}\n",
      "{'n_estimators': 60, 'max_leaf_nodes': 52, 'max_features': 3, 'max_depth': 26}\n",
      "{'n_estimators': 160, 'max_leaf_nodes': 2, 'max_features': 11, 'max_depth': 17}\n",
      "{'n_estimators': 80, 'max_leaf_nodes': 42, 'max_features': 7, 'max_depth': 14}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 22, 'max_features': 5, 'max_depth': 8}\n",
      "{'n_estimators': 140, 'max_leaf_nodes': 32, 'max_features': 11, 'max_depth': 5}\n",
      "{'n_estimators': 170, 'max_leaf_nodes': 32, 'max_features': 5, 'max_depth': 14}\n",
      "{'n_estimators': 160, 'max_leaf_nodes': 27, 'max_features': 9, 'max_depth': 20}\n",
      "{'n_estimators': 160, 'max_leaf_nodes': 27, 'max_features': 9, 'max_depth': 20}\n",
      "{'n_estimators': 100, 'max_leaf_nodes': 22, 'max_features': 7, 'max_depth': 2}\n"
     ]
    }
   ],
   "source": [
    "# otrzymamy liste 10 najlepszych zestawow parametrow z posrod wylosowanych\n",
    "ranks_ll5 = random.cv_results_['rank_test_score']\n",
    "ranks_ll5.sort()\n",
    "n = 10\n",
    "\n",
    "for i in range(n):\n",
    "    print(random.cv_results_['params'][ranks_ll5[i]])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Niestety losowane parametry nie ukazały nam zbyt wiele prawidłowości, dlatego model zostanie wytrenowany na wszystkich parametrach. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [],
   "source": [
    "# dodajmy do modelu wylosowane najlepsze parametry\n",
    "new_params_ll5 =  {\"max_depth\": list(range(2,36,3)),\n",
    "              \"max_features\" : list(range(3,12,2)),\n",
    "              \"max_leaf_nodes\" : list(range(12,58,5)),\n",
    "              \"n_estimators\" : list(range(70,180,20))}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 5040 candidates, totalling 25200 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done 25200 out of 25200 | elapsed: 309.8min finished\n"
     ]
    }
   ],
   "source": [
    "# wybierzmy teraz najlepszy zestaw parametrow z okrojonej listy parametrow\n",
    "grid_search_ll5 = GridSearchCV(clf5, param_grid = new_params_ll5, cv = 5, verbose = 1, scoring = make_scorer(f1_score, average = \"macro\")).fit(X,Y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'max_depth': 29, 'max_features': 11, 'max_leaf_nodes': 52, 'n_estimators': 70}"
      ]
     },
     "execution_count": 231,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# mamy proponowany najlepszy zestaw parametrow\n",
    "grid_search_ll5.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9621 \n",
      " acc test:  0.9521\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll5 =  RandomForestClassifier(max_depth= 29, max_features = 11,max_leaf_nodes= 52, n_estimators = 70, random_state=1).fit(X_train_scaled_PCA, Y_train)\n",
    "\n",
    "Y_pred_ll5 =  model_ll5.predict(X_test_scaled_PCA)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll5.score(X_train_scaled_PCA, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll5.score(X_test_scaled_PCA, Y_test),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       973\n",
      "           1       0.00      0.00      0.00        49\n",
      "\n",
      "    accuracy                           0.95      1022\n",
      "   macro avg       0.48      0.50      0.49      1022\n",
      "weighted avg       0.91      0.95      0.93      1022\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1221: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll5 = confusion_matrix(Y_test, Y_pred_ll5)\n",
    "raport_ll5 = classification_report(Y_test, Y_pred_ll5)\n",
    "print(raport_ll5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Niestety ponownie wynik dobry tylko dla 1 klasy, pozostalo nam jeszcze sprawdzic czy zmiana wag cos pomoze."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 50 candidates, totalling 250 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Using backend SequentialBackend with 1 concurrent workers.\n",
      "[Parallel(n_jobs=1)]: Done 250 out of 250 | elapsed:  2.3min finished\n"
     ]
    }
   ],
   "source": [
    "# wybierzmy teraz najlepszy zestaw wag dla modelu\n",
    "grid_search_ll6 = GridSearchCV(model_ll5, param_grid = param_grid_weight, cv = 5, verbose = 1, scoring = make_scorer(f1_score, average = \"macro\")).fit(X,Y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'class_weight': {0: 0.1163265306122449, 1: 0.883673469387755}}"
      ]
     },
     "execution_count": 236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#w ten sposob uzyskalismy najlepszy model przy ponizszych wagach\n",
    "grid_search_ll6.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9408 \n",
      " acc test:  0.9237\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll6 =  RandomForestClassifier(max_depth= 14, max_features = 9,max_leaf_nodes= 22, n_estimators = 70,\n",
    "                                   class_weight= {0: 0.1163265306122449, 1: 0.883673469387755}, random_state= 1).fit(X_train_scaled_PCA, Y_train)\n",
    "\n",
    "Y_pred_ll6 =  model_ll6.predict(X_test_scaled_PCA)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll6.score(X_train_scaled_PCA, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll6.score(X_test_scaled_PCA, Y_test),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.96      0.96       973\n",
      "           1       0.19      0.18      0.19        49\n",
      "\n",
      "    accuracy                           0.92      1022\n",
      "   macro avg       0.58      0.57      0.57      1022\n",
      "weighted avg       0.92      0.92      0.92      1022\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll6 = confusion_matrix(Y_test, Y_pred_ll6)\n",
    "raport_ll6 = classification_report(Y_test, Y_pred_ll6)\n",
    "print(raport_ll6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sprawdzmy jeszcze przypisanie do klasy decyzyjnej przy 0.4 i 0.6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sprobujmy przypisac ze decyzja o przypisaniu jest podejmowana gdy prawdopodobienstwo wynosi >0,4, >0.6, >0.8\n",
    "decyzje2 = pd.DataFrame( model_ll6.predict_proba(X_test_scaled_PCA)[::,1], columns= [\"a\"])\n",
    "\n",
    "lista_04_2 = []\n",
    "lista_06_2 = []\n",
    "\n",
    "for i in decyzje2[\"a\"]:\n",
    "    if i >0.6:\n",
    "        lista_06_2.append(1)\n",
    "    else:\n",
    "        lista_06_2.append(0)\n",
    "    if i >0.4:\n",
    "        lista_04_2.append(1)\n",
    "    else:\n",
    "        lista_04_2.append(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.99      0.97       973\n",
      "           1       0.22      0.08      0.12        49\n",
      "\n",
      "    accuracy                           0.94      1022\n",
      "   macro avg       0.59      0.53      0.54      1022\n",
      "weighted avg       0.92      0.94      0.93      1022\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# przeanalizujmy jak w tych wypadkach wyglada nasza macierz pomylek\n",
    "matrix_ll6_06 = confusion_matrix(Y_test, lista_06_2)\n",
    "raport_ll6_06 = classification_report(Y_test, lista_06_2)\n",
