Week1/Session2_solutions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "d0fff099-32b4-4d5f-94f2-dc957e8d7b58",
    "_uuid": "0a7336426503aa654b923d813483e106306716f9"
   },
   "source": [
    "# Week 1 - The What Why and when of Machine Learning\n",
    "Main idea is to follow that notebook in order to have an initial aproach -  this doesn't means that you only have to answer to the questions or implement code. Feel free to add all the interesting content that you find and take conclusions from the data. Please just use this as a template"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 0. Library import and list input files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "_cell_guid": "a4c3c51c-f465-4cfc-913a-ee1225e57724",
    "_uuid": "15592da27c5f06e6ab4fb0a293a851cc8489453b",
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import sys\n",
    "\n",
    "\n",
    "#Ignore Warnings - save some confusion\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "#Pandas more columns\n",
    "pd.options.display.max_columns = None\n",
    "pd.set_option('display.max_columns', None)\n",
    "\n",
    "# Add input as import path\n",
    "sys.path.insert(0,'../input')\n",
    "\n",
    "# Plot style\n",
    "plt.style.use('fivethirtyeight')\n",
    "\n",
    "# Import the data from the dataset\n",
    "train_data = pd.read_csv('../input/train.csv',index_col='id')\n",
    "test_data = pd.read_csv('../input/test.csv',index_col='id')\n",
    "\n",
    "trd = train_data # too much to write each time\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "d96be262-1ded-4cd3-9193-03d7dc0b86bb",
    "_uuid": "823783aab011b613db23022d071be8684faa60e0"
   },
   "source": [
    "# 1. Feature engineering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "b764727f-3455-4d7b-8eac-101b52644cf4",
    "_uuid": "028fec15cf0c5f625c122db5e726c9e6f1073905"
   },
   "source": [
    "### 1.1 Data engineering.\n",
    "Objectives:\n",
    "* Create three new features using the actual features.\n",
    "* Create plots like in the first session. (Categorical / Numerical)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "\n",
    "# Todo 1 : Create a new feature that describes if people travels alone or not.\n",
    "#dataset['TravelAlone'] = ...\n",
    "\n",
    "# Todo 2 : Create a new feature that represents the FamilySize - for each passenger a new column with a number 1..n\n",
    "#dataset['FamilySize'] = ...\n",
    "\n",
    "# Todo 3 : Create a new feature based in the Title name for example Sr, Mrs, Rev.\n",
    "# Clue 1: As we can see different classes of title name exist. Decide how to group and decide \n",
    "#train_data['name'].str.split(\",\",expand=True)[1].str.split(\".\",expand=True)[0].unique()\n",
    "# Your work\n",
    "#dataset['RareName'] = ....\n",
    "\n",
    "# Optional : Plot the data in plots like in the previous training notebook (Categorical / Numerical)\n",
    "\n",
    "\n",
    "#Objective 1 - create 3 new features\n",
    "#Feature #1  TravelAlone\n",
    "trd[\"TravelAlone\"] = trd[\"sibsp\"] == 0\n",
    "#display(trd[[\"sibsp\", \"TravelAlone\"]]) #To Check if all is well\n",
    "\n",
    "#Feature #2 - nr_cabins\n",
    "# Oh man - can I make this a lambda??\n",
    "def count_words(sample):\n",
    "    return len(str(sample).split())\n",
    "trd[\"nr_cabins\"] = trd[\"cabin\"].apply(count_words)\n",
    "\n",
    "\n",
    "#Feature #3 - Fare per cabin\n",
    "trd[\"fare_per_cabin\"] = trd[\"fare\"].astype(float)/trd[\"nr_cabins\"].astype(float)\n",
    "#display(trd[[\"fare\", \"fare_per_cabin\", \"nr_cabins\"]]) #All is well - just double checking"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "b764727f-3455-4d7b-8eac-101b52644cf4",
    "_uuid": "028fec15cf0c5f625c122db5e726c9e6f1073905"
   },
   "source": [
    "### 1.2 Data correlation.\n",
    "Objectives:\n",
    "* Create a heatmap that show the data correlation\n",
    "* Understand the data correlation\n",
    "\n",
    "Questions:\n",
    "* Q1: Which are the top three pairs of data with more correlation? (ToDo 1..2 and 3..4 ) \n",
    "* Q2: Is the sex of the passengers related with survival?\n",
