{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "aUF-lH3fFBsa"
   },
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2336,
     "status": "ok",
     "timestamp": 1573001775617,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "8JSOq_4DFCKB",
    "outputId": "97e1bf86-b8b7-4b63-b795-0a771d1e0336"
   },
   "outputs": [],
   "source": [
    "file_url = '../bank-full.csv'\n",
    "bankData = pd.read_csv(file_url, sep=\";\")\n",
    "bankData.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 173
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2328,
     "status": "ok",
     "timestamp": 1573001775618,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "5_K4qSskFOiX",
    "outputId": "4f6d26d9-9d9c-41fb-bb4c-bec08a74da4c"
   },
   "outputs": [],
   "source": [
    "# Relationship between housing and propensity for term deposits\n",
    "bankData.groupby(['housing', 'y'])['y'].agg(houseTot='count').reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 173
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2322,
     "status": "ok",
     "timestamp": 1573001775619,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "HBJQtKjUFfC1",
    "outputId": "42939e01-5c7f-4e13-8f93-a97fba2d62bb"
   },
   "outputs": [],
   "source": [
    "# Relationship between having a loan and propensity for term deposits\n",
    "bankData.groupby(['loan', 'y'])['y'].agg(loanTot='count').reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2315,
     "status": "ok",
     "timestamp": 1573001775619,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "ktpNZGC2Qz70",
    "outputId": "ca650459-4e00-427e-e913-c84e46d4e26b"
   },
   "outputs": [],
   "source": [
    "# Taking the quantiles for 25%, 50% