SLC - In-Class Lab Exercise_Day4_Ensemble.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VBvK9DHGNWPc"
   },
   "source": [
    "<table align=\"center\" width=100%>\n",
    "    <tr>\n",
    "        <td width=\"15%\">\n",
    "            <img src=\"in_class.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"center\">\n",
    "                <font color=\"#21618C\" size=8px>\n",
    "                    <b> Inclass - Lab <br>(Day 4)\n",
    "                    </b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "g5ZbqmF-A-z7"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vh797P5UNWPg"
   },
   "source": [
    "### About the data set (Employee data)\n",
    "\n",
    "The dataset contains information about employees. The aim is to find which employees might undergo attrition.<br> Attribute information: \n",
    "\n",
    "**Age:** Age of the employee \n",
    "\n",
    "**BusinessTravel:** How much travel is involved in the job for the employee:No Travel, Travel Frequently, Tavel Rarely\n",
    "\n",
    "**Department:** Department of the employee: Human Resources, Reserach & Development, Sales\n",
    "\n",
    "**Commute:** Number of miles of daily commute for the employee\n",
    "\n",
    "**Education:** Employee education field: Human Resources, Life Sciences, Marketing, Medical Sciences, Technical, Others\n",
    "\n",
    "**EnvironmentSatisfaction:** Satisfaction of employee with office environment\n",
    "\n",
    "**Gender:** Employee gender\n",
    "\n",
    "**JobInvolvement:** Job involvement rating\n",
    "\n",
    "**JobLevel:**  Job level for employee designation\n",
    "\n",
    "**JobSatisfaction:** Employee job satisfaction rating\n",
    "\n",
    "**MonthlyIncome:** Employee monthly salary\n",
    "\n",
    "**OverTime:** Has the employee been open to working overtime: Yes or No\n",
    "\n",
    "**PercentSalaryHike:** Percent increase in salary\n",
    "\n",
    "**PerformanceRating:** Overall employee performance rating\n",
    "\n",
    "**YearsAtCompany:** Number of years the employee has worked with the company\n",
    "\n",
    "**Attrition:** Employee leaving the company: Yes or No"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Sno8wi3mNWPh"
   },
   "source": [
    "## Table of Content\n",
    "\n",
    "1. **[ADABoost](#ada)**\n",
    "2. **[Gradient Boosting](#gb)**\n",
    "3. **[Extreme Gradient Boosting (XGBoost)](#xgb)**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mMb9RQT1NWPh"
   },
   "source": [
    "**Import the required libraries**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "executionInfo": {
     "elapsed": 1512,
     "status": "ok",
     "timestamp": 1637212436242,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "47ovCMZRNWPi"
   },
   "outputs": [],
   "source": [
    "# import 'Pandas' \n",
    "import pandas as pd \n",
    "\n",
    "# import 'Numpy' \n",
    "import numpy as np\n",
    "\n",
    "# import subpackage of Matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.colors import ListedColormap\n",
    "\n",
    "# import 'Seaborn' \n",
    "import seaborn as sns\n",
    "\n",
    "# to suppress warnings \n",
    "from warnings import filterwarnings\n",
    "filterwarnings('ignore')\n",
    "\n",
    "# display all columns of the dataframe\n",
    "pd.options.display.max_columns = None\n",
    "\n",
    "# display all rows of the dataframe\n",
    "pd.options.display.max_rows = None\n",
    " \n",
    "# to display the float values upto 6 decimal places     \n",
    "pd.options.display.float_format = '{:.6f}'.format\n",
    "\n",
    "# import train-test split \n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# import StandardScaler to perform scaling\n",
    "from sklearn.preprocessing import StandardScaler \n",
    "\n",
    "# resample the dataset\n",
    "from sklearn.utils import resample\n",
    "from sklearn.utils import shuffle\n",
    "\n",
    "# import various functions from sklearn \n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import roc_curve\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.ensemble import StackingClassifier,VotingClassifier\n",
    "\n",
    "\n",
    "# import the XGBoost function for classification\n",
    "from xgboost import XGBClassifier\n",
    "\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "executionInfo": {
     "elapsed": 380,
     "status": "ok",
     "timestamp": 1637212529121,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "VfgA7_skNWPk"
   },
   "outputs": [],
   "source": [
    "# set the plot size using 'rcParams'\n",
    "# once the plot size is set using 'rcParams', it sets the size of all the forthcoming plots in the file\n",
    "# pass width and height in inches to 'figure.figsize' \n",
    "plt.rcParams['figure.figsize'] = [15,8]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 53275,
     "status": "ok",
     "timestamp": 1637212606959,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "d-FaLdZ9NjY-",
    "outputId": "61a2cbb9-4938-41da-e04c-2b35f3ed6729"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mounted at /content/drive\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XC-hLl7rNWPk"
   },
   "source": [
    "#### Load the csv file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 313
    },
    "executionInfo": {
     "elapsed": 933,
     "status": "ok",
     "timestamp": 1637212609439,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "6Y0q8XviNWPl",
    "outputId": "8088bdcf-d352-4b68-cc96-d5269d1e176e"
   },
   "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>Attrition</th>\n",
       "      <th>BusinessTravel</th>\n",
       "      <th>Department</th>\n",
       "      <th>DistanceFromHome</th>\n",
       "      <th>EducationField</th>\n",
       "      <th>EnvironmentSatisfaction</th>\n",
       "      <th>Gender</th>\n",
       "      <th>JobInvolvement</th>\n",
       "      <th>JobLevel</th>\n",
       "      <th>JobRole</th>\n",
       "      <th>JobSatisfaction</th>\n",
       "      <th>MonthlyIncome</th>\n",
       "      <th>NumCompaniesWorked</th>\n",
       "      <th>OverTime</th>\n",
       "      <th>PercentSalaryHike</th>\n",
       "      <th>PerformanceRating</th>\n",
       "      <th>YearsAtCompany</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>33</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Travel_Frequently</td>\n",
       "      <td>Research &amp; Development</td>\n",
       "      <td>3</td>\n",
       "      <td>Life Sciences</td>\n",
       "      <td>1</td>\n",
       "      <td>Male</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Research Scientist</td>\n",
       "      <td>1</td>\n",
       "      <td>3348</td>\n",
       "      <td>1</td>\n",
       "      <td>Yes</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>32</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Travel_Rarely</td>\n",
       "      <td>Sales</td>\n",
       "      <td>4</td>\n",
       "      <td>Medical</td>\n",
       "      <td>4</td>\n",
       "      <td>Male</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>Sales Executive</td>\n",
       "      <td>4</td>\n",
       "      <td>10400</td>\n",
       "      <td>1</td>\n",
       "      <td>No</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>40</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Travel_Rarely</td>\n",
       "      <td>Research &amp; Development</td>\n",
       "      <td>9</td>\n",
       "      <td>Life Sciences</td>\n",
       "      <td>4</td>\n",
       "      <td>Male</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>Laboratory Technician</td>\n",
       "      <td>1</td>\n",
       "      <td>2018</td>\n",
       "      <td>3</td>\n",
       "      <td>No</td>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>42</td>\n",
       "      <td>No</td>\n",
       "      <td>Travel_Rarely</td>\n",
       "      <td>Research &amp; Development</td>\n",
       "      <td>7</td>\n",
       "      <td>Medical</td>\n",
       "      <td>2</td>\n",
       "      <td>Female</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>Research Scientist</td>\n",
       "      <td>2</td>\n",
       "      <td>2372</td>\n",
       "      <td>6</td>\n",
       "      <td>Yes</td>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>43</td>\n",
       "      <td>No</td>\n",
       "      <td>Travel_Frequently</td>\n",
       "      <td>Research &amp; Development</td>\n",
       "      <td>27</td>\n",
       "      <td>Life Sciences</td>\n",
       "      <td>3</td>\n",
       "      <td>Female</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>Manufacturing Director</td>\n",
       "      <td>1</td>\n",
       "      <td>10820</td>\n",
       "      <td>8</td>\n",
       "      <td>No</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Age Attrition     BusinessTravel              Department  DistanceFromHome  \\\n",
       "0   33       Yes  Travel_Frequently  Research & Development                 3   \n",
       "1   32       Yes      Travel_Rarely                   Sales                 4   \n",
       "2   40       Yes      Travel_Rarely  Research & Development                 9   \n",
       "3   42        No      Travel_Rarely  Research & Development                 7   \n",
       "4   43        No  Travel_Frequently  Research & Development                27   \n",
       "\n",
       "  EducationField  EnvironmentSatisfaction  Gender  JobInvolvement  JobLevel  \\\n",
       "0  Life Sciences                        1    Male               3         1   \n",
       "1        Medical                        4    Male               1         3   \n",
       "2  Life Sciences                        4    Male               3         1   \n",
       "3        Medical                        2  Female               4         2   \n",
       "4  Life Sciences                        3  Female               3         3   \n",
       "\n",
       "                  JobRole  JobSatisfaction  MonthlyIncome  NumCompaniesWorked  \\\n",
       "0      Research Scientist                1           3348                   1   \n",
       "1         Sales Executive                4          10400                   1   \n",
       "2   Laboratory Technician                1           2018                   3   \n",
       "3      Research Scientist                2           2372                   6   \n",
       "4  Manufacturing Director                1          10820                   8   \n",
       "\n",
       "  OverTime  PercentSalaryHike  PerformanceRating  YearsAtCompany  \n",
       "0      Yes                 11                  3              10  \n",
       "1       No                 11                  3              14  \n",
       "2       No                 14                  3               5  \n",
       "3      Yes                 16                  3               1  \n",
       "4       No                 11                  3               8  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load the csv file\n",
    "df_employee = pd.read_csv('/content/drive/My Drive/SLC/emp_attrition.csv')\n",
    "\n",
    "# display the first five observations\n",
    "df_employee.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 311,
     "status": "ok",
     "timestamp": 1634803306095,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "PY2VVAeoNWPm",
    "outputId": "794588fe-adfd-4491-bf4a-da584c6d6295"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Age                         int64\n",
       "Attrition                  object\n",
       "BusinessTravel             object\n",
       "Department                 object\n",
       "DistanceFromHome            int64\n",
       "EducationField             object\n",
       "EnvironmentSatisfaction     int64\n",
       "Gender                     object\n",
       "JobInvolvement              int64\n",
       "JobLevel                    int64\n",
       "JobRole                    object\n",
       "JobSatisfaction             int64\n",
       "MonthlyIncome               int64\n",
       "NumCompaniesWorked          int64\n",
       "OverTime                   object\n",
       "PercentSalaryHike           int64\n",
       "PerformanceRating           int64\n",
       "YearsAtCompany              int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check data types of variables\n",
    "df_employee.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_oPKSQEfNWPn"
   },
   "source": [
    "**Data Preparation**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "b3Ysll5dNWPn"
   },
   "source": [
    "**Remove the outlier using IQR method, if present.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 355,
     "status": "ok",
     "timestamp": 1637212624747,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "pD6JSh6cNWPo",
    "outputId": "a41953ad-2667-4b15-b046-d6a525fef663"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1580, 18)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the size of data\n",
    "df_employee.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 4,
     "status": "ok",
     "timestamp": 1637212626141,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "nyAYFeMfNWPo",
    "outputId": "7738e2c4-e7b7-44c3-c4c2-5808afad4c48"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1487, 18)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# calculate the first quartile\n",
    "Q1 = df_employee.quantile(0.25)\n",
    "\n",
    "#calculate the third quartile\n",
    "Q3 = df_employee.quantile(0.75)\n",
    "\n",
    "# The Interquartile Range (IQR) is defined as the difference between the third and first quartile\n",
    "# calculate IQR\n",
    "IQR = Q3 - Q1\n",
    "\n",
    "# remove the outliers from the dataframe\n",
    "df_employee = df_employee[~((df_employee < (Q1 - 1.5 * IQR)) | (df_employee > (Q3 + 1.5 * IQR))).any(axis=1)]\n",
    "\n",
    "# check the shape of the data\n",
    "df_employee.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "executionInfo": {
     "elapsed": 338,
     "status": "ok",
     "timestamp": 1637212628898,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "AQi7tACrNWPp"
   },
   "outputs": [],
   "source": [
    "# reset the row index\n",
    "df_employee = df_employee.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vCbxPmaRNWPp"
   },
   "source": [
    "**Encode the categorical variables.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EdKox4SWNWPp"
   },
   "source": [
    "#### Split the dependent and independent variables."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "executionInfo": {
     "elapsed": 381,
     "status": "ok",
     "timestamp": 1637212631139,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "LHZNAQN-NWPp"
   },
   "outputs": [],
   "source": [
    "# store the target variable \"Attrition\" in a dataframe 'df_target'\n",
    "df_target = df_employee['Attrition']\n",
    "\n",
    "# store all the independent variables in a dataframe 'df_feature' \n",
    "# drop the column \"Attrition\" using drop()\n",
    "# 'axis = 1' drops the specified column\n",
    "df_feature = df_employee.drop('Attrition', axis = 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "executionInfo": {
     "elapsed": 379,
     "status": "ok",
     "timestamp": 1637212633211,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "zFPqqhfcNWPq",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# if the value in the target variable is 'Yes' then replace it with 1 else with 0\n",
    "for i in range(len(df_target)):\n",
    "    if df_target[i] == 'Yes':\n",
    "        df_target[i] = 1\n",
    "    else:\n",
    "        df_target[i] = 0\n",
    "        \n",
    "# change the datatype of the target variable to integer\n",
    "df_target = df_target.astype('int')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4fW1AFNWNWPq"
   },
   "source": [
    "#### Filter numerical and categorical variables."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 344,
     "status": "ok",
     "timestamp": 1637212635997,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "gcKY2VSTNWPq",
    "outputId": "a37838d8-15e2-44c3-fdad-d4490bb04649"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['Age', 'DistanceFromHome', 'EnvironmentSatisfaction', 'JobInvolvement',\n",
       "       'JobLevel', 'JobSatisfaction', 'MonthlyIncome', 'NumCompaniesWorked',\n",
       "       'PercentSalaryHike', 'PerformanceRating', 'YearsAtCompany'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# filter the numerical features in the dataset\n",
    "# 'select_dtypes' is used to select the variables with given data type\n",
    "# 'include = [np.number]' will include all the numerical variables\n",
    "df_num = df_feature.select_dtypes(include = [np.number])\n",
    "\n",
    "# display numerical features\n",
    "df_num.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 343,
     "status": "ok",
     "timestamp": 1637212638462,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "O3lWmgl9NWPq",
    "outputId": "8c69ad08-5328-4bd3-f40c-b1301ef83597"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['BusinessTravel', 'Department', 'EducationField', 'Gender', 'JobRole',\n",
       "       'OverTime'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# filter the categorical features in the dataset\n",
    "# 'select_dtypes' is used to select the variables with given data type\n",
    "# 'include = [np.object]' will include all the categorical variables\n",
    "df_cat = df_feature.select_dtypes(include = [np.object])\n",
    "\n",
    "# display categorical features\n",
    "df_cat.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HDURvuBpNWPr"
   },
   "source": [
    "**Encode the each categorical variable and create (n-1) dummy variables for n categories of the variable.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "executionInfo": {
     "elapsed": 6,
     "status": "ok",
     "timestamp": 1637212641663,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "7gWbfpn0NWPr"
   },
   "outputs": [],
   "source": [
    "# use 'get_dummies' from pandas to create dummy variables\n",
    "# use 'drop_first' to create (n-1) dummy variables\n",
    "dummy_var = pd.get_dummies(data = df_cat, drop_first = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jORW6-qMNWPr"
   },
   "source": [
    "#### Concatenate numerical and dummy encoded categorical variables."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 243
    },
    "executionInfo": {
     "elapsed": 400,
     "status": "ok",
     "timestamp": 1637212644283,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "uEAhNWgwNWPr",
    "outputId": "75e203fa-1ebc-408d-ab84-1d9e4754ade3"
   },
   "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>DistanceFromHome</th>\n",
       "      <th>EnvironmentSatisfaction</th>\n",
       "      <th>JobInvolvement</th>\n",
       "      <th>JobLevel</th>\n",
       "      <th>JobSatisfaction</th>\n",
       "      <th>MonthlyIncome</th>\n",
       "      <th>NumCompaniesWorked</th>\n",
       "      <th>PercentSalaryHike</th>\n",
       "      <th>PerformanceRating</th>\n",
       "      <th>YearsAtCompany</th>\n",
       "      <th>BusinessTravel_Travel_Frequently</th>\n",
       "      <th>BusinessTravel_Travel_Rarely</th>\n",
       "      <th>Department_Research &amp; Development</th>\n",
       "      <th>Department_Sales</th>\n",
       "      <th>EducationField_Life Sciences</th>\n",
       "      <th>EducationField_Marketing</th>\n",
       "      <th>EducationField_Medical</th>\n",
       "      <th>EducationField_Other</th>\n",
       "      <th>EducationField_Technical Degree</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>JobRole_Human Resources</th>\n",
       "      <th>JobRole_Laboratory Technician</th>\n",
       "      <th>JobRole_Manager</th>\n",
       "      <th>JobRole_Manufacturing Director</th>\n",
       "      <th>JobRole_Research Director</th>\n",
       "      <th>JobRole_Research Scientist</th>\n",
       "      <th>JobRole_Sales Executive</th>\n",
       "      <th>JobRole_Sales Representative</th>\n",
       "      <th>OverTime_Yes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>33</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3348</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>32</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>10400</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>40</td>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2018</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>42</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2372</td>\n",
       "      <td>6</td>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>43</td>\n",
       "      <td>27</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>10820</td>\n",
       "      <td>8</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Age  DistanceFromHome  EnvironmentSatisfaction  JobInvolvement  JobLevel  \\\n",
       "0   33                 3                        1               3         1   \n",
       "1   32                 4                        4               1         3   \n",
       "2   40                 9                        4               3         1   \n",
       "3   42                 7                        2               4         2   \n",
       "4   43                27                        3               3         3   \n",
       "\n",
       "   JobSatisfaction  MonthlyIncome  NumCompaniesWorked  PercentSalaryHike  \\\n",
       "0                1           3348                   1                 11   \n",
       "1                4          10400                   1                 11   \n",
       "2                1           2018                   3                 14   \n",
       "3                2           2372                   6                 16   \n",
       "4                1          10820                   8                 11   \n",
       "\n",
       "   PerformanceRating  YearsAtCompany  BusinessTravel_Travel_Frequently  \\\n",
       "0                  3              10                                 1   \n",
       "1                  3              14                                 0   \n",
       "2                  3               5                                 0   \n",
       "3                  3               1                                 0   \n",
       "4                  3               8                                 1   \n",
       "\n",
       "   BusinessTravel_Travel_Rarely  Department_Research & Development  \\\n",
       "0                             0                                  1   \n",
       "1                             1                                  0   \n",
       "2                             1                                  1   \n",
       "3                             1                                  1   \n",
       "4                             0                                  1   \n",
       "\n",
       "   Department_Sales  EducationField_Life Sciences  EducationField_Marketing  \\\n",
       "0                 0                             1                         0   \n",
       "1                 1                             0                         0   \n",
       "2                 0                             1                         0   \n",
       "3                 0                             0                         0   \n",
       "4                 0                             1                         0   \n",
       "\n",
       "   EducationField_Medical  EducationField_Other  \\\n",
       "0                       0                     0   \n",
       "1                       1                     0   \n",
       "2                       0                     0   \n",
       "3                       1                     0   \n",
       "4                       0                     0   \n",
       "\n",
       "   EducationField_Technical Degree  Gender_Male  JobRole_Human Resources  \\\n",
       "0                                0            1                        0   \n",
       "1                                0            1                        0   \n",
       "2                                0            1                        0   \n",
       "3                                0            0                        0   \n",
       "4                                0            0                        0   \n",
       "\n",
       "   JobRole_Laboratory Technician  JobRole_Manager  \\\n",
       "0                              0                0   \n",
       "1                              0                0   \n",
       "2                              1                0   \n",
       "3                              0                0   \n",
       "4                              0                0   \n",
       "\n",
       "   JobRole_Manufacturing Director  JobRole_Research Director  \\\n",