    "* Q3: If two features have a lot of correlation we want to preserve both or only use one?\n",
    "* Q4: Are the previous new created features related with the survivance in the accident? Choose the one that you think that is the best.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "_cell_guid": "ac3cbd95-2242-4251-90ca-dd65c34e17b3",
    "_uuid": "6745659346d0b3d3a9b6424ce123307f20be533f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before Answering. Some basics -> Data Correlation for the features pclass/age\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd0AAAFnCAYAAADjbJN9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIYQQNfk//ZwyD6G+vYoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7faea00c42b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "A1 : Most Correlated pairs\n",
      "fare <--> fare_per_cabin\n",
      "TravelAlone <--> sibsp\n",
      "sibsp <--> TravelAlone\n",
      "pclass <--> fare\n",
      "fare_per_cabin <--> fare\n",
      "\n",
      "\n",
      "\n",
      "\n",
      "A2: SEX-Survived correlation ratio. =  0.524846293771\n",
      "It is correlated\n",
      "Not Heavily correlated\n",
      "\n",
      "\n",
      "\n",
      "\n",
      "A3. We want to preserve both.\n",
      "However, we also want to see how each of the features correlates with every other\n",
      "In the case as answered in A2, we want to keep 'fare' for sure as it appears in 3 top correlations\n",
      "\n",
      "\n",
      "\n",
      "\n",
      "A4. Using the heatmap of the new features to evaluate this\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd0AAAEOCAYAAADMl83vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fae9f9c6240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TravelAlone       0.115157\n",
      "nr_cabins         0.104661\n",
      "fare_per_cabin    0.260290\n",
      "Name: survived, dtype: float64\n",
      "Looking at the heatmap, I think fare_per_cabin has the most correlation with survived\n",
      "But! some already existing(not newly created) like age-survived features are better off\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# PreTodo. Copying train_data into another variable. Some examples\n",
    "sample_data = trd.copy()\n",
    "sample_data.drop(columns=[\"body\"], inplace=True) #Dropping target variable\n",
    "sex_types = {\"male\":1, \"female\":0}\n",
    "sample_data[\"sex\"] = sample_data[\"sex\"].apply(sex_types.get) # Converting into numerical label\n",
    "#Work with your train data , the same way we work with sample_data\n",
    "\n",
    "\n",
    "\n",
    "# Answer 1: Compute the correlation for all the features ( Remember that the survived column is our label )\n",
    "print(\"Before Answering. Some basics -> Data Correlation for the features pclass/age\")\n",
    "data_correlation = np.abs(sample_data.corr()) #Absolute data correlation - Only magnitudes are important\n",
    "# Plot the data in a fancy plot like the sns.heatmap\n",
    "sns.heatmap(data_correlation) \n",
    "plt.show()\n",
    "# I can look at the data_correlation map to find the max-correlated pairs or ..\n",
    "data_corr_arr = np.array(data_correlation)\n",
    "#forcing identity elements to -1. We want unique pairs, not age-age, survived-survived , etc\n",
    "for i in range(len(data_corr_arr)):\n",
    "    data_corr_arr[i,i] = -1\n",
    "\n",
    "num_largest = 5\n",
    "indices = (-data_corr_arr).argpartition(num_largest, axis=None)[:num_largest]\n",
    "x, y = np.unravel_index(indices, data_corr_arr.shape)\n",
    "print(\"A1 : Most Correlated pairs\")\n",
    "for ix,iy in zip(x,y):\n",
    "    print(data_correlation.columns[ix],\"<-->\",data_correlation.columns[iy])\n",
    "\n",
    "# ToDo 3: Compute the correlation between the sex feature and survived label \n",
    "print(\"\\n\\n\\n\")\n",
    "sex_survived_corr_ratio = data_correlation[\"sex\"][\"survived\"]\n",
    "print(\"A2: SEX-Survived correlation ratio. = \", sex_survived_corr_ratio)\n",
    "if sex_survived_corr_ratio > 0.5:\n",
    "    print(\"It is correlated\")\n",
    "if sex_survived_corr_ratio <0.8:\n",
    "    print(\"Not Heavily correlated\")\n",
    "\n",
    "print(\"\\n\\n\\n\")    \n",