    "print(raport_ll6_06)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.91      0.93       973\n",
      "           1       0.14      0.31      0.19        49\n",
      "\n",
      "    accuracy                           0.88      1022\n",
      "   macro avg       0.55      0.61      0.56      1022\n",
      "weighted avg       0.92      0.88      0.90      1022\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# przeanalizujmy jak w tych wypadkach wyglada nasza macierz pomylek\n",
    "matrix_ll6_04 = confusion_matrix(Y_test, lista_04_2)\n",
    "raport_ll6_04 = classification_report(Y_test, lista_04_2)\n",
    "print(raport_ll6_04)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**10.Jako kolejna metode walki z niezbalansowanym zbiorem wypróbujmy duplikacje sampli w klasie 1**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "#powiekszmy sobie nasz zbior dla klasy 1. o 2000 losowych obserwacji\n",
    "losowe_indeksy = np.random.randint(1,249, 2000)\n",
    "zbior1 = df_dummy[df_dummy.stroke == 1]\n",
    "\n",
    "duplikaty = pd.DataFrame()\n",
    "for i in losowe_indeksy:\n",
    "    zbior1 = zbior1.append(zbior1.iloc[i:i+1,])\n",
    "    \n",
    "zbior1.reset_index()\n",
    "    \n",
    "df_dummy_duplicated = df_dummy[df_dummy.stroke == 0].append(zbior1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# zdefiniujmy zbiór X i Y bez podzialu na testowy do cross validacji\n",
    "X_duplicated = df_dummy_duplicated.iloc[:,0:3].join(df_dummy.iloc[:,4:])\n",
    "Y_duplicated  = df_dummy_duplicated [\"stroke\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "# podzielmy też na zbiór treningowy i testowy\n",
    "X_train_duplicated, X_test_duplicated, Y_train_duplicated,Y_test_duplicated = train_test_split(X_duplicated,Y_duplicated, test_size = 0.2, random_state=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Spróbujmy znaleźć najlepsze parametry dla nowego modelu, wykorzystamy model random forrest."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [],
   "source": [
    "# stwórzmy najpierw zestaw parametrów do testów\n",
    "params_ll7 =  {\"max_depth\": list(range(2,36,3)),\n",
    "              \"max_features\" : list(range(3,12,2)),\n",
    "              \"max_leaf_nodes\" : list(range(2,58,5)),\n",
    "              \"n_estimators\" : list(range(70,200,20))}\n",
    "\n",
    "\n",
    "r=  0   # policzymy ile jest mozliwych kombinacji parametrow\n",
    "for j in params_ll2.values():\n",
    "    s=len(j)\n",
    "    r += s"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 300,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomizedSearchCV(cv=5, estimator=RandomForestClassifier(), n_iter=15.0,\n",
       "                   param_distributions={'max_depth': [2, 5, 8, 11, 14, 17, 20,\n",
       "                                                      23, 26, 29, 32, 35],\n",
       "                                        'max_features': [3, 5, 7, 9, 11],\n",
       "                                        'max_leaf_nodes': [2, 7, 12, 17, 22, 27,\n",
       "                                                           32, 37, 42, 47, 52,\n",
       "                                                           57],\n",
       "                                        'n_estimators': [70, 90, 110, 130, 150,\n",
       "                                                         170, 190]},\n",
       "                   scoring=make_scorer(f1_score, average=macro))"
      ]
     },
     "execution_count": 300,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# wylosujmy najlepsze kombinacje parametrów dla tego modelu.\n",
    "clf7 = RandomForestClassifier(random_state=1) \n",
    "\n",
    "random  = RandomizedSearchCV(clf7, param_distributions = params_ll7, cv = 5, n_iter = round(r/3,0), scoring = make_scorer(f1_score, average = \"macro\"))\n",
    "\n",
    "random.fit(X_train_duplicated, Y_train_duplicated)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'n_estimators': 110, 'max_leaf_nodes': 27, 'max_features': 3, 'max_depth': 17}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 22, 'max_features': 5, 'max_depth': 2}\n",
      "{'n_estimators': 170, 'max_leaf_nodes': 2, 'max_features': 9, 'max_depth': 14}\n",
      "{'n_estimators': 70, 'max_leaf_nodes': 37, 'max_features': 3, 'max_depth': 2}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 17, 'max_features': 5, 'max_depth': 23}\n",
      "{'n_estimators': 130, 'max_leaf_nodes': 17, 'max_features': 9, 'max_depth': 20}\n",
      "{'n_estimators': 70, 'max_leaf_nodes': 27, 'max_features': 9, 'max_depth': 29}\n",
      "{'n_estimators': 170, 'max_leaf_nodes': 32, 'max_features': 11, 'max_depth': 35}\n",
      "{'n_estimators': 190, 'max_leaf_nodes': 42, 'max_features': 5, 'max_depth': 2}\n",
      "{'n_estimators': 150, 'max_leaf_nodes': 2, 'max_features': 7, 'max_depth': 20}\n"
     ]
    }
   ],
   "source": [
    "# otrzymamy liste 10 najlepszych zestawow parametrow z posrod wylosowanych\n",
    "ranks_ll7 = random.cv_results_['rank_test_score']\n",
    "ranks_ll7.sort()\n",
    "n = 10\n",
    "\n",
    "for i in range(n):\n",
    "    print(random.cv_results_['params'][ranks_ll7[i]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.6971 \n",
      " acc test:  0.6881\n"
     ]
    }
   ],
   "source": [
    "# sprawdzmy od razu czy jest sens dalej szukac parametrow, czyli jak sie sprawuje nasz model ze wskazanymi parametrami\n",
    "model_ll7 =  RandomForestClassifier(max_depth= 20, max_features = 11,max_leaf_nodes= 27, n_estimators = 110,random_state=1).fit(X_train_duplicated, Y_train_duplicated)\n",
    "\n",
    "Y_pred_ll7 =  model_ll7.predict(X_test_duplicated)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_ll7.score(X_train_duplicated, Y_train_duplicated),4), \"\\n\",\n",
    "      \"acc test: \", round(model_ll7.score(X_test_duplicated, Y_test_duplicated),4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Wynik Acc bardzo niski, jak to wyglada dalej:\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.69      0.98      0.81       946\n",
      "           1       0.55      0.06      0.11       436\n",
      "\n",
      "    accuracy                           0.69      1382\n",
      "   macro avg       0.62      0.52      0.46      1382\n",
      "weighted avg       0.65      0.69      0.59      1382\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_ll7 = confusion_matrix(Y_test_duplicated, Y_pred_ll7)\n",
    "raport_ll7 = classification_report(Y_test_duplicated, Y_pred_ll7)\n",
    "print(raport_ll7)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ta metoda w tym przypadku nie zdała egzaminu.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**11.Ostatnią metodą bedzie regresja logistyczna na przeskalowanych danych + PCA**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_lr3 = LogisticRegression(random_state=1)\n",
    "model_lr3.fit(X_train_scaled_PCA, Y_train)\n",
    "Y_pred_lr3 = model_lr3.predict(X_test_scaled_PCA)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9587 \n",
      " acc test:  0.9521\n"
     ]
    }
   ],
   "source": [
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_lr3.score(X_train_scaled_PCA, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_lr3.score(X_test_scaled_PCA, Y_test),4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_lr3 = confusion_matrix(Y_test, Y_pred_lr3)\n",
    "raport_lr3 = classification_report(Y_test, Y_pred_lr3)\n",
    "print(raport_lr3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Niestety nic z tego, nie widzę potrzeby dalszych prób."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**12. Podsumowanie**  \n",
    "Wykorzystujac przedstawione tutaj metody, najlepszy wyniki, przedstawiony za pomoca F1 score dla obu klas, uzyskaliśmy stosujac model model_ll3 o parametrach jak ponizej."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 247,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(class_weight={0: 0.1163265306122449,\n",
       "                                     1: 0.883673469387755},\n",
       "                       max_depth=14, max_features=9, max_leaf_nodes=22,\n",
       "                       n_estimators=70)"
      ]
     },
     "execution_count": 247,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_ll3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "nasz wynik to: \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.91      0.94       973\n",
      "           1       0.21      0.45      0.28        49\n",
      "\n",
      "    accuracy                           0.89      1022\n",
      "   macro avg       0.59      0.68      0.61      1022\n",
      "weighted avg       0.93      0.89      0.91      1022\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"nasz wynik to: \\n\", raport_ll3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEGCAYAAABo25JHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAYe0lEQVR4nO3dfZBV9X3H8fd3H3lmkV0ReVpQUDFqkI0aNWo0RiS2TFo7MSaxMclYGsmkaR60mUmdaTpTbdLW2GgYxhKrzYROq6kkpVpTTTFVI6g8CKgsGGABZXl+WNjdu/fbP+6FXJbdu3eXe87Ze3+f1wwznHN+y36O4P3sefodc3dERCRcFUkHEBGRZKkIREQCpyIQEQmcikBEJHAqAhGRwFUlHaC/6uvrvbGxMekYIiIl5bXXXtvt7g09bSu5ImhsbGTlypVJxxARKSlmtqW3bTo1JCISOBWBiEjgVAQiIoFTEYiIBE5FICISuMiKwMwWm9kuM3uzl+1mZg+ZWbOZrTGzS6PKIiIivYvyiOAxYE6e7TcD07O/7gJ+FGEWERHpRWTPEbj7cjNrzDNkHvC4Z+bBfsXM6sxsvLvvjCqTiMTrydda2LLnSNIxykZT4xlcM6PHZ8JOS5IPlE0AtuUst2TXnVIEZnYXmaMGJk+eHEs4ETk9uw+38/V/Ww2AWcJhysT8a88puyLo6Z9Gj2/JcfdFwCKApqYmvUlHpAS8vmUfAP82/8N8qPGMhNNIPkneNdQCTMpZngjsSCiLiBTZ61v3U1VhXDRhdNJRpA9JHhEsBRaY2RLgcuCArg+IRKc91cXLm/aQ6ornoPrFja1cePYohlRXxvL9ZOAiKwIz+ylwHVBvZi3AfUA1gLsvBJYBc4FmoA24M6osIgI/X72Tb2TP2cflT66dFuv3k4GJ8q6hT/ex3YG7o/r+InKyo51dADz+hcsYM6wm8u9nBjPGjYz8+8jpK7lpqEXk9FwwfhQNI2uTjiGDiIpApMR1pZ09R9r7HHfoWGcMaaQUqQhESty3n1rLv67c1vfArOpK3dQvJ1MRiJS49w4eY0LdUP70unP6HDtu1BDqYrg+IKVFRSBSBupH1vLZK6YkHUNKlKahFhEJnI4IRErEEy//lodf2HTK+r1HOph59qj4A0nZUBGIlIiVW/ZxuD3FJy4af8q2j55f/InIJBwqApESUj+ihgduvTjpGFJmdI1ARCRwOiIQ6YdDxzq57+l1HG5Pxf6917QcYEi1fnaT4lMRiPTDhp2HeOqN7Uw6YyjDa+L936duWDXXnqdrAVJ8KgKRAbj/Dy7mqnPrk44hUhQ6zhQRCZyOCCQ4L23azRtb9w/oa1v2tRU5jUjyVAQSnPueXsfGXYcH/PW1VRWMGzWkiIlEkqUikOB0uTP3orN48FOzBvT1FQZVlTqrKuVDRSBBqjCjpkof5iKgIpBByN15fes+jrR3RfLnH+2I5s8VKVUqAhl0Vm3bzx/+6OVIv8fIIfqnL3Kc/m+QQef4T+zfnXdhZLNqXjBes3WKHKcikEFrxriRzJ5yRtIxRMqeikBidbSji45UOu+YJObxEQmZikBis6n1MHMeXE5nlxc0XrdoisRDRSCxaT3UTmeXc8eHp9A4dnjesSNqq7hk4uiYkomETUUgsZvzgbO48hxN2CYyWOjYW0QkcDoikNPyt8+8xX+u3VnQ2GOdepBLZDBSEchpeXHjbo51dvHhaWMLGj+8toqLJ9ZFnEpE+kNFIKftwrNH8+BtA5vATUSSp2sEIiKBi/SIwMzmAD8AKoFH3f3+bttHA/8CTM5m+b67/zjKTHKqje8f4pFfbSKVLuz+/lxb97bRMLI2glQiEpfIisDMKoGHgRuBFmCFmS119/U5w+4G1rv775lZA/C2mf3E3TuiyiWnem7D+/zsje00jh1GhVm/vnbs8Bqu1rt7RUpalEcElwHN7r4ZwMyWAPOA3CJwYKSZGTAC2AtofoGEPPNn1zCkujLpGCISsyivEUwAtuUst2TX5fohcAGwA1gLfNXdT5mIxszuMrOVZraytbU1qrwiIkGK8oigp3MM3U9C3wSsAq4HzgGeM7MX3f3gSV/kvghYBNDU1NT/E9nCC2/tormX9/S++u7emNOIyGASZRG0AJNylieS+ck/153A/e7uQLOZvQucD7waYa4gfeWnb+Sd1XPcqFqqNcmbSJCiLIIVwHQzmwpsB24Dbu82ZitwA/CimY0DzgM2R5gpWKl0mjuvauTrHz+vx+21VRVUVvTvQrGIlIfIisDdU2a2AHiWzO2ji919nZnNz25fCHwXeMzM1pI5lXSPu++OKlPoaiorGFGrZwhF5GSRfiq4+zJgWbd1C3N+vwP4eJQZQrVlzxHeO3DsxHI6/7tgRCRg+vGwDHWlnZseXM6xzpM//YfV6K9bRE6lT4YylHbnWGeaP5o9kU/Oytyxa2bMmqzJ3kTkVCqCMjZl7DCu1FO/ItIHFUGZSHWlT8wV1NmlCwIiUjgVQRk41tnF1Q88z+7DJ0/RVFmh5wJEpG8qgjLQ1tHF7sMdfOyCccyeMgaAygr45KyJCScTkVKgIigjH5lezx9f2Zh0DBEpMTp3ICISOBVBiXt39xE+++hvko4hIiVMRVDiNuw8yPqdB7n+/DO5ZkZD0nFEpASpCMrEPXPOZ2r98KRjiEgJUhGIiARORSAiEjgVgYhI4FQEIiKBUxGIiAROTxYPYkfaU/x89Q468kwit277wRgTiUg5UhEMYv+9/j3ufWptn+NqqioYM7w6hkQiUo5UBINYZ1dmWulffOVqxo8e0uu4oTWVevuYiAyYPj1KQN2wasaOqE06hoiUKRXBIOTuNO86zNY9bUlHEZEAqAgGoXU7DnLLP/76xPKQ6soE04hIuVMRDEIHj3UCmfmDrjp3LPU6LSQiEdJzBIPYrMl1XDyxLukYIlLmVAQiIoFTEYiIBE5FICISOBWBiEjgVAQiIoFTEQxCP/jlxqQjiEhAIi0CM5tjZm+bWbOZ3dvLmOvMbJWZrTOz/40yT6nYkn2i+ILxoxJOIiIhiOyBMjOrBB4GbgRagBVmttTd1+eMqQMeAea4+1YzOzOqPKWkssK4dfZERg/VjKIiEr0ojwguA5rdfbO7dwBLgHndxtwOPOXuWwHcfVeEeUREpAdRFsEEYFvOckt2Xa4ZwBgz+5WZvWZmd/T0B5nZXWa20sxWtra2RhRXRCRMURaB9bDOuy1XAbOBTwA3Ad8xsxmnfJH7IndvcvemhoaG4icVEQlYlJPOtQCTcpYnAjt6GLPb3Y8AR8xsOXAJ8E6EuUREJEeURwQrgOlmNtXMaoDbgKXdxjwNfMTMqsxsGHA5sCHCTCIi0k1kRwTunjKzBcCzQCWw2N3Xmdn87PaF7r7BzJ4B1gBp4FF3fzOqTKVg/Y6DbN9/NOkYIhKQSN9H4O7LgGXd1i3stvw94HtR5iglf/HUGgAaRuodBCISDz1ZPMi0p9JcPHE037rpvKSjiEggVASDjJkxbtQQzHq66UpEpPj0qsqEpbrSbNx1GM/eWNve2ZVsIBEJjoogYQ+/sIl/+OXJd8teOGF0QmlEJEQqgoQdONrJkOoKHvzUrBPrLp2i9xSLSHxUBINAdUUFcz5wVtIxRCRQ/b5YbGaVZvaZKMKIiEj8ej0iMLNRwN1kJopbCjwHLAC+AawCfhJHwHKy5NWtfOfpN09cGAbocmdkrQ7MRCQ5+T6BngD2AS8DXwK+CdQA89x9VQzZys7GXYcB+JNrp520Xi+gEZEk5SuCae5+EYCZPQrsBia7+6FYkpWp2qpKvnnT+UnHEBE5Id81gs7jv3H3LuBdlYCISPnJd0RwiZkd5HfvFRias+zurvMZfXjmzfdYtHzTieVt+zSZnIgMPr0WgbtXxhmkHD3/1vu8uf0gl087A4DzzxrJRXpYTEQGmXx3DQ0B5gPnkpkmerG7p+IKVi7GjqjhiS9ennQMEZFe5btG8M9AE7AWmAv8XSyJREQkVvmuEczMuWvon4BX44lUPpa/s5u0d39Ns4jI4JKvCHLvGkppWuT+O3isk1RaRSAig1u+Ivhg9i4hyNwppLuG+qmmqoI/uuTspGOIiOSVrwhWu/usPNtFRKQM5LtYrHMapyGddva3dfY9UEQkYfmOCM40sz/vbaO7/30EecrG2u0HAGjr0BvHRGRwy1cElcAIfvdksfRDeyoNwO9/UNcIRGRwy1cEO939r2JLUqYqdLeViAxy+YpAn2ADsG1vG+2pNNv3tyUdRUSkIPmK4IbYUpSJF97exZ0/XnHSutqqfr8ETkQkVvkmndsbZ5BysL+tA4Dv3DKThpG1DK+pZNbkMQmnEhHJT+9IjMAN559JY/3wpGOIiBRE5y1ERAKnI4Ii+fyPX+WlTXsA3SkkIqVFRVAka1oOcE7DCH7vkvFMHDM06TgiIgVTERRR05QxfPm6c5OOISLSL5FeIzCzOWb2tpk1m9m9ecZ9yMy6zOzWKPOIiMipIisCM6sEHgZuBmYCnzazmb2MewB4NqosUTvQ1sneIx1JxxARGZAojwguA5rdfbO7dwBLgHk9jPsK8CSwK8Iskdq46xAAdcOqE04iItJ/URbBBGBbznJLdt0JZjYB+CSwMN8fZGZ3mdlKM1vZ2tpa9KDFctnUM5KOICLSb1EWQU/3UHZ/x8GDwD3unneuZndf5O5N7t7U0NBQtIAiIhLtXUMtwKSc5YnAjm5jmoAl2fch1wNzzSzl7v8RYS4REckRZRGsAKab2VRgO3AbcHvuAHefevz3ZvYY8AuVgIhIvCIrAndPmdkCMncDVQKL3X2dmc3Pbs97XUBEROIR6QNl7r4MWNZtXY8F4O6fjzKLiIj0TJPOiYgETkUgIhI4FYGISOBUBCIigdPso6dh18FjtKfS7DrUnnQUEZEBUxEM0K837uaz//Sbk9bVVOoAS0RKj4pggPYcyRwFfGvOeTSMqGV4bRWzp+hF9SJSelQEp2nOhWcxrWFE0jFERAZM5zJERAKnIhigb/37GgBML6oXkRKnIhig9lSaIdUVTDljWNJRREROi4pggKorjS9cNZWKCh0RiEhpUxGIiAROdw3104rf7uWRF5rp7Or+sjURkdKkI4J++uX69/nVO63MmlzH1efWJx1HROS06YhgAGqrKvjZl69KOoaISFHoiEBEJHAqgn54edMeVvx2b9IxRESKSkXQD9979i1e37qf88aNTDqKiEjRqAj6ocvh2hkNPL3g6qSjiIgUjYpARCRwKgIRkcCpCEREAqciEBEJnIpARCRwerK4AOm0s/twO52pdNJRRESKTkVQgL