       "0                               0                          0   \n",
       "1                               0                          0   \n",
       "2                               0                          0   \n",
       "3                               0                          0   \n",
       "4                               1                          0   \n",
       "\n",
       "   JobRole_Research Scientist  JobRole_Sales Executive  \\\n",
       "0                           1                        0   \n",
       "1                           0                        1   \n",
       "2                           0                        0   \n",
       "3                           1                        0   \n",
       "4                           0                        0   \n",
       "\n",
       "   JobRole_Sales Representative  OverTime_Yes  \n",
       "0                             0             1  \n",
       "1                             0             0  \n",
       "2                             0             0  \n",
       "3                             0             1  \n",
       "4                             0             0  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# concat the dummy variables with numeric features to create a dataframe of all independent variables\n",
    "# 'axis=1' concats the dataframes along columns \n",
    "X = pd.concat([df_num, dummy_var], axis = 1)\n",
    "\n",
    "# display first five observations\n",
    "X.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "c1QiteVxNWPs"
   },
   "source": [
    "### Let's begin with some hands-on practice exercises"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "5y81qIfKNWPs"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>1. Convert the Age variable into categorical by dividing it in the ranges; 18-30, 31-40, 41-50, 51-60.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 330
    },
    "executionInfo": {
     "elapsed": 349,
     "status": "ok",
     "timestamp": 1637212723964,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "wcOwvRTWNWPs",
    "outputId": "7d191412-bbbf-44e0-a4c6-eb5469ec6958"
   },
   "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>DistanceFromHome</th>\n",
       "      <th>EnvironmentSatisfaction</th>\n",
       "      <th>JobInvolvement</th>\n",
       "      <th>JobLevel</th>\n",
       "      <th>JobSatisfaction</th>\n",
       "      <th>MonthlyIncome</th>\n",
       "      <th>NumCompaniesWorked</th>\n",
       "      <th>PercentSalaryHike</th>\n",
       "      <th>PerformanceRating</th>\n",
       "      <th>YearsAtCompany</th>\n",
       "      <th>BusinessTravel_Travel_Frequently</th>\n",
       "      <th>BusinessTravel_Travel_Rarely</th>\n",
       "      <th>Department_Research &amp; Development</th>\n",
       "      <th>Department_Sales</th>\n",
       "      <th>EducationField_Life Sciences</th>\n",
       "      <th>EducationField_Marketing</th>\n",
       "      <th>EducationField_Medical</th>\n",
       "      <th>EducationField_Other</th>\n",
       "      <th>EducationField_Technical Degree</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>JobRole_Human Resources</th>\n",
       "      <th>JobRole_Laboratory Technician</th>\n",
       "      <th>JobRole_Manager</th>\n",
       "      <th>JobRole_Manufacturing Director</th>\n",
       "      <th>JobRole_Research Director</th>\n",
       "      <th>JobRole_Research Scientist</th>\n",
       "      <th>JobRole_Sales Executive</th>\n",
       "      <th>JobRole_Sales Representative</th>\n",
       "      <th>OverTime_Yes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>A31-40</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3348</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>A31-40</td>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>10400</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>A31-40</td>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2018</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>A41-50</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2372</td>\n",
       "      <td>6</td>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>A41-50</td>\n",
       "      <td>27</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>10820</td>\n",
       "      <td>8</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      Age  DistanceFromHome  EnvironmentSatisfaction  JobInvolvement  \\\n",
       "0  A31-40                 3                        1               3   \n",
       "1  A31-40                 4                        4               1   \n",
       "2  A31-40                 9                        4               3   \n",
       "3  A41-50                 7                        2               4   \n",
       "4  A41-50                27                        3               3   \n",
       "\n",
       "   JobLevel  JobSatisfaction  MonthlyIncome  NumCompaniesWorked  \\\n",
       "0         1                1           3348                   1   \n",
       "1         3                4          10400                   1   \n",
       "2         1                1           2018                   3   \n",
       "3         2                2           2372                   6   \n",
       "4         3                1          10820                   8   \n",
       "\n",
       "   PercentSalaryHike  PerformanceRating  YearsAtCompany  \\\n",
       "0                 11                  3              10   \n",
       "1                 11                  3              14   \n",
       "2                 14                  3               5   \n",
       "3                 16                  3               1   \n",
       "4                 11                  3               8   \n",
       "\n",
       "   BusinessTravel_Travel_Frequently  BusinessTravel_Travel_Rarely  \\\n",
       "0                                 1                             0   \n",
       "1                                 0                             1   \n",
       "2                                 0                             1   \n",
       "3                                 0                             1   \n",
       "4                                 1                             0   \n",
       "\n",
       "   Department_Research & Development  Department_Sales  \\\n",
       "0                                  1                 0   \n",
       "1                                  0                 1   \n",
       "2                                  1                 0   \n",
       "3                                  1                 0   \n",
       "4                                  1                 0   \n",
       "\n",
       "   EducationField_Life Sciences  EducationField_Marketing  \\\n",
       "0                             1                         0   \n",
       "1                             0                         0   \n",
       "2                             1                         0   \n",
       "3                             0                         0   \n",
       "4                             1                         0   \n",
       "\n",
       "   EducationField_Medical  EducationField_Other  \\\n",
       "0                       0                     0   \n",
       "1                       1                     0   \n",
       "2                       0                     0   \n",
       "3                       1                     0   \n",
       "4                       0                     0   \n",
       "\n",
       "   EducationField_Technical Degree  Gender_Male  JobRole_Human Resources  \\\n",
       "0                                0            1                        0   \n",
       "1                                0            1                        0   \n",
       "2                                0            1                        0   \n",
       "3                                0            0                        0   \n",
       "4                                0            0                        0   \n",
       "\n",
       "   JobRole_Laboratory Technician  JobRole_Manager  \\\n",
       "0                              0                0   \n",
       "1                              0                0   \n",
       "2                              1                0   \n",
       "3                              0                0   \n",
       "4                              0                0   \n",
       "\n",
       "   JobRole_Manufacturing Director  JobRole_Research Director  \\\n",
       "0                               0                          0   \n",
       "1                               0                          0   \n",
       "2                               0                          0   \n",
       "3                               0                          0   \n",
       "4                               1                          0   \n",
       "\n",
       "   JobRole_Research Scientist  JobRole_Sales Executive  \\\n",
       "0                           1                        0   \n",
       "1                           0                        1   \n",
       "2                           0                        0   \n",
       "3                           1                        0   \n",
       "4                           0                        0   \n",
       "\n",
       "   JobRole_Sales Representative  OverTime_Yes  \n",
       "0                             0             1  \n",
       "1                             0             0  \n",
       "2                             0             0  \n",
       "3                             0             1  \n",
       "4                             0             0  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# iterating a for loop \n",
    "# categorizing the variable 'Age'\n",
    "for i in range(X.shape[0]):\n",
    "    if X.Age[i]>=18 and X.Age[i]<=30:\n",
    "        X.Age[i] = \"A18-30\"\n",
    "    elif(X.Age[i]>=31 and X.Age[i]<=40):\n",
    "        X.Age[i] = \"A31-40\"\n",
    "    elif(X.Age[i]>=41 and X.Age[i]<=50):\n",
    "        X.Age[i] = \"A41-50\"\n",
    "    else:\n",
    "        X.Age[i] = \"A51-60\"\n",
    "    \n",
    "X.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "6V77EG9-NWPs"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>2. Employee from which age category has the highest salary?.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 388,
     "status": "ok",
     "timestamp": 1637212793549,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "8yJ8g3FrNWPs",
    "outputId": "f57ee966-c3c0-4e8e-c007-6189a08715e6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Age\n",
       "A18-30    11935\n",
       "A31-40    12061\n",
       "A41-50    11957\n",
       "A51-60    10976\n",
       "Name: MonthlyIncome, dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# group the dataframe by 'Age'\n",
    "grp = X.groupby('Age')\n",
    "\n",
    "# find out the maximum value from the column MonthlyIncome for each age group\n",
    "grp.MonthlyIncome.max()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "clM_r1PPNWPt"
   },
   "source": [
    "**Interpretation:** An employee between the age category 31 to 40 has the highest income i.e. 12061."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "i9yr0gx3NWPt"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>3. Dummy encode the categorical variable \"Age\".</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 243
    },
    "executionInfo": {
     "elapsed": 345,
     "status": "ok",
     "timestamp": 1637212921461,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "T_el__76NWPt",
    "outputId": "1a66212f-ac48-4bbc-d9b8-117366e47e36"
   },
   "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>DistanceFromHome</th>\n",
       "      <th>EnvironmentSatisfaction</th>\n",
       "      <th>JobInvolvement</th>\n",
       "      <th>JobLevel</th>\n",
       "      <th>JobSatisfaction</th>\n",
       "      <th>MonthlyIncome</th>\n",
       "      <th>NumCompaniesWorked</th>\n",
       "      <th>PercentSalaryHike</th>\n",
       "      <th>PerformanceRating</th>\n",
       "      <th>YearsAtCompany</th>\n",
       "      <th>BusinessTravel_Travel_Frequently</th>\n",
       "      <th>BusinessTravel_Travel_Rarely</th>\n",
       "      <th>Department_Research &amp; Development</th>\n",
       "      <th>Department_Sales</th>\n",
       "      <th>EducationField_Life Sciences</th>\n",