    "print(\"A3. We want to preserve both.\")\n",
    "print(\"However, we also want to see how each of the features correlates with every other\")\n",
    "print(\"In the case as answered in A2, we want to keep 'fare' for sure as it appears in 3 top correlations\")\n",
    "\n",
    "print(\"\\n\\n\\n\")    \n",
    "print(\"A4. Using the heatmap of the new features to evaluate this\")\n",
    "new_data_corr =  data_correlation[[\"TravelAlone\",\"nr_cabins\",\"fare_per_cabin\"]]\n",
    "#print(new_data_corr.iloc(data_correlation.columns.index(\"survived\")))\n",
    "sns.heatmap(new_data_corr)\n",
    "idx_survived = list(data_correlation.columns).index(\"survived\")\n",
    "plt.show()\n",
    "print(np.abs(new_data_corr.iloc[idx_survived]))\n",
    "print(\"Looking at the heatmap, I think fare_per_cabin has the most correlation with survived\")\n",
    "print(\"But! some already existing(not newly created) like age-survived features are better off\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "_cell_guid": "b764727f-3455-4d7b-8eac-101b52644cf4",
    "_uuid": "028fec15cf0c5f625c122db5e726c9e6f1073905"
   },
   "source": [
    "# 2. Data visualization.\n",
    "Objectives:\n",
    "* Use PCA to reduce the dimensions of our data\n",
    "* Analyze the separability of our data\n",
    "\n",
    "Questions:\n",
    "* Q1: Is the data separable?\n",
    "* Q2: You think that we need to reduce the dimensionality of our data?\n",
    "* Q3: Find whitch infromation returns the explained_variance_ratio_ method\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "ename": "SyntaxError",
     "evalue": "invalid syntax (<ipython-input-105-1bf45be22fb6>, line 17)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;36m  File \u001b[0;32m\"<ipython-input-105-1bf45be22fb6>\"\u001b[0;36m, line \u001b[0;32m17\u001b[0m\n\u001b[0;31m    train_data_subset[\"embarked\"] = ....\u001b[0m\n\u001b[0m                                        ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.decomposition import PCA as sklearnPCA\n",
    "# \n",
    "\"\"\"\n",
    "First of all we need to scale our data we have different features measured in different \"units\" what we need is to normali\n",
    "this features to be in the same range, the easiest way to do this is with the MinMaxScaler\n",
    "\"\"\"\n",
    "\n",
    "# NOT ANSWERING THESE AS  MORE SUBJECTIVE THAN OBJECTIVE. WILL DISCUSS IN CLASS\n",
    "\n",
    "# 1. Select a subset of data\n",
    "#train_data_subset = train_data[[\"age\", \"sex\", \"pclass\", \"embarked\", \"TravelAlone\"]]\n",
    "train_data_subset = train_data[[\"age\", \"pclass\", \"sibsp\", \"embarked\", \"sex\"]]\n",
    "\n",
    "# Todo 1: Convert the categorical labels to numerical for example Sex can be converted to a boolean\n",
    "train_data_subset[\"sex\"] = ...\n",
    "train_data_subset[\"embarked\"] = ....\n",
    "\n",
    "# 2. Remove NA with 0 probably exist a better way :)\n",
    "train_data_subset.fillna(0, inplace=True)\n",
    "\n",
    "# 3. Scale the data\n",
    "scaler = MinMaxScaler()\n",
    "train_data_scaled = pd.DataFrame(scaler.fit_transform(train_data_subset), columns=train_data_subset.columns)\n",
    "\n",
    "#After\n",
    "print(\"Before scale:\")\n",
    "print(train_data_subset[0:3])\n",
    "print(\"\\n\")\n",
    "#Before\n",
    "print(\"After scale:\")\n",
    "print(train_data_scaled[0:3])\n",
    "print(\"\\n\")\n",
    "\n",
    "# 4. Compute PCA\n",
    "pca = sklearnPCA(n_components=2) #2-dimensional PCA\n",
    "transformed = pd.DataFrame(pca.fit_transform(train_data_scaled))\n",
    "train_data.index = range(len(train_data.index))\n",
    "\n",
    "# 5. Plot the data with the reduced dimentions\n",
    "plt.scatter(transformed[train_data[\"survived\"]==0][0], transformed[train_data[\"survived\"]==0][1], label='Died', c='red')\n",
    "plt.scatter(transformed[train_data[\"survived\"]==1][1], transformed[train_data[\"survived\"]==1][1], label='Survived', c='blue')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "# 6. Interesting methods\n",
    "print(pca.explained_variance_ratio_ )\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": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}