/+zw0s/r93AfhY3biE04iIFJeKoADvHzpG/YhavnbjdD48bWzScUREikpFUKDRQ6v4zOVTko4hIlJ0ulgsIhK4SI8IzGwO8AOgEnjU3e/vtv0zwD3ZxcPAn7r76igzFeLhF5p54uUtJ5b3tXUwSe8mFpEyFVkRmFkl8DBwI9ACrDCzpe6+PmfYu8C17r7PzG4GFgGXR5WpUL95dy8dXWluvOB3F4avPFfXBkSkPEV5RHAZ0OzumwHMbAkwDzhRBO7+Us74V4CJEebpl8lnDOOBWy9OOoaISOSivEYwAdiWs9ySXdebLwL/1dMGM7vLzFaa2crW1tYiRhQRkSiPCKyHdT2+8d3MPkqmCHqc39ndF5E5bURTU1NR3xqfTjv3LV3HewePnVi3bvsBXRMQkWBEWQQtwKSc5YnAju6DzOxi4FHgZnffE2GeHu050sETr2zhrFFDGDO8BoAzRw3hxpl6cExEwhBlEawAppvZVGA7cBtwe+4AM5sMPAV8zt3fiTBLn+6+/lw+d4WeExCR8ERWBO6eMrMFwLNkbh9d7O7rzGx+dvtC4C+BscAjZgaQcvemqDKJiMipIn2OwN2XAcu6rVuY8/svAV+KMkNffr76lLNVIiJBCfrJ4s6uNH/1i8zdrJPGDE04jYhIMoIuAs/ef/S1j83guvPOTDaMiEhCgi6C46oqe7rTVUQkDMHOPnq4PcWK3+5NOoaISOKCLYKH/mcji5ZvBmBEbbD/GUREwi2Cto4Uo4ZU8fgXL+cDZ49KOo6ISGKCLQKA6soKPjipLukYIiKJCvZi8dGONF1e1GmLRERKUpBFsG1vG0++3kJnKp10FBGRxAVZBHuOdADwWc0tJCISZhEcd8U0vXVMRCToIhARkQCL4OVNe7j7J68nHUNEZNAIrgjWtOxn+/6jfKppEpdOHpN0HBGRxAX7HMF9vz+TYTXB7r6IyAnBHRGIiMjJVAQiIoFTEYiIBE5FICISOBWBiEjggiuCZWt3Jh1BRGRQCaoIDrR1srrlAAA1lUHtuohIr4L6NEylM7ONfueWmVSpCEREgMCK4LhqvaxeROSEIItARER+R0UgIhI4FYGISOBUBCIigQuqCDq79LJ6EZHugimC17fu4+oHngegskJ3DYmIHBdMEezcf4xU2pl/7Tl84qLxSccRERk0Ii0CM5tjZm+bWbOZ3dvDdjOzh7Lb15jZpVHmAfiDSydQN6wm6m8jIlIyIisCM6sEHgZuBmYCnzazmd2G3QxMz/66C/hRVHlERKRnUR4RXAY0u/tmd+8AlgDzuo2ZBzzuGa8AdWam8zYiIjGKsggmANtylluy6/o7BjO7y8xWmtnK1tbWAYU5a/QQ5l50FiNq9Z5iEZFcUX4q9nRrTvf7NwsZg7svAhYBNDU1Dege0NlTxjB7yuyBfKmISFmL8oigBZiUszwR2DGAMSIiEqEoi2AFMN3MpppZDXAbsLTbmKXAHdm7h64ADri73hwjIhKjyE4NuXvKzBYAzwKVwGJ3X2dm87PbFwLLgLlAM9AG3BlVHhER6VmkV07dfRmZD/vcdQtzfu/A3VFmEBGR/IJ5slhERHqmIhARCZyKQEQkcCoCEZHAWeZ6bekws1ZgywC/vB7YXcQ4pUD7HAbtcxhOZ5+nuHtDTxtKrghOh5mtdPempHPESfscBu1zGKLaZ50aEhEJnIpARCRwoRXBoqQDJED7HAbtcxgi2eegrhGIiMipQjsiEBGRblQEIiKBK8siMLM5Zva2mTWb2b09bDczeyi7fY2ZXZpEzmIqYJ8/k93XNWb2kpldkkTOYuprn3PGfcjMuszs1jjzRaGQfTaz68xslZmtM7P/jTtjsRXwb3u0mf3czFZn97mkZzE2s8VmtsvM3uxle/E/v9y9rH6RmfJ6EzANqAFWAzO7jZkL/BeZN6RdAfwm6dwx7POVwJjs728OYZ9zxj1PZhbcW5POHcPfcx2wHpicXT4z6dwx7PO3gQeyv28A9gI1SWc/jX2+BrgUeLOX7UX//CrHI4LLgGZ33+zuHcASYF63MfOAxz3jFaDOzMbHHbSI+txnd3/J3fdlF18h8za4UlbI3zPAV4AngV1xhotIIft8O/CUu28FcPdS3+9C9tmBkWZmwAgyRZCKN2bxuPtyMvvQm6J/fpVjEUwAtuUst2TX9XdMKenv/nyRzE8UpazPfTazCcAngYWUh0L+nmcAY8zsV2b2mpndEVu6aBSyzz8ELiDzmtu1wFfdPR1PvEQU/fMr0hfTJMR6WNf9HtlCxpSSgvfHzD5KpgiujjRR9ArZ5weBe9y9K/PDYskrZJ+rgNnADcBQ4GUze8Xd34k6XEQK2eebgFXA9cA5wHNm9qK7H4w6XEKK/vlVjkXQAkzKWZ5I5ieF/o4pJQXtj5ldDDwK3Ozue2LKFpVC9rkJWJItgXpgrpml3P0/4olYdIX+297t7keAI2a2HLgEKNUiKGSf7wTu98wJ9GYzexc4H3g1noixK/rnVzmeGloBTDezqWZWA9wGLO02ZilwR/bq+xXAAXffGXfQIupzn81sMvAU8LkS/ukwV5/77O5T3b3R3RuBfwe+XMIlAIX9234a+IiZVZnZMOByYEPMOYupkH3eSuYICDMbB5wHbI41ZbyK/vlVdkcE7p4yswXAs2TuOFjs7uvMbH52+0Iyd5DMBZqBNjI/UZSsAvf5L4GxwCPZn5BTXsIzNxa4z2WlkH129w1m9gywBkgDj7p7j7chloIC/56/CzxmZmvJnDa5x91LdnpqM/spcB1Qb2YtwH1ANUT3+aUpJkREAleOp4ZERKQfVAQiIoFTEYiIBE5FICISOBWBiEjgVAQiBcrOYLoq51djdqbPA2b2hpltMLP7smNz179lZt9POr9Ib8ruOQKRCB119w/mrjCzRuBFd7/FzIYDq8zsF9nNx9cPBd4ws5+5+//FG1mkbzoiECmS7LQOr5GZ7yZ3/VEyc+GU8sSGUsZUBCKFG5pzWuhn3Tea2Vgy88Ov67Z+DDAdWB5PTJH+0akhkcKdcmoo6yNm9gaZKR3uz06BcF12/Royc9/c7+7vxZhVpGAqApHT96K739LbejObAfw6e41gVdzhRPqiU0MiEcvO9vo3wD1JZxHpiYpAJB4LgWvMbGrSQUS60+yjIiKB0xGBiEjgVAQiIoFTEYiIBE5FICISOBWBiEjgVAQiIoFTEYiIBO7/ATjOzqIb2UKWAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#wyliczmy AUC-ROC dla naszego najlepszego modelu\n",
    "Y_pred_proba_ll3 = model_ll3.predict_proba(X_test)[::,1]\n",
    "FPR, TPR,_ = roc_curve(Y_test,Y_pred_proba_ll3)\n",
    "AUC = roc_auc_score(Y_test,Y_pred_proba_ll3)\n",
    "\n",
    "plt.plot(FPR, TPR, label = \"wynik \" + str(AUC))\n",
    "plt.xlabel(\"FPR\")\n",
    "plt.ylabel(\"TPR\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Szukamy pkt, w którym mamy najwyzsze TPR i wzglednie niskie FPR,\n",
    "możemy uznać, że właściwym progiem klasyfikacji dla tego modelu bedzie nawet 0.2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Sprawdzmy jak rzeczywiscie prezentuje sie modele przy klasyfikacji od progu 0.8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "#przypiszmy, więc decyzje na podstawie prawdopodobienstwo 0.2 i wyzej do klasy 1\n",
    "\n",
    "decyzje = pd.DataFrame( model_ll3.predict_proba(X_test)[::,1], columns= [\"a\"])\n",
    "\n",
    "lista_02 = []\n",
    "for i in decyzje[\"a\"]:\n",
    "    if i >0.2:\n",
    "        lista_02.append(1)\n",
    "    else:\n",
    "        lista_02.append(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.99      0.68      0.80       975\n",
      "           1       0.11      0.85      0.20        47\n",
      "\n",
      "    accuracy                           0.69      1022\n",
      "   macro avg       0.55      0.76      0.50      1022\n",
      "weighted avg       0.95      0.69      0.78      1022\n",
      "\n"
     ]
    }
   ],
   "source": [
    "matrix_ll3_02 = confusion_matrix(Y_test, lista_02)\n",
    "raport_ll3_02 = classification_report(Y_test, lista_02)\n",
    "print(raport_ll3_02)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Nasz finalny model charakteryzuje się niską acc, bo 0.69, ale za to:**\n",
    "- poprawnie rozpoznaje 99% klasy 0,\n",
    "- jezeli juz przypisze kogos do klasy 1, to z 85% prawdopodobienstwem \n",
    "\n",
    "\n",
    "Niestety, przypisuje poprawnie tylko 11% objektów klasy 1."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**II Zaawansowane modele**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**1.wykorzystajmy w naszej analizie model XGBoost**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\xgboost\\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].\n",
      "  warnings.warn(label_encoder_deprecation_msg, UserWarning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc train:  0.9982 \n",
      " acc test:  0.946\n"
     ]
    }
   ],
   "source": [
    "# prosty model xgb, nr1\n",
    "\n",
    "model_xgb1= XGBClassifier(objective = 'binary:logistic',random_state=1).fit(X_train, Y_train, eval_metric = \"logloss\")\n",
    "\n",
    "Y_pred_xgb1 =  model_xgb1.predict(X_test)\n",
    "\n",
    "#ocena modelu na danych treningowych i testowych \n",
    "print(\"acc train: \",round(model_xgb1.score(X_train, Y_train),4), \"\\n\",\n",
    "      \"acc test: \", round(model_xgb1.score(X_test, Y_test),4))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.99      0.97       935\n",
      "           1       0.25      0.06      0.10        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.60      0.53      0.54       982\n",
      "weighted avg       0.92      0.95      0.93       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_xgb1 = confusion_matrix(Y_test, Y_pred_xgb1)\n",
    "raport_xgb1 = classification_report(Y_test, Y_pred_xgb1)\n",
    "print(raport_xgb1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Widzimy, że prostym modelem XGB nie poprawiliśmy wyników dla 1 klasy.  \n",
    "Spróbujmy znaleźć bardziej optymalny zestaw parametrów przy użyciu optymalizatowra hyperopt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [],
   "source": [
    "#dane musza byc w odpowiednim formacie, dlatego dostosujmy df do formatu array\n",
    "X_train_array = X_train.values\n",
    "X_test_array = X_test.values\n",
    "\n",
    "#dostosujmy wartosci Y, za pomoca LabelEncoder\n",
    "le  = LabelEncoder()\n",
    "Y_train_encoded = le.fit_transform(Y_train)\n",
    "Y_test_encoded = le.fit_transform(Y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_xgb_hyperopt(X,Y,cv, num_eval):\n",
    "    def objective(parametry): #zdefiniujmy funkcje celu\n",
    "        model = XGBClassifier(**parametry, random_state = 10, objective = \"binary:logistic\",\n",
    "                              n_jobs = 2,use_label_encoder=False, scale_pos_weight = 10)\n",
    "                \n",
    "        for a, b in cv.split(X, Y):\n",
    "            eval_set = [(X[a],Y[a]),(X[b],Y[b])]\n",
    "            model.fit(X[a],Y[a],eval_set = eval_set, eval_metric = 'logloss', early_stopping_rounds = 10, verbose = 0) \n",
    "            y_pred = model.predict(X[b])\n",
    "            score =cohen_kappa_score(Y[b],y_pred)\n",
    "           # print(\"kappa score: {:.4f}\".format(score))\n",
    "            return{'loss': -score, 'status':STATUS_OK}\n",
    "        \n",
    "    parametry ={\n",
    "        'max_depth': scope.int(hp.quniform('max_depth',5,15,1)),\n",
    "        'n_estimators' : hp.choice('n_estimators', np.arange(20, 150, 2, dtype=int)),\n",
    "        'learning_rate' : hp.quniform('learning_rate', 0.001, 0.5, 0.005),\n",
    "        'min_child_weight': hp.quniform('min_child_weight', 1, 10, 1),\n",
    "        'subsample': hp.uniform('subsample', 0.7, 1),\n",
    "        'gamma' : hp.quniform('gamma', 0.5, 1, 0.05),\n",
    "        'colsample_bytree' : hp.quniform('colsample_bytree', 0.5, 1, 0.05)\n",
    "        }  \n",
    "    \n",
    "    trials = Trials()\n",
    "    \n",
    "    best = fmin(fn=objective,\n",
    "                space = parametry,\n",
    "                algo = tpe.suggest,\n",
    "                max_evals = num_eval,\n",
    "                trials = trials)\n",
    "    hyperparams = space_eval(parametry, best)\n",
    "    \n",
    "    print(\"Najlepsze parametry: \", hyperparams)\n",
    "    return hyperparams"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "#zdefiniujmy jeszcze rodzaj walidacji, jako wykorzystamy\n",
    "kF = StratifiedKFold(n_splits = 5, shuffle= True, random_state= 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|██████| 20/20 [00:10<00:00,  1.86trial/s, best loss: -0.14540595771437548]\n",
      "Najlepsze parametry:  {'colsample_bytree': 0.6000000000000001, 'gamma': 0.75, 'learning_rate': 0.405, 'max_depth': 6, 'min_child_weight': 10.0, 'n_estimators': 22, 'subsample': 0.7201127678892102}\n"
     ]
    }
   ],
   "source": [
    "# odaplenie hyeropt\n",
    "best_XGB = run_xgb_hyperopt(X_train_array, Y_train_encoded, cv=kF, num_eval=20)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "# uruchomie teraz nasz najlepszy model\n",
    "model_xgb2 = XGBClassifier(**best_XGB, objective= \"binary:logistic\",use_label_encoder=False,random_state=1).fit(X_train,Y_train_encoded, eval_metric = 'logloss')\n",
    "Y_pred_xgb2 = model_xgb2.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.50      0.02      0.04        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.73      0.51      0.51       982\n",
      "weighted avg       0.93      0.95      0.93       982\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_xgb2 = confusion_matrix(Y_test_encoded, Y_pred_xgb2)\n",
    "raport_xgb2 = classification_report(Y_test_encoded, Y_pred_xgb2)\n",
    "print(raport_xgb2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Jak widzimy bardziej zaawansowana metoda, jako jest XGBoost wraz z optymalizatorem parametrów przy użyciu hyperopta nie poprawiła znacznie naszego najlepszego wyniku osiagnietego przez model RandomForrest.\n",
    "Pokazuje nam z 98% dokladnościa przypadki braku ataku serca, ale bardzo słabo przewiduje faktyczne ataki serca"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**III Sieć neuronowa**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 299,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from keras import layers\n",
    "from tensorflow.keras import layers, callbacks\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from imblearn.over_sampling import SMOTE\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [],
   "source": [
    "# transformujemy Y_test i Y_train\n",
    "\n",
    "Y_train = np.array(Y_train)\n",
    "Y_test = np.array(Y_test)\n",
    "\n",
    "Y_test = OneHotEncoder().fit_transform(Y_test.reshape(-1,1)).toarray()\n",
    "Y_train = OneHotEncoder().fit_transform(Y_train.reshape(-1,1)).toarray()\n",
    "\n",
    "# zmienmy jeszcze format\n",
    "Y_test = Y_test.astype(np.float32)\n",
    "Y_train = Y_train.astype(np.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 335,
   "metadata": {},
   "outputs": [],
   "source": [
    "#definicja modelu\n",
    "def model_nn(X,Y, units, drop, classNum, epoch, lr = 0.01, bn = False):\n",
    "    '''\n",
    "    X - dane treningowe w postaci array bez labeli,\n",
    "    Y - labeli dla danych treningowych\n",
    "    units - lista unitow dla poszczegolnych warstw\n",
    "    drop - poziom dropoutu dla poszczegolnej wartswy\n",
    "    classNum = ilosc klas \n",
    "    epoch - liczba epok podczas trenowania \n",
    "    lr - learning rate\n",
    "    bn - czy zastosowac batch normalization \n",
    "    '''\n",
    "    model = tf.keras.models.Sequential()\n",
    "    model.add(layers.InputLayer(input_shape = X.shape[1]))\n",
    "    for i, j in enumerate(units):\n",
    "        model.add(layers.Dense(j, activation='relu', name=\"ukryta_\" + str(i)))\n",
    "        model.add(layers.Dropout(drop))\n",
    "        if bn == True:\n",
    "            model.add(layers.BatchNormalization())\n",
    "                  \n",
    "    model.add(layers.Dense(classNum, name=\"WYJSCIE\"))\n",
    "    model.add(layers.Softmax())\n",
    "    \n",
    "    model.summary()\n",
    "    \n",
    "    optymalizator = tf.optimizers.Adam(lr)\n",
    "    \n",
    "    model.compile(optimizer=optymalizator,\n",
    "              loss=tf.keras.losses.BinaryCrossentropy(),\n",
    "              metrics=['AUC'])\n",
    "    \n",
    "    results = model.fit(X,Y, validation_split= 0.2, epochs=epoch, verbose=1,\n",
    "              callbacks=[callbacks.EarlyStopping(monitor= 'val_auc', min_delta=0.1, \n",
    "                                                    patience=5)])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_19\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 512)               8704      \n",
      "_________________________________________________________________\n",
      "dropout_42 (Dropout)         (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_43 (Dropout)         (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "ukryta_2 (Dense)             (None, 64)                16448     \n",