       "      <th>EducationField_Marketing</th>\n",
       "      <th>EducationField_Medical</th>\n",
       "      <th>EducationField_Other</th>\n",
       "      <th>EducationField_Technical Degree</th>\n",
       "      <th>Gender_Male</th>\n",
       "      <th>JobRole_Human Resources</th>\n",
       "      <th>JobRole_Laboratory Technician</th>\n",
       "      <th>JobRole_Manager</th>\n",
       "      <th>JobRole_Manufacturing Director</th>\n",
       "      <th>JobRole_Research Director</th>\n",
       "      <th>JobRole_Research Scientist</th>\n",
       "      <th>JobRole_Sales Executive</th>\n",
       "      <th>JobRole_Sales Representative</th>\n",
       "      <th>OverTime_Yes</th>\n",
       "      <th>Age_A31-40</th>\n",
       "      <th>Age_A41-50</th>\n",
       "      <th>Age_A51-60</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3348</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>10400</td>\n",
       "      <td>1</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>9</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2018</td>\n",
       "      <td>3</td>\n",
       "      <td>14</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2372</td>\n",
       "      <td>6</td>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>27</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>10820</td>\n",
       "      <td>8</td>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   DistanceFromHome  EnvironmentSatisfaction  JobInvolvement  JobLevel  \\\n",
       "0                 3                        1               3         1   \n",
       "1                 4                        4               1         3   \n",
       "2                 9                        4               3         1   \n",
       "3                 7                        2               4         2   \n",
       "4                27                        3               3         3   \n",
       "\n",
       "   JobSatisfaction  MonthlyIncome  NumCompaniesWorked  PercentSalaryHike  \\\n",
       "0                1           3348                   1                 11   \n",
       "1                4          10400                   1                 11   \n",
       "2                1           2018                   3                 14   \n",
       "3                2           2372                   6                 16   \n",
       "4                1          10820                   8                 11   \n",
       "\n",
       "   PerformanceRating  YearsAtCompany  BusinessTravel_Travel_Frequently  \\\n",
       "0                  3              10                                 1   \n",
       "1                  3              14                                 0   \n",
       "2                  3               5                                 0   \n",
       "3                  3               1                                 0   \n",
       "4                  3               8                                 1   \n",
       "\n",
       "   BusinessTravel_Travel_Rarely  Department_Research & Development  \\\n",
       "0                             0                                  1   \n",
       "1                             1                                  0   \n",
       "2                             1                                  1   \n",
       "3                             1                                  1   \n",
       "4                             0                                  1   \n",
       "\n",
       "   Department_Sales  EducationField_Life Sciences  EducationField_Marketing  \\\n",
       "0                 0                             1                         0   \n",
       "1                 1                             0                         0   \n",
       "2                 0                             1                         0   \n",
       "3                 0                             0                         0   \n",
       "4                 0                             1                         0   \n",
       "\n",
       "   EducationField_Medical  EducationField_Other  \\\n",
       "0                       0                     0   \n",
       "1                       1                     0   \n",
       "2                       0                     0   \n",
       "3                       1                     0   \n",
       "4                       0                     0   \n",
       "\n",
       "   EducationField_Technical Degree  Gender_Male  JobRole_Human Resources  \\\n",
       "0                                0            1                        0   \n",
       "1                                0            1                        0   \n",
       "2                                0            1                        0   \n",
       "3                                0            0                        0   \n",
       "4                                0            0                        0   \n",
       "\n",
       "   JobRole_Laboratory Technician  JobRole_Manager  \\\n",
       "0                              0                0   \n",
       "1                              0                0   \n",
       "2                              1                0   \n",
       "3                              0                0   \n",
       "4                              0                0   \n",
       "\n",
       "   JobRole_Manufacturing Director  JobRole_Research Director  \\\n",
       "0                               0                          0   \n",
       "1                               0                          0   \n",
       "2                               0                          0   \n",
       "3                               0                          0   \n",
       "4                               1                          0   \n",
       "\n",
       "   JobRole_Research Scientist  JobRole_Sales Executive  \\\n",
       "0                           1                        0   \n",
       "1                           0                        1   \n",
       "2                           0                        0   \n",
       "3                           1                        0   \n",
       "4                           0                        0   \n",
       "\n",
       "   JobRole_Sales Representative  OverTime_Yes  Age_A31-40  Age_A41-50  \\\n",
       "0                             0             1           1           0   \n",
       "1                             0             0           1           0   \n",
       "2                             0             0           1           0   \n",
       "3                             0             1           0           1   \n",
       "4                             0             0           0           1   \n",
       "\n",
       "   Age_A51-60  \n",
       "0           0  \n",
       "1           0  \n",
       "2           0  \n",
       "3           0  \n",
       "4           0  "
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# use 'get_dummies' from pandas to create dummy variables\n",
    "# use 'drop_first' to create (n-1) dummy variables\n",
    "X = pd.get_dummies(data = X, drop_first = True)\n",
    "X.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 340,
     "status": "ok",
     "timestamp": 1637212954830,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "jSqLQVCrfr55",
    "outputId": "345ed37a-7b5d-46d5-8abb-6356e3fbe80e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['DistanceFromHome', 'EnvironmentSatisfaction', 'JobInvolvement',\n",
       "       'JobLevel', 'JobSatisfaction', 'MonthlyIncome', 'NumCompaniesWorked',\n",
       "       'PercentSalaryHike', 'PerformanceRating', 'YearsAtCompany',\n",
       "       'BusinessTravel_Travel_Frequently', 'BusinessTravel_Travel_Rarely',\n",
       "       'Department_Research & Development', 'Department_Sales',\n",
       "       'EducationField_Life Sciences', 'EducationField_Marketing',\n",
       "       'EducationField_Medical', 'EducationField_Other',\n",
       "       'EducationField_Technical Degree', 'Gender_Male',\n",
       "       'JobRole_Human Resources', 'JobRole_Laboratory Technician',\n",
       "       'JobRole_Manager', 'JobRole_Manufacturing Director',\n",
       "       'JobRole_Research Director', 'JobRole_Research Scientist',\n",
       "       'JobRole_Sales Executive', 'JobRole_Sales Representative',\n",
       "       'OverTime_Yes', 'Age_A31-40', 'Age_A41-50', 'Age_A51-60'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3mbHC3ziNWPt"
   },
   "source": [
    "<a id=\"ada\"> </a>\n",
    "## 1. ADABoost"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FqmEeaLvNWPt"
   },
   "source": [
    "## We shall use the employee dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "6j6ZudhkNWPt"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>4. Build a adaboost model with 40 number of estimators to predict whether or not an employee will leave the company based on all the dummy encoded categorical features only. Calculate the accuracy for the model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "executionInfo": {
     "elapsed": 506,
     "status": "ok",
     "timestamp": 1637213355430,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "99jt4gu1NWPu"
   },
   "outputs": [],
   "source": [
    "# all the dummy encoded categorical variables are present after the index position 9\n",
    "# thus, taking a subset of X from columns at index position 10 till the last column\n",
    "X_subset = X.iloc[:, 10:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "executionInfo": {
     "elapsed": 4,
     "status": "ok",
     "timestamp": 1637213356893,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "6NfiwFKwqWLu"
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import AdaBoostClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 365,
     "status": "ok",
     "timestamp": 1637213472215,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "vE_FZvBINWPu",
    "outputId": "69f008c1-e951-429b-ed24-44eda3111ebf"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The Accuracy is:  0.7013\n"
     ]
    }
   ],
   "source": [
    "# split data into train subset and test subset\n",
    "# set 'random_state' to generate the same dataset each time you run the code \n",
    "# 'test_size' returns the proportion of data to be included in the test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_subset, df_target,random_state = 10, test_size = 0.2)\n",
    "\n",
    "# instantiate the 'AdaBoostClassifier'\n",
    "# n_estimators: number of estimators to consider\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "ada_model = AdaBoostClassifier(n_estimators = 40, random_state = 10)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "ada_model.fit(X_train, y_train)\n",
    "\n",
    "# predict the attrition on the test data\n",
    "y_pred = ada_model.predict(X_test)\n",
    "\n",
    "# print the accuracy\n",
    "# rounding the result upto 4 digits\n",
    "print(\"The Accuracy is: \", round(accuracy_score(y_test, y_pred), 4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "Z2_du_bpNWPu"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>5. Build a adaboost full model with 40 number of estimators to predict if an employee will leave the company. Generate a classification report.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 464,
     "status": "ok",
     "timestamp": 1637213649603,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Ei6B1mPnNWPu",
    "outputId": "4fe6d990-740b-47c0-9f4a-c774e328e254"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train (1189, 32)\n",
      "y_train (1189,)\n",
      "X_test (298, 32)\n",
      "y_test (298,)\n"
     ]
    }
   ],
   "source": [
    "# split data into train subset and test subset\n",
    "# set 'random_state' to generate the same dataset each time you run the code \n",
    "# 'test_size' returns the proportion of data to be included in the test set\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, df_target, random_state = 10, test_size = 0.2)\n",
    "\n",
    "# check the dimensions of the train & test subset using 'shape'\n",
    "# print dimension of train set\n",
    "print('X_train', X_train.shape)\n",
    "print('y_train', y_train.shape)\n",
    "\n",
    "# print dimension of test set\n",
    "print('X_test', X_test.shape)\n",
    "print('y_test', y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 395,
     "status": "ok",
     "timestamp": 1637213659689,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "TscPHFmANWPv",
    "outputId": "6de1f074-ff04-44b7-829d-d6cfe2dc2520"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.83      0.79      0.81       170\n",
      "           1       0.74      0.78      0.76       128\n",
      "\n",
      "    accuracy                           0.79       298\n",
      "   macro avg       0.78      0.78      0.78       298\n",
      "weighted avg       0.79      0.79      0.79       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# instantiate the 'AdaBoostClassifier'\n",
    "# n_estimators: number of estimators to consider\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "ada_model = AdaBoostClassifier(n_estimators = 40, random_state = 10)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "ada_model.fit(X_train, y_train)\n",
    "\n",
    "# predict the attrition on the test data\n",
    "y_pred_full = ada_model.predict(X_test)\n",
    "\n",
    "# print the classification report\n",
    "print(classification_report(y_test, y_pred_full))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "ZMqLuOWVNWPv"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>6.  Find the area under the receiver operating characteristic curve for full model built in question 5. </b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 407
    },
    "executionInfo": {
     "elapsed": 616,
     "status": "ok",
     "timestamp": 1637213720005,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "mA4R4Bo2NWPv",
    "outputId": "78a6b1aa-fef2-4613-83ce-67572e3d61b0"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# the roc_curve() returns the values for false positive rate, true positive rate and threshold\n",
    "# pass the actual target values and predicted probabilities to the function\n",
    "fpr, tpr, thresholds = roc_curve(y_test, y_pred_full)\n",
    "\n",
    "# plot the straight line showing worst prediction for the model\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "\n",
    "# plot the ROC curve\n",
    "plt.plot(fpr, tpr)\n",
    "\n",
    "# add the AUC score to the plot\n",
    "# 'x' and 'y' gives position of the text\n",
    "# 's' is the text \n",
    "# use round() to round-off the AUC score upto 4 digits\n",
    "plt.text(x = 0.02, y = 0.8, s = ('AUC Score:',round(metrics.roc_auc_score(y_test, y_pred_full),4)))\n",
    "                               \n",
    "# plot the grid\n",
    "plt.grid(True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "hAnOza2FNWPv"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>7. Plot a confusion matrix for the full model built above.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 410
    },
    "executionInfo": {
     "elapsed": 436,
     "status": "ok",
     "timestamp": 1637213748270,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "j09ESXL-NWPv",
    "outputId": "13d655e4-e198-4865-f9e1-11355bb09b0a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a confusion matrix\n",
    "# pass the actual and predicted target values to the confusion_matrix()\n",
    "cm = confusion_matrix(y_test, y_pred_full)\n",
    "\n",
    "# label the confusion matrix  \n",
    "# pass the matrix as 'data'\n",
    "# pass the required column names to the parameter, 'columns'\n",
    "# pass the required row names to the parameter, 'index'\n",
    "conf_matrix = pd.DataFrame(data = cm,columns = ['Predicted Attrition: No','Predicted Attrition: Yes'], \n",
    "                           index = ['Actual Attrition :No','Actual Attrition :Yes'])\n",
    "\n",
    "# plot a heatmap to visualize the confusion matrix\n",
    "# 'annot' prints the value of each grid \n",
    "# 'fmt = d' returns the integer value in each grid\n",
    "# 'cmap' assigns color to each grid\n",
    "# as we do not require different colors for each grid in the heatmap,\n",
    "# use 'ListedColormap' to assign the specified color to the grid\n",
    "# 'cbar = False' will not return the color bar to the right side of the heatmap\n",
    "# 'linewidths' assigns the width to the line that divides each grid\n",
    "# 'annot_kws = {'size':25})' assigns the font size of the annotated text \n",
    "sns.heatmap(conf_matrix, annot = True, fmt = 'd', cmap = ListedColormap(['lightskyblue']), cbar = False, \n",
    "            linewidths = 0.1, annot_kws = {'size':25})\n",
    "\n",
    "# set the font size of x-axis ticks using 'fontsize'\n",
    "plt.xticks(fontsize = 10)\n",
    "\n",
    "# set the font size of y-axis ticks using 'fontsize'\n",
    "plt.yticks(fontsize = 10)\n",
    "\n",
    "# display the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "75qf_UOQNWPw"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>8. Calculate the specificity, sensitivity, % of misclassified and correctly classified observations. What can you say about the model by looking at the sensitivity and specificity values? Is this a good model?</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 447,
     "status": "ok",
     "timestamp": 1634800129478,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "QvhIkIAINWPw",
    "outputId": "670b42d4-5548-422c-e939-12e8054519ff"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Percentage of correctly classified observations:  78.52\n",
      "Percentage of misclassified observations:  21.48\n",
      "\n",
      "Specificity:  0.7882\n",
      "Sensitivity:  0.7812\n"
     ]
    }
   ],
   "source": [
    "# True Negative from confusion matrix\n",
    "TN = cm[0][0]\n",
    "# True Positive from confusion matrix\n",
    "TP = cm[1][1]\n",
    "# False Positive from confusion matrix\n",
    "FP = cm[0][1]\n",
    "# False Negative from confusion matrix\n",
    "FN = cm[1][0]\n",
    "\n",
    "# add all the values of a confusion matrix\n",
    "total = TN + TP + FP + FN\n",
    "\n",
    "# find the percentage of correctly classified observations\n",
    "# round the result upto 2 decimal points\n",
    "correct_classified_percentage = ((TN+TP)/total)*100\n",
    "print(\"Percentage of correctly classified observations: \", round(correct_classified_percentage, 2))\n",
    "\n",
    "# find the percentage of  misclassified observations\n",
    "# round the result upto 2 decimal points\n",
    "misclassified_percentage = ((FP+FN)/total)*100\n",
    "print(\"Percentage of misclassified observations: \", round(misclassified_percentage, 2))\n",
    "\n",
    "# find the specificity\n",
    "# round the result upto 4 decimal points\n",
    "specificity = TN/(TN+FP)\n",
    "print(\"\\nSpecificity: \", round(specificity, 4))\n",
    "\n",
    "# find the sensitivity\n",
    "# round the result upto 4 decimal points\n",
    "sensitivity = TP/(TP+FN)\n",
    "print(\"Sensitivity: \", round(sensitivity, 4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XtnHkLB5NWPw"
   },
   "source": [
    "**Intrepretation:** There is no significant difference between the sensitivity and the specificity. Thus we can say that the model is performing good and it is a stable model. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "2wfNLdJZNWPw"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>9. Find the best value of hyperparameters \"n_estimators\" and \"learning_rate\" for the adaboost full model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XUUhvd8dNWPw"
   },
   "source": [
    "Given list of values:\n",
    "\n",
    "       n_estimators = [100, 120, 150, 200, 250, 300, 350],\n",
    "       learning_rate = [0.1, 0.01, 0.001, 0.15, 0.015]         "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 70386,
     "status": "ok",
     "timestamp": 1637215546149,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "9Ey9mqIyNWPx",
    "outputId": "35365e2b-6b48-46c3-d998-39d651440444"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for AdaBoost Classifier:  {'learning_rate': 0.15, 'n_estimators': 350} \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# create a dictionary with hyperparameters and its values\n",
    "# n_estimators: number of estimators to consider\n",
    "# max_depth: assigns maximum depth of the tree\n",
    "# learning_rate: Boosting learning rate\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "tuned_paramaters = [{'n_estimators': [100, 120, 150, 200, 250, 300, 350],\n",
    "                     'learning_rate': [0.1, 0.01, 0.001, 0.15, 0.015]}]\n",
    " \n",
    "# instantiate the 'AdaBoostClassifier'\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "ada_model = AdaBoostClassifier(random_state = 10)\n",
    "\n",
    "# use GridSearchCV() to find the optimal value of the hyperparameters\n",
    "# estimator: pass the decision tree classifier model\n",
    "# param_grid: pass the list 'tuned_parameters'\n",
    "# cv: number of folds in k-fold i.e. here cv = 5\n",
    "# n_jobs=-1 tells the model to use all the central processing units \n",
    "ada_grid = GridSearchCV(estimator = ada_model, \n",
    "                         param_grid = tuned_paramaters, \n",
    "                         cv = 5,\n",
    "                         n_jobs=-1)\n",
    "\n",
    "# fit the model on X_train and y_train using fit()\n",
    "ada_grid.fit(X_train, y_train)\n",
    "\n",
    "# get the best parameters\n",
    "print('Best parameters for AdaBoost Classifier: ', ada_grid.best_params_, '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "Wwhki18gNWPx"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>10. Build a adaboost model using the tuned hyperparameters, perform 10 fold cross validation and get the average ROC-AUC score.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 8392,
     "status": "ok",
     "timestamp": 1637215620616,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "vq3PWqS6NWPx",
    "outputId": "3d2543ec-ce0a-434e-f218-0fd908e63400"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Mean ROC-AUC score after 10 fold cross validation:  0.84\n"
     ]
    }
   ],
   "source": [
    "# instantiate the 'AdaBoostClassifier'\n",
    "# n_estimators: number of estimators to consider\n",
    "# 'max_depth': assigns maximum depth of the tree\n",
    "# learning_rate: Boosting learning rate\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "ada_model = AdaBoostClassifier(learning_rate = ada_grid.best_params_['learning_rate'], \n",
    "                               n_estimators = ada_grid.best_params_['n_estimators'], \n",
    "                               random_state = 10)\n",
    "\n",
    "# using cross_val_score() for k-fold cross validation\n",
    "# estimator: pass the machine learning function\n",
    "# pass the X_train and y_train sets\n",
    "# cv: stands for number of folds. Similar to k in KFold\n",
    "# scoring: pass the scoring parameter e.g. 'roc_auc' for roc-auc score\n",
    "scores = cross_val_score(estimator = ada_model, \n",
    "                         X = X_train, \n",
    "                         y = y_train, \n",
    "                         cv = 10, \n",
    "                         scoring = 'roc_auc')\n",
    "\n",
    "# get the average of the ROC-AUC scores obtained after 10 fold cross validation\n",
    "# round the result upto 2 decimal points\n",
    "print(\"Mean ROC-AUC score after 10 fold cross validation: \", round(scores.mean(), 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "executionInfo": {
     "elapsed": 517,
     "status": "ok",
     "timestamp": 1637216163441,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Q6cBNR1X48se"
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import cross_val_score,KFold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "executionInfo": {
     "elapsed": 62798,
     "status": "ok",
     "timestamp": 1637216919143,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "d3xND3J5qOkt"
   },
   "outputs": [],
   "source": [
    "kf=KFold(n_splits=5,shuffle=True,random_state=0)\n",
    "RF_be=[]\n",
    "RF_ve=[]\n",
    "for i in np.arange(1,100):\n",
    "  RF=RandomForestClassifier(n_estimators=i,criterion='entropy',random_state=0)\n",
    "  scores=cross_val_score(RF,X,df_target,cv=kf,scoring='roc_auc')\n",
    "  RF_be.append(1-np.mean(scores))\n",
    "  RF_ve.append(np.std(scores,ddof=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 419
    },
    "executionInfo": {
     "elapsed": 593,
     "status": "ok",
     "timestamp": 1637216924064,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "yKnIwYjstkpt",