      "_________________________________________________________________\n",
      "dropout_44 (Dropout)         (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 130       \n",
      "_________________________________________________________________\n",
      "softmax_14 (Softmax)         (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 156,610\n",
      "Trainable params: 156,610\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "99/99 [==============================] - 3s 12ms/step - loss: 0.6765 - auc: 0.9549 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 2/30\n",
      "99/99 [==============================] - 1s 8ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 3/30\n",
      "99/99 [==============================] - 1s 7ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 4/30\n",
      "99/99 [==============================] - 1s 6ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 5/30\n",
      "99/99 [==============================] - 1s 6ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 6/30\n",
      "99/99 [==============================] - 1s 7ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n"
     ]
    }
   ],
   "source": [
    "model_nn1 = model_nn(X_train_array, Y_train,[512,256,64],0.3,2,30)\n",
    "Y_pred_nn1 = model_nn1.predict(X_test_array).argmax(axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "31/31 [==============================] - 0s 3ms/step - loss: 0.7341 - auc: 0.9521\n",
      "loss: 0.7340565323829651\n",
      "auc: 0.9521384239196777\n"
     ]
    }
   ],
   "source": [
    "# szybka ewaluacja\n",
    "eval_res = model_nn1.evaluate(X_test_array, Y_test)\n",
    "eval_names = model_nn1.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn1 = confusion_matrix(Y_test.argmax(axis = 1), Y_pred_nn1)\n",
    "raport_nn1 = classification_report(Y_test.argmax(axis = 1), Y_pred_nn1)\n",
    "print(raport_nn1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model nie rozpoznaje w ogóle klasy 2, spróbujmy innych konfiguracji."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_22\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 256)               4352      \n",
      "_________________________________________________________________\n",
      "dropout_49 (Dropout)         (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_50 (Dropout)         (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 258       \n",
      "_________________________________________________________________\n",
      "softmax_17 (Softmax)         (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 37,506\n",
      "Trainable params: 37,506\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "99/99 [==============================] - 2s 7ms/step - loss: 0.7510 - auc: 0.9510 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 2/30\n",
      "99/99 [==============================] - 0s 4ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 3/30\n",
      "99/99 [==============================] - 0s 4ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 4/30\n",
      "99/99 [==============================] - 0s 4ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 5/30\n",
      "99/99 [==============================] - 0s 4ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "Epoch 6/30\n",
      "99/99 [==============================] - 0s 4ms/step - loss: 0.6299 - auc: 0.9589 - val_loss: 0.6439 - val_auc: 0.9580\n",
      "31/31 [==============================] - 0s 2ms/step - loss: 0.7341 - auc: 0.9521\n",
      "loss: 0.7340565323829651\n",
      "auc: 0.9521384239196777\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      1.00      0.98       935\n",
      "           1       0.00      0.00      0.00        47\n",
      "\n",
      "    accuracy                           0.95       982\n",
      "   macro avg       0.48      0.50      0.49       982\n",
      "weighted avg       0.91      0.95      0.93       982\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n",
      "C:\\Users\\Dawid\\anaconda3\\lib\\site-packages\\sklearn\\metrics\\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "model_nn2 = model_nn(X_train_array, Y_train,[256,128],0.5,2,30,0.1)\n",
    "Y_pred_nn2 = model_nn2.predict(X_test_array).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn2.evaluate(X_test_array, Y_test)\n",
    "eval_names = model_nn2.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn2 = confusion_matrix(Y_test.argmax(axis = 1), Y_pred_nn2)\n",
    "raport_nn2 = classification_report(Y_test.argmax(axis = 1), Y_pred_nn2)\n",
    "print(raport_nn2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 313,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Sprobujmy wykorzystac inne metode replikacje danych, dla poradzenia sobie z problemem niezbalansowania.\n",
    "\n",
    "smote = SMOTE(sampling_strategy = 'minority')\n",
    "x_sm, y_sm = smote.fit_resample(X,Y)\n",
    "\n",
    "\n",
    "# podzielmy jeszcze raz zbior na zbior treningowy i testowy\n",
    "X_train1, X_test1, Y_train1,Y_test1 = train_test_split(x_sm,y_sm, test_size = 0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 314,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Przetworzmy dane do formatu wejsciowego do modelu\n",
    "\n",
    "X_train_array_d = X_train1.values\n",
    "X_test_array_d = X_test1.values\n",
    "\n",
    "\n",
    "# transformujemy Y_test_duplicated i Y_test_duplicated\n",
    "\n",
    "Y_train_d = np.array(Y_train1)\n",
    "Y_test_d = np.array(Y_test1)\n",
    "\n",
    "Y_test_d = OneHotEncoder().fit_transform(Y_test_d.reshape(-1,1)).toarray()\n",
    "# zmienmy jeszcze format\n",
    "Y_test_d = Y_test_d.astype(np.float32)\n",
    "\n",
    "# transformujemy y_train\n",
    "Y_train_d = OneHotEncoder().fit_transform(Y_train_d.reshape(-1,1)).toarray()\n",
    "# zmienmy jeszcze format\n",
    "Y_train_d = Y_train_d.astype(np.float32)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 316,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_115\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 256)               4352      \n",
      "_________________________________________________________________\n",
      "dropout_236 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_131 (Bat (None, 256)               1024      \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_237 (Dropout)        (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_132 (Bat (None, 128)               512       \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 258       \n",
      "_________________________________________________________________\n",
      "softmax_110 (Softmax)        (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 39,042\n",
      "Trainable params: 38,274\n",
      "Non-trainable params: 768\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "188/188 [==============================] - 3s 6ms/step - loss: 0.7052 - auc: 0.7295 - val_loss: 0.5790 - val_auc: 0.8224\n",
      "Epoch 2/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.6581 - auc: 0.7610 - val_loss: 0.4520 - val_auc: 0.8688\n",