    "outputId": "adc55c5c-aaa3-4882-8dc1-c8f1c88fbb1e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f716d568f90>]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_axis=np.arange(len(RF_ve))\n",
    "plt.plot(x_axis,RF_ve)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 473,
     "status": "ok",
     "timestamp": 1637216751161,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "0L4tA__ctvnh",
    "outputId": "56d90727-3bce-40fd-c540-b9180c8b8dc0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.004113382345374829, 24)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.min(RF_ve),np.argmin(RF_ve)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 364,
     "status": "ok",
     "timestamp": 1637216943832,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "x-KKXMCeu-6-",
    "outputId": "301a8607-6c53-4b1e-9511-946bacb24f9f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.0018823769014442585, 4)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.min(RF_ve),np.argmin(RF_ve)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "executionInfo": {
     "elapsed": 348,
     "status": "ok",
     "timestamp": 1637227606969,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "J3V30_-D5GJa"
   },
   "outputs": [],
   "source": [
    "LR=LogisticRegression()\n",
    "RF=RandomForestClassifier(criterion='entropy',n_estimators=5,random_state=0)\n",
    "AB=AdaBoostClassifier(learning_rate= 0.15, n_estimators=350,random_state=10)\n",
    "GB=GradientBoostingClassifier(n_estimators = 300, \n",
    "                                          max_depth = 6, \n",
    "                                          learning_rate = .15,\n",
    "                                          random_state = 8)\n",
    "XGB=XGBClassifier(gamma=2, learning_rate=0.15,max_depth=6,n_estimators=100)\n",
    "stacked = VotingClassifier(estimators = [('RandomForest',RF),('StackedGB', stack_model_gdBoost), ('StackedAB', stack_model_AdaBoost)],voting='hard')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {
    "executionInfo": {
     "elapsed": 370,
     "status": "ok",
     "timestamp": 1637227613024,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "9lpCEcNU43J1"
   },
   "outputs": [],
   "source": [
    "models = []\n",
    "models.append(('Logistic', LR))\n",
    "models.append(('RandomForest', RF))\n",
    "models.append(('AdaBoost', AB))\n",
    "models.append(('GBoost',GB))\n",
    "models.append(('XGBoost',XGB))\n",
    "models.append(('StackedAB',stack_model_AdaBoost))\n",
    "models.append(('StackedGB',stack_model_gdBoost))\n",
    "models.append(('VotingClassifier',stacked))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 0
    },
    "executionInfo": {
     "elapsed": 88086,
     "status": "ok",
     "timestamp": 1637227702711,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Nn7Pe2wb55NU",
    "outputId": "f78135d4-aaf1-441b-ddae-d1fa051a5fda"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Logistic: 0.255832 (0.057636)\n",
      "RandomForest: 0.021402 (0.017388)\n",
      "AdaBoost: 0.153366 (0.037460)\n",
      "GBoost: 0.013270 (0.010812)\n",
      "XGBoost: 0.029889 (0.014400)\n",
      "StackedAB: 0.009402 (0.010172)\n",
      "StackedGB: 0.010988 (0.010957)\n",
      "VotingClassifier: nan (nan)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "results = []\n",
    "names = []\n",
    "for name, model in models:\n",
    "\tkfold = KFold(shuffle=True,n_splits=10,random_state=0)\n",
    "\tcv_results = cross_val_score(model, X, df_target,cv=kfold, scoring='roc_auc')\n",
    "\tresults.append(cv_results)\n",
    "\tnames.append(name)\n",
    "\tprint(\"%s: %f (%f)\" % (name, np.mean(1-cv_results),np.std(cv_results,ddof=1)))\n",
    "   # boxplot algorithm comparison\n",
    "fig = plt.figure()\n",
    "fig.suptitle('Algorithm Comparison')\n",
    "ax = fig.add_subplot(111)\n",
    "plt.boxplot(results)\n",
    "ax.set_xticklabels(names)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 348,
     "status": "ok",
     "timestamp": 1634805438886,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "iITnixyf_0iJ",
    "outputId": "adbe5d90-9118-4899-dbd0-f360e8c7021b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[array([0.77608696, 0.63800657, 0.83954802, 0.69430147, 0.73148148,\n",
       "        0.76868906, 0.74279019, 0.73988971, 0.70550887, 0.80537772]),\n",
       " array([0.9990942 , 0.98211026, 0.99661017, 0.99669118, 1.        ,\n",
       "        0.97327555, 0.99197909, 0.99411765, 0.99887955, 0.99432961]),\n",
       " array([0.8192029 , 0.8090544 , 0.87702448, 0.80367647, 0.85748003,\n",
       "        0.84795955, 0.89005047, 0.83161765, 0.8162465 , 0.91402963])]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Nux6Qi4XNWPx"
   },
   "source": [
    "<a id=\"gb\"> </a>\n",
    "## 2. Gradient Boosting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "executionInfo": {
     "elapsed": 392,
     "status": "ok",
     "timestamp": 1637218771117,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "IV9cJQJs6n7r"
   },
   "outputs": [],
   "source": [
    "age=np.array([13,14,15,25,35,49,68,71,73])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 8,
     "status": "ok",
     "timestamp": 1637218772624,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "udpQY_bu7BP-",
    "outputId": "eb75aa9c-4f92-48db-afac-d08e3cbded61"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "577.1111111111111"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#mse at root\n",
    "np.mean((age-np.mean(age))**2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "executionInfo": {
     "elapsed": 363,
     "status": "ok",
     "timestamp": 1637219097570,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "5OkvRvEK7Iiv"
   },
   "outputs": [],
   "source": [
    "#est1_pred\n",
    "y1_pred=np.array([19.25,19.25,19.25,57.2,19.25,57.2,57.2,57.2,57.2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 15,
     "status": "ok",
     "timestamp": 1637219101063,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "sHFqYgcY7IqO",
    "outputId": "54379469-c96f-4dac-bcbd-880c976730e0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "221.50555555555556"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean((age-y1_pred)**2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "executionInfo": {
     "elapsed": 1451,
     "status": "ok",
     "timestamp": 1637219710643,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "dXSQ7B-R7iHD"
   },
   "outputs": [],
   "source": [
    "#combined_pred\n",
    "y_pred_out=np.array([15.68,15.68,15.68,53.63,15.68,64.33,53.63,64.33,64.33])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 6,
     "status": "ok",
     "timestamp": 1637219711929,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "OixH9E0e7iJ9",
    "outputId": "47f978f8-0d75-43f0-d963-2cc381b499f5"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "196.06334444444448"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean((age-y_pred_out)**2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 429,
     "status": "ok",
     "timestamp": 1637219780068,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "ynfmSadK5zcX",
    "outputId": "b5f39031-2951-4a45-eaa3-e4d24bfe823a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ -2.68,  -1.68,  -0.68, -28.63,  19.32, -15.33,  14.37,   6.67,\n",
       "         8.67])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "age-y_pred_out"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "JH10ny4nNWPx"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>11. Build the gradient boosting model 100 estimators and maximum depth 3. Also, find the accuracy score.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 426,
     "status": "ok",
     "timestamp": 1637221021636,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "AyhEJ37hNWPx",
    "outputId": "1cc429e1-e4e4-4324-dbce-2c57c7b7c8ab"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(random_state=8)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# instantiate the 'GradientBoostingClassifier' \n",
    "# n_estimators: number of estimators to consider\n",
    "# 'max_depth': assigns maximum depth of the tree\n",
    "# pass the 'random_state' to obtain the same results for each time you run the code\n",
    "gboost_model = GradientBoostingClassifier(n_estimators = 100, max_depth = 3, random_state = 8)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "gboost_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "bPA_6M0QNWPy"
   },
   "source": [
    "Accuracy Score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 415,
     "status": "ok",
     "timestamp": 1637221035164,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "1ubhInGTNWPy",
    "outputId": "8eae751f-288c-40ff-8176-c8d463ffd868"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The accuracy score is  0.8893\n"
     ]
    }
   ],
   "source": [
    "# predict the attrition on the test data\n",
    "y_pred_gb = gboost_model.predict(X_test)\n",
    "\n",
    "print('The accuracy score is ', round(accuracy_score(y_test, y_pred_gb), 4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "psVNe8J6NWPy"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>12. Find the best value of hyperparameters \"n_estimators\", \"learning_rate\" and \"max_depth\" for the gradient boosting full model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FPog-cwoNWPy"
   },
   "source": [
    "Given list of values:\n",
    "\n",
    "        n_estimators = [100, 120, 150, 200, 250, 300, 350],\n",
    "        learning_rate = [0.1, 0.01, 0.001, 0.15, 0.015], \n",
    "        max_depth = [2, 3, 4, 5, 6]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 374845,
     "status": "ok",
     "timestamp": 1637221557929,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "e_xtNf5nNWPy",
    "outputId": "3206339d-b616-481c-a2ed-205284f93dbd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Gradient Boositng Classifier:  {'learning_rate': 0.15, 'max_depth': 6, 'n_estimators': 300} \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# create a dictionary with hyperparameters and its values\n",
    "# learning_rate: Boosting learning rate\n",
    "# n_estimators: number of estimators to consider\n",
    "# pass the range of values to 'max_depth' that assigns maximum depth of the tree\n",
    "tuned_paramaters = [{'n_estimators': [100, 120, 150, 200, 250, 300, 350],\n",
    "                     'learning_rate': [0.1, 0.01, 0.001, 0.15, 0.015], \n",
    "                     'max_depth': [2, 3, 4, 5, 6]}]\n",
    " \n",
    "# instantiate the 'GradientBoostingClassifier'\n",
    "# pass the 'random_state' to obtain the same results for each time you run the code\n",
    "gboost_model = GradientBoostingClassifier(random_state = 8)\n",
    "\n",
    "# use GridSearchCV() to find the optimal value of the hyperparameters\n",
    "# estimator: pass the decision tree classifier model\n",
    "# param_grid: pass the list 'tuned_parameters'\n",
    "# cv: number of folds in k-fold i.e. here cv = 5\n",
    "# n_jobs=-1 tells the model to use all the central processing units \n",
    "gb_grid = GridSearchCV(estimator = gboost_model, \n",
    "                         param_grid = tuned_paramaters, \n",
    "                         cv = 5, \n",
    "                         n_jobs=-1)\n",
    "\n",
    "# fit the model on X_train and y_train using fit()\n",
    "gb_grid.fit(X_train, y_train)\n",
    "\n",
    "# get the best parameters\n",
    "print('Best parameters for Gradient Boositng Classifier: ', gb_grid.best_params_, '\\n')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "vXm30GRVNWPy"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>13. Build the gradient boosting model and an adaboost model with their best parameters and generate a classification report for both of them. Which model is the best and why?</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wfvux2wXNWPz"
   },
   "source": [
    "Gradient Boosting Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1776,
     "status": "ok",
     "timestamp": 1637221615262,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "FuxTeF7zNWPz",
    "outputId": "de15d390-9b47-4479-92c0-a382beec5ab0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(learning_rate=0.15, max_depth=6, n_estimators=300,\n",
       "                           random_state=8)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# instantiate the 'GradientBoostingClassifier' \n",
    "# n_estimators: number of estimators to consider\n",
    "# learning_rate: Boosting learning rate\n",