      "Epoch 3/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.5863 - auc: 0.7842 - val_loss: 0.4592 - val_auc: 0.8843\n",
      "Epoch 4/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.5967 - auc: 0.7755 - val_loss: 0.5035 - val_auc: 0.8436\n",
      "Epoch 5/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.6159 - auc: 0.7521 - val_loss: 0.5042 - val_auc: 0.8408\n",
      "Epoch 6/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.7346 - auc: 0.7333 - val_loss: 0.5034 - val_auc: 0.8234\n",
      "59/59 [==============================] - 1s 3ms/step - loss: 0.5070 - auc: 0.8212\n",
      "loss: 0.5069677233695984\n",
      "auc: 0.821212887763977\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.66      0.93      0.77       937\n",
      "           1       0.88      0.53      0.66       943\n",
      "\n",
      "    accuracy                           0.73      1880\n",
      "   macro avg       0.77      0.73      0.72      1880\n",
      "weighted avg       0.77      0.73      0.72      1880\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_nn3 = model_nn(X_train_array_d, Y_train_d,[256,128],0.5,2,30,0.1, True)\n",
    "Y_pred_nn3 = model_nn3.predict(X_test_array_d).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn3.evaluate(X_test_array_d, Y_test_d)\n",
    "eval_names = model_nn3.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn3 = confusion_matrix(Y_test_d.argmax(axis = 1), Y_pred_nn3)\n",
    "raport_nn3 = classification_report(Y_test_d.argmax(axis = 1), Y_pred_nn3)\n",
    "print(raport_nn3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Model po replikacji danych osiagnal znaczna poprawe, sprobujmy jeszcze kilku eksperymentów."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 317,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_116\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 512)               8704      \n",
      "_________________________________________________________________\n",
      "dropout_238 (Dropout)        (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_133 (Bat (None, 512)               2048      \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 1026      \n",
      "_________________________________________________________________\n",
      "softmax_111 (Softmax)        (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 11,778\n",
      "Trainable params: 10,754\n",
      "Non-trainable params: 1,024\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "188/188 [==============================] - 3s 6ms/step - loss: 0.5793 - auc: 0.8039 - val_loss: 0.4340 - val_auc: 0.8843\n",
      "Epoch 2/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.4643 - auc: 0.8627 - val_loss: 0.5613 - val_auc: 0.7667\n",
      "Epoch 3/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.4149 - auc: 0.8914 - val_loss: 0.8991 - val_auc: 0.7131\n",
      "Epoch 4/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.4051 - auc: 0.8977 - val_loss: 0.7781 - val_auc: 0.7182\n",
      "Epoch 5/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.4027 - auc: 0.8991 - val_loss: 0.6987 - val_auc: 0.7300\n",
      "Epoch 6/30\n",
      "188/188 [==============================] - 1s 4ms/step - loss: 0.4053 - auc: 0.8962 - val_loss: 0.4470 - val_auc: 0.8695\n",
      "59/59 [==============================] - 0s 3ms/step - loss: 0.4481 - auc: 0.8679\n",
      "loss: 0.44808194041252136\n",
      "auc: 0.8678510189056396\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.70      0.96      0.81       937\n",
      "           1       0.93      0.60      0.73       943\n",
      "\n",
      "    accuracy                           0.78      1880\n",
      "   macro avg       0.82      0.78      0.77      1880\n",
      "weighted avg       0.82      0.78      0.77      1880\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_nn4 = model_nn(X_train_array_d, Y_train_d,[512],0.5,2,30,0.001, True)\n",
    "Y_pred_nn4 = model_nn4.predict(X_test_array_d).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn4.evaluate(X_test_array_d, Y_test_d)\n",
    "eval_names = model_nn4.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn4 = confusion_matrix(Y_test_d.argmax(axis = 1), Y_pred_nn4)\n",
    "raport_nn4 = classification_report(Y_test_d.argmax(axis = 1), Y_pred_nn4)\n",
    "print(raport_nn4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 321,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_120\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 512)               8704      \n",
      "_________________________________________________________________\n",
      "dropout_243 (Dropout)        (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_138 (Bat (None, 512)               2048      \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 128)               65664     \n",
      "_________________________________________________________________\n",
      "dropout_244 (Dropout)        (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_139 (Bat (None, 128)               512       \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 258       \n",
      "_________________________________________________________________\n",
      "softmax_115 (Softmax)        (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 77,186\n",
      "Trainable params: 75,906\n",
      "Non-trainable params: 1,280\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "188/188 [==============================] - 3s 7ms/step - loss: 0.6373 - auc: 0.7772 - val_loss: 0.5732 - val_auc: 0.8212\n",
      "Epoch 2/30\n",
      "188/188 [==============================] - 1s 5ms/step - loss: 0.5864 - auc: 0.8074 - val_loss: 0.4780 - val_auc: 0.8597\n",
      "Epoch 3/30\n",
      "188/188 [==============================] - 1s 5ms/step - loss: 0.5444 - auc: 0.8233 - val_loss: 0.4571 - val_auc: 0.8704\n",
      "Epoch 4/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.5189 - auc: 0.8352 - val_loss: 0.4369 - val_auc: 0.8804\n",
      "Epoch 5/30\n",
      "188/188 [==============================] - 1s 5ms/step - loss: 0.5044 - auc: 0.8423 - val_loss: 0.4126 - val_auc: 0.8967\n",
      "Epoch 6/30\n",
      "188/188 [==============================] - 1s 5ms/step - loss: 0.4840 - auc: 0.8550 - val_loss: 0.3940 - val_auc: 0.9046\n",
      "59/59 [==============================] - 1s 3ms/step - loss: 0.4082 - auc: 0.8993\n",
      "loss: 0.4082390367984772\n",
      "auc: 0.8993103504180908\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.89      0.73      0.81       937\n",
      "           1       0.78      0.91      0.84       943\n",
      "\n",
      "    accuracy                           0.82      1880\n",
      "   macro avg       0.83      0.82      0.82      1880\n",
      "weighted avg       0.83      0.82      0.82      1880\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_nn5 = model_nn(X_train_array_d, Y_train_d,[512,128],0.2,2,30,0.0001, True)\n",