    "# pass the range of values to 'max_depth' that assigns maximum depth of the tree\n",
    "# consider the best parameters obtained in Q12\n",
    "gboost_model = GradientBoostingClassifier(n_estimators = 300, \n",
    "                                          max_depth = 6, \n",
    "                                          learning_rate = .15,\n",
    "                                          random_state = 8)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "gboost_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 423,
     "status": "ok",
     "timestamp": 1637221623227,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "aGwrp6i3NWPz",
    "outputId": "6ed63cb0-edb2-447e-81b9-a75ab289d08a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.95      0.96       170\n",
      "           1       0.93      0.98      0.95       128\n",
      "\n",
      "    accuracy                           0.96       298\n",
      "   macro avg       0.96      0.96      0.96       298\n",
      "weighted avg       0.96      0.96      0.96       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# predict the attrition on the test data\n",
    "y_pred_gb = gboost_model.predict(X_test)\n",
    "\n",
    "# print the classification report\n",
    "print(classification_report(y_test, y_pred_gb))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "l8F35GH1NWPz"
   },
   "source": [
    "Ada Boosting Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1081,
     "status": "ok",
     "timestamp": 1637221671224,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "8biGPVf3NWPz",
    "outputId": "cd356596-e3c9-407e-cce3-40e9bdcab5e4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AdaBoostClassifier(learning_rate=0.15, n_estimators=350, random_state=8)"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# instantiate the 'AdaBoostClassifier'\n",
    "# pass the 'random_state' to obtain the same results for each code implementation\n",
    "# consider the best parameters obtained in Q9\n",
    "ada_model = AdaBoostClassifier(learning_rate = ada_grid.best_params_['learning_rate'], \n",
    "                               n_estimators = ada_grid.best_params_['n_estimators'], \n",
    "                               random_state = 8)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "ada_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 423,
     "status": "ok",
     "timestamp": 1637221677299,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "ODw7b5VnNWPz",
    "outputId": "d8b78ce0-c4b7-4287-a6ab-348f95480e0e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.80      0.79      0.80       170\n",
      "           1       0.73      0.74      0.74       128\n",
      "\n",
      "    accuracy                           0.77       298\n",
      "   macro avg       0.77      0.77      0.77       298\n",
      "weighted avg       0.77      0.77      0.77       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# predict the attrition on the test data\n",
    "y_pred_ada = ada_model.predict(X_test)\n",
    "\n",
    "# print the classification report\n",
    "print(classification_report(y_test, y_pred_ada))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "QwrDGeKQNWP0"
   },
   "source": [
    "The gradient boosting model(gbm) is giving the highest accuracy and also the values of sensitivity and the specificity are better and high as compared to that of adaboost model. Thus, we can say that gbm is performing better than adaboost model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "iiS9iSfQNWP0"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>14. Plot the area under the receiver operating characteristic curve for tuned gradient boosting model built in question 13. Also plot a confusion matrix.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 483
    },
    "executionInfo": {
     "elapsed": 6,
     "status": "ok",
     "timestamp": 1634773454953,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "LtioBxS1NWP0",
    "outputId": "04b8e240-aaf3-4baf-b92b-57089c811993"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# the roc_curve() returns the values for false positive rate, true positive rate and threshold\n",
    "# pass the actual target values and predicted probabilities to the function\n",
    "fpr, tpr, thresholds = roc_curve(y_test, y_pred_gb)\n",
    "\n",
    "# plot the straight line showing worst prediction for the model\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "\n",
    "# plot the ROC curve\n",
    "plt.plot(fpr, tpr)\n",
    "\n",
    "# add the AUC score to the plot\n",
    "# 'x' and 'y' gives position of the text\n",
    "# 's' is the text \n",
    "# use round() to round-off the AUC score upto 4 digits\n",
    "plt.text(x = 0.02, y = 0.8, s = ('AUC Score:',round(metrics.roc_auc_score(y_test, y_pred_gb),4)))\n",
    "                               \n",
    "# plot the grid\n",
    "plt.grid(True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 483
    },
    "executionInfo": {
     "elapsed": 17,
     "status": "ok",
     "timestamp": 1634773455927,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "kuhzzUuWNWP0",
    "outputId": "6ea0fe2d-cee7-4581-870f-d91a0c1dce2a"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a confusion matrix\n",
    "# pass the actual and predicted target values to the confusion_matrix()\n",
    "cm = confusion_matrix(y_test, y_pred_gb)\n",
    "\n",
    "# label the confusion matrix  \n",
    "# pass the matrix as 'data'\n",
    "# pass the required column names to the parameter, 'columns'\n",
    "# pass the required row names to the parameter, 'index'\n",
    "conf_matrix = pd.DataFrame(data = cm,columns = ['Predicted Attrition: No','Predicted Attrition: Yes'], \n",
    "                           index = ['Actual Attrition :No','Actual Attrition :Yes'])\n",
    "\n",
    "# plot a heatmap to visualize the confusion matrix\n",
    "# 'annot' prints the value of each grid \n",
    "# 'fmt = d' returns the integer value in each grid\n",
    "# 'cmap' assigns color to each grid\n",
    "# as we do not require different colors for each grid in the heatmap,\n",
    "# use 'ListedColormap' to assign the specified color to the grid\n",
    "# 'cbar = False' will not return the color bar to the right side of the heatmap\n",
    "# 'linewidths' assigns the width to the line that divides each grid\n",
    "# 'annot_kws = {'size':25})' assigns the font size of the annotated text \n",
    "sns.heatmap(conf_matrix, annot = True, fmt = 'd', cmap = ListedColormap(['lightskyblue']), cbar = False, \n",
    "            linewidths = 0.1, annot_kws = {'size':25})\n",
    "\n",
    "# set the font size of x-axis ticks using 'fontsize'\n",
    "plt.xticks(fontsize = 10)\n",
    "\n",
    "# set the font size of y-axis ticks using 'fontsize'\n",
    "plt.yticks(fontsize = 10)\n",
    "\n",
    "# display the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 403,
     "status": "ok",
     "timestamp": 1634808479948,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "ppR5jY4uLX6v",
    "outputId": "b804dc97-9d6d-4200-c2e1-1c85ec834bfb"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9176470588235294"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "156/170"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 339,
     "status": "ok",
     "timestamp": 1634808495740,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "wmQ58SZELb-3",
    "outputId": "1bec89f8-04f6-4a40-fb43-c92beb68103f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9765625"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "125/128"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cqNBdURqNWP0"
   },
   "source": [
    "<a id=\"xgb\"> </a>\n",
    "## 3. Extreme Gradient Boosting (XGBoost)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cCZD1EWkNWP0"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                    <b>15. Build the XGBoost model with a learning rate of 0.01 and gamma equal to 2. Calculate the roc_auc score and generate a classification rerport.\n",
    "                    </b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 322,
     "status": "ok",
     "timestamp": 1637225055828,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "xYYuF9B_NWP0",
    "outputId": "39761607-8536-4876-ad5e-2714f50d4d8f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(gamma=2, learning_rate=0.01)"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# instantiate the 'XGBClassifier'\n",
    "# set the learning rate as 0.01\n",
    "# set te value of gamma as 2\n",
    "xgb_model = XGBClassifier(learning_rate = 0.01, gamma = 2)\n",
    "\n",
    "# fit the model using fit() on train data\n",
    "xgb_model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "executionInfo": {
     "elapsed": 457,
     "status": "ok",
     "timestamp": 1637225170316,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Et2iPOasNWP1"
   },
   "outputs": [],
   "source": [
    "# predict the attrition on the test data\n",
    "y_pred_xgb = xgb_model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 5,
     "status": "ok",
     "timestamp": 1637225172387,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Rbu7A9uaNWP1",
    "outputId": "42bdd509-9242-4c41-d06e-abf9650771c0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The Roc_Auc Score is  0.750643382352941\n"
     ]
    }
   ],
   "source": [
    "# print the roc_auc score\n",
    "score = roc_auc_score(y_test, y_pred_xgb)\n",
    "print('The Roc_Auc Score is ', score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 656,
     "status": "ok",
     "timestamp": 1637225174876,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Xu2tEJgtNWP1",
    "outputId": "c0ec5818-13e8-450b-c00f-c60896087a66"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.77      0.83      0.80       170\n",
      "           1       0.75      0.67      0.71       128\n",
      "\n",
      "    accuracy                           0.76       298\n",
      "   macro avg       0.76      0.75      0.75       298\n",
      "weighted avg       0.76      0.76      0.76       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# print the classification report\n",
    "print(classification_report(y_test, y_pred_xgb))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "ZeJCNPeoNWP1"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>16. Find the best value of hyperparameters \"n_estimators\", \"learning_rate\" and \"max_depth\" for the extreme gradient boosting full model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "V9R5PvK9NWP1"
   },
   "source": [
    "Given list of values:\n",
    "\n",
    "        n_estimators = [100, 120, 150],\n",
    "        learning_rate = [0.1, 0.01, 0.001, 0.15, 0.015], \n",
    "        gamma = [2, 3, 4, 5, 6],\n",
    "        max_depth = [2, 3, 4, 5, 6]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 251801,
     "status": "ok",
     "timestamp": 1637225510304,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "F8EFTzR3NWP1",
    "outputId": "65d50432-36c9-4055-dffc-b124fecce49c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best parameters for Extreme Gradient Boositng Classifier:  {'gamma': 2, 'learning_rate': 0.15, 'max_depth': 6, 'n_estimators': 100} \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# create a dictionary with hyperparameters and its values\n",
    "# gamma: Minimum loss reduction required to make a further partition on a leaf node of the tree.\n",
    "# learning_rate: Boosting learning rate\n",
    "# n_estimators: number of estimators to consider\n",
    "# pass the range of values to 'max_depth' that assigns maximum depth of the tree\n",
    "tuned_paramaters = [{'n_estimators': [100, 120, 150],\n",
    "                     'learning_rate': [0.1, 0.01, 0.001, 0.15, 0.015], \n",
    "                     'gamma': [2, 3, 4, 5, 6],\n",
    "                     'max_depth': [2, 3, 4, 5, 6]}]\n",
    " \n",
    "# instantiate the 'XGBClassifier'\n",
    "# pass the 'random_state' to obtain the same results for each time you run the code\n",
    "# n_jobs=-1 tells the model to use all the central processing units \n",
    "xgb_model = XGBClassifier(random_state = 8, n_jobs=-1)\n",
    "\n",
    "# use GridSearchCV() to find the optimal value of the hyperparameters\n",
    "# estimator: pass the decision tree classifier model\n",
    "# param_grid: pass the list 'tuned_parameters'\n",
    "# cv: number of folds in k-fold i.e. here cv = 5\n",