    "Y_pred_nn5 = model_nn5.predict(X_test_array_d).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn5.evaluate(X_test_array_d, Y_test_d)\n",
    "eval_names = model_nn5.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn5 = confusion_matrix(Y_test_d.argmax(axis = 1), Y_pred_nn5)\n",
    "raport_nn5 = classification_report(Y_test_d.argmax(axis = 1), Y_pred_nn5)\n",
    "print(raport_nn5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 331,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_129\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 512)               8704      \n",
      "_________________________________________________________________\n",
      "dropout_267 (Dropout)        (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_162 (Bat (None, 512)               2048      \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_268 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_163 (Bat (None, 256)               1024      \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 514       \n",
      "_________________________________________________________________\n",
      "softmax_124 (Softmax)        (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 143,618\n",
      "Trainable params: 142,082\n",
      "Non-trainable params: 1,536\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "188/188 [==============================] - 3s 8ms/step - loss: 0.6335 - auc: 0.7867 - val_loss: 0.5433 - val_auc: 0.8287\n",
      "Epoch 2/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.5407 - auc: 0.8292 - val_loss: 0.4705 - val_auc: 0.8718\n",
      "Epoch 3/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.5094 - auc: 0.8421 - val_loss: 0.4093 - val_auc: 0.8952\n",
      "Epoch 4/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.4774 - auc: 0.8588 - val_loss: 0.3689 - val_auc: 0.9165\n",
      "Epoch 5/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.4390 - auc: 0.8798 - val_loss: 0.3418 - val_auc: 0.9273\n",
      "Epoch 6/30\n",
      "188/188 [==============================] - 1s 6ms/step - loss: 0.4038 - auc: 0.8984 - val_loss: 0.3254 - val_auc: 0.9362\n",
      "59/59 [==============================] - 1s 4ms/step - loss: 0.3422 - auc: 0.9292\n",
      "loss: 0.34222978353500366\n",
      "auc: 0.9291889667510986\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.87      0.83      0.85       937\n",
      "           1       0.84      0.88      0.86       943\n",
      "\n",
      "    accuracy                           0.86      1880\n",
      "   macro avg       0.86      0.86      0.86      1880\n",
      "weighted avg       0.86      0.86      0.86      1880\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_nn6 = model_nn(X_train_array_d, Y_train_d,[512,256],0.1,2,30,0.0001, True)\n",
    "Y_pred_nn6 = model_nn6.predict(X_test_array_d).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn6.evaluate(X_test_array_d, Y_test_d)\n",
    "eval_names = model_nn6.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn6 = confusion_matrix(Y_test_d.argmax(axis = 1), Y_pred_nn6)\n",
    "raport_nn6 = classification_report(Y_test_d.argmax(axis = 1), Y_pred_nn6)\n",
    "print(raport_nn6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 332,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_130\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "ukryta_0 (Dense)             (None, 1024)              17408     \n",
      "_________________________________________________________________\n",
      "dropout_269 (Dropout)        (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_164 (Bat (None, 1024)              4096      \n",
      "_________________________________________________________________\n",
      "ukryta_1 (Dense)             (None, 256)               262400    \n",
      "_________________________________________________________________\n",
      "dropout_270 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "batch_normalization_165 (Bat (None, 256)               1024      \n",
      "_________________________________________________________________\n",
      "WYJSCIE (Dense)              (None, 2)                 514       \n",
      "_________________________________________________________________\n",
      "softmax_125 (Softmax)        (None, 2)                 0         \n",
      "=================================================================\n",
      "Total params: 285,442\n",
      "Trainable params: 282,882\n",
      "Non-trainable params: 2,560\n",
      "_________________________________________________________________\n",
      "Epoch 1/30\n",
      "188/188 [==============================] - 4s 11ms/step - loss: 0.5769 - auc: 0.8142 - val_loss: 0.5280 - val_auc: 0.8448\n",
      "Epoch 2/30\n",
      "188/188 [==============================] - 2s 10ms/step - loss: 0.4946 - auc: 0.8544 - val_loss: 0.4077 - val_auc: 0.8969\n",
      "Epoch 3/30\n",
      "188/188 [==============================] - 2s 9ms/step - loss: 0.4556 - auc: 0.8731 - val_loss: 0.3377 - val_auc: 0.9304\n",
      "Epoch 4/30\n",
      "188/188 [==============================] - 2s 10ms/step - loss: 0.4264 - auc: 0.8896 - val_loss: 0.3113 - val_auc: 0.9454\n",
      "Epoch 5/30\n",
      "188/188 [==============================] - 2s 10ms/step - loss: 0.4025 - auc: 0.9022 - val_loss: 0.3153 - val_auc: 0.9385\n",
      "Epoch 6/30\n",
      "188/188 [==============================] - 2s 10ms/step - loss: 0.3712 - auc: 0.9165 - val_loss: 0.2882 - val_auc: 0.9500\n",
      "59/59 [==============================] - 1s 5ms/step - loss: 0.2992 - auc: 0.9465\n",
      "loss: 0.29919472336769104\n",
      "auc: 0.9464707970619202\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.81      0.87       937\n",
      "           1       0.83      0.94      0.88       943\n",
      "\n",
      "    accuracy                           0.88      1880\n",
      "   macro avg       0.88      0.88      0.88      1880\n",
      "weighted avg       0.88      0.88      0.88      1880\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model_nn6 = model_nn(X_train_array_d, Y_train_d,[1024,256],0.1,2,30,0.0001, True)\n",
    "Y_pred_nn6 = model_nn6.predict(X_test_array_d).argmax(axis = 1)\n",
    "\n",
    "# szybka ewaluacja\n",
    "eval_res = model_nn6.evaluate(X_test_array_d, Y_test_d)\n",
    "eval_names = model_nn6.metrics_names\n",
    "print(f\"{eval_names[0]}: {eval_res[0]}\")\n",
    "print(f\"{eval_names[1]}: {eval_res[1]}\")\n",
    "\n",
    "#obejrzyjmy jeszcze macierz pomyłek \n",
    "matrix_nn6 = confusion_matrix(Y_test_d.argmax(axis = 1), Y_pred_nn6)\n",
    "raport_nn6 = classification_report(Y_test_d.argmax(axis = 1), Y_pred_nn6)\n",
    "print(raport_nn6)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Podsumowanie "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nasz model po zastosowanie zaawansowanej metody replikacji danych mniejszej klasy, poprzez interpolacje i dostrojenia parametrow modelu w sieci neuronowej uzyskał całkiem dobre wyniki na poziomie 88% dokladnosci\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}