    "# n_jobs=-1 tells the model to use all the central processing units \n",
    "xgb_grid = GridSearchCV(estimator = xgb_model, \n",
    "                         param_grid = tuned_paramaters, \n",
    "                         cv = 5, \n",
    "                         n_jobs=-1)\n",
    "\n",
    "# fit the model on X_train and y_train using fit()\n",
    "xgb_grid.fit(X_train, y_train)\n",
    "\n",
    "# get the best parameters\n",
    "print('Best parameters for Extreme Gradient Boositng Classifier: ', xgb_grid.best_params_, '\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "executionInfo": {
     "elapsed": 392,
     "status": "ok",
     "timestamp": 1637225936979,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "6dfU88aZNWP2"
   },
   "outputs": [],
   "source": [
    "# predict the attrition on the test data\n",
    "y_pred_xgb = xgb_grid.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 537,
     "status": "ok",
     "timestamp": 1637225939450,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "XUZrzEbZNWP2",
    "outputId": "5270a181-5225-4e00-d6c7-205d4862d98b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.98      0.85      0.91       170\n",
      "           1       0.83      0.98      0.90       128\n",
      "\n",
      "    accuracy                           0.91       298\n",
      "   macro avg       0.91      0.91      0.91       298\n",
      "weighted avg       0.92      0.91      0.91       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# print the classification report\n",
    "print(classification_report(y_test, y_pred_xgb))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "ua2ctWMLNWP2"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>17. Plot the connfusion matrix for tuned extreme gradient boosting model built in question 16.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 410
    },
    "executionInfo": {
     "elapsed": 819,
     "status": "ok",
     "timestamp": 1637226030018,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "0zRmstSuNWP2",
    "outputId": "cad671f1-d332-452e-a1d3-9e1794fc784d"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a confusion matrix\n",
    "# pass the actual and predicted target values to the confusion_matrix()\n",
    "cm = confusion_matrix(y_test, y_pred_xgb)\n",
    "\n",
    "# label the confusion matrix  \n",
    "# pass the matrix as 'data'\n",
    "# pass the required column names to the parameter, 'columns'\n",
    "# pass the required row names to the parameter, 'index'\n",
    "conf_matrix = pd.DataFrame(data = cm,columns = ['Predicted Attrition: No','Predicted Attrition: Yes'], \n",
    "                           index = ['Actual Attrition :No','Actual Attrition :Yes'])\n",
    "\n",
    "# plot a heatmap to visualize the confusion matrix\n",
    "# 'annot' prints the value of each grid \n",
    "# 'fmt = d' returns the integer value in each grid\n",
    "# 'cmap' assigns color to each grid\n",
    "# as we do not require different colors for each grid in the heatmap,\n",
    "# use 'ListedColormap' to assign the specified color to the grid\n",
    "# 'cbar = False' will not return the color bar to the right side of the heatmap\n",
    "# 'linewidths' assigns the width to the line that divides each grid\n",
    "# 'annot_kws = {'size':25})' assigns the font size of the annotated text \n",
    "sns.heatmap(conf_matrix, annot = True, fmt = 'd', cmap = ListedColormap(['lightskyblue']), cbar = False, \n",
    "            linewidths = 0.1, annot_kws = {'size':25})\n",
    "\n",
    "# set the font size of x-axis ticks using 'fontsize'\n",
    "plt.xticks(fontsize = 10)\n",
    "\n",
    "# set the font size of y-axis ticks using 'fontsize'\n",
    "plt.yticks(fontsize = 10)\n",
    "\n",
    "# display the plot\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "_ZeqS0ofNWP2"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>18. Plot the area under the receiver operating characteristic curve for tuned extreme gradient boosting model built in question 16.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 407
    },
    "executionInfo": {
     "elapsed": 523,
     "status": "ok",
     "timestamp": 1637226046913,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "2dqB0J4FNWP2",
    "outputId": "5f2bb8e6-841b-4323-c686-d2f62b66228b"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# the roc_curve() returns the values for false positive rate, true positive rate and threshold\n",
    "# pass the actual target values and predicted probabilities to the function\n",
    "fpr, tpr, thresholds = roc_curve(y_test, y_pred_xgb)\n",
    "\n",
    "# plot the straight line showing worst prediction for the model\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "\n",
    "# plot the ROC curve\n",
    "plt.plot(fpr, tpr)\n",
    "\n",
    "# add the AUC score to the plot\n",
    "# 'x' and 'y' gives position of the text\n",
    "# 's' is the text \n",
    "# use round() to round-off the AUC score upto 4 digits\n",
    "plt.text(x = 0.02, y = 0.8, s = ('AUC Score:',round(metrics.roc_auc_score(y_test, y_pred_xgb),4)))\n",
    "                               \n",
    "# plot the grid\n",
    "plt.grid(True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "JpzuN1xdNWP3"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>19. Use the stacking technique with the decision tree and Random Forest model as base learners. Consider the Adaboost model as a final estimator. Also, compute the ROC-AUC score of the model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1877,
     "status": "ok",
     "timestamp": 1637226768264,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "Zuk3rb7bNWP3",
    "outputId": "a02bd905-f985-48b6-fd6f-2cfd065e329e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StackingClassifier(estimators=[('Decision Tree', DecisionTreeClassifier()),\n",
       "                               ('Random Forest', RandomForestClassifier())],\n",
       "                   final_estimator=AdaBoostClassifier(random_state=8))"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# consider the decision tree and random forest algorithms as base learners\n",
    "# pass the value of K to the parameter, 'n_neighbors'\n",
    "base_learners = [('Decision Tree', DecisionTreeClassifier()),\n",
    "                 ('Random Forest', RandomForestClassifier())]\n",
    "\n",
    "# initialize stacking classifier \n",
    "# pass the base learners to the parameter, 'estimators'\n",
    "# pass the AdaBoost model as the 'final_estimator'/ meta model\n",
    "# pass the 'random_state' to obtain the same results for each time you run the code\n",
    "stack_model_AdaBoost = StackingClassifier(estimators = base_learners, \n",
    "                        final_estimator = AdaBoostClassifier(random_state = 8))\n",
    "\n",
    "# fit the model on train dataset\n",
    "stack_model_AdaBoost.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 428,
     "status": "ok",
     "timestamp": 1637226788702,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "KvIPXiSbNWP3",
    "outputId": "bcf04078-18b8-41aa-cced-98b61f067f81"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score for the model with AdaBoost as final estimator: 0.9752297794117648\n"
     ]
    }
   ],
   "source": [
    "# calculate the predicted probabilities of target variable\n",
    "# consider only the probabilities of positive class by using the subset '[:,1]'\n",
    "y_pred_prob = stack_model_AdaBoost.predict_proba(X_test)[:, 1]\n",
    "\n",
    "# use 'roc_auc_score()' to calculate the AUC score \n",
    "print('AUC score for the model with AdaBoost as final estimator:', roc_auc_score(y_test, y_pred_prob))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 568,
     "status": "ok",
     "timestamp": 1637226791160,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "92WgixbXNWP3",
    "outputId": "8bb8a719-75e4-4439-d358-03965776e8ca",
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.97      0.97       170\n",
      "           1       0.96      0.96      0.96       128\n",
      "\n",
      "    accuracy                           0.97       298\n",
      "   macro avg       0.97      0.97      0.97       298\n",
      "weighted avg       0.97      0.97      0.97       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = stack_model_AdaBoost.predict(X_test)\n",
    "print(classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true,
    "id": "bLWrmwZINWP3"
   },
   "source": [
    "<table align=\"left\">\n",
    "    <tr>\n",
    "        <td width=\"6%\">\n",
    "            <img src=\"question_icon.png\">\n",
    "        </td>\n",
    "        <td>\n",
    "            <div align=\"left\", style=\"font-size:120%\">\n",
    "                <font color=\"#21618C\">\n",
    "                        <b>20. Use the stacking technique with the decision tree and Random Forest model as base learners. Consider the Gradient Boosting model as a final estimator. Also, compute the ROC-AUC score of the model.</b>\n",
    "                </font>\n",
    "            </div>\n",
    "        </td>\n",
    "    </tr>\n",
    "</table>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1820,
     "status": "ok",
     "timestamp": 1637227074085,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "R3kMQv6wNWP3",
    "outputId": "e871c369-700e-449d-ca3a-61380f48b2f9"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "StackingClassifier(estimators=[('Decision Tree', DecisionTreeClassifier()),\n",
       "                               ('Random Forest', RandomForestClassifier())],\n",
       "                   final_estimator=GradientBoostingClassifier(random_state=8))"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# consider the decision tree and random forest algorithms as base learners\n",
    "# pass the value of K to the parameter, 'n_neighbors'\n",
    "base_learners = [('Decision Tree', DecisionTreeClassifier()),\n",
    "                 ('Random Forest', RandomForestClassifier())]\n",
    "\n",
    "# initialize stacking classifier \n",
    "# pass the base learners to the parameter, 'estimators'\n",
    "# pass the GradientBoosting model as the 'final_estimator'/ meta model\n",
    "# pass the 'random_state' to obtain the same results for each time you run the code\n",
    "stack_model_gdBoost = StackingClassifier(estimators = base_learners, final_estimator = GradientBoostingClassifier(random_state = 8))\n",
    "\n",
    "# fit the model on train dataset\n",
    "stack_model_gdBoost.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 10,
     "status": "ok",
     "timestamp": 1634773891166,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "3vFg-3MENWP4",
    "outputId": "1e74aa51-af55-4191-f1fc-81eab02c1828"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUC score for the model with GradientBoosting as final estimator: 0.9795726102941176\n"
     ]
    }
   ],
   "source": [
    "# calculate the predicted probabilities of target variable\n",
    "# consider only the probabilities of positive class by using the subset '[:,1]'\n",
    "y_pred_prob = stack_model_gdBoost.predict_proba(X_test)[:, 1]\n",
    "\n",
    "# use 'roc_auc_score()' to calculate the AUC score \n",
    "print('AUC score for the model with GradientBoosting as final estimator:', roc_auc_score(y_test, y_pred_prob))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 478,
     "status": "ok",
     "timestamp": 1634773891640,
     "user": {
      "displayName": "Mahesh Anand",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AOh14Ghy47CXBCGfpGm71ElmYPDEY5c3AQB5tdq4ROpemw=s64",
      "userId": "06788086636039591358"
     },
     "user_tz": -330
    },
    "id": "H7GWwi1CNWP4",
    "outputId": "bfb70ac2-a5c9-4272-84d9-55c08763b979"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.98      0.97       170\n",
      "           1       0.97      0.96      0.96       128\n",
      "\n",
      "    accuracy                           0.97       298\n",
      "   macro avg       0.97      0.97      0.97       298\n",
      "weighted avg       0.97      0.97      0.97       298\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = stack_model_gdBoost.predict(X_test)\n",
    "print(classification_report(y_test, y_pred))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [
    "3mbHC3ziNWPt"
   ],
   "name": "SLC - In-Class Lab Exercise_Day4_Ensemble.ipynb",
   "provenance": []
  },
  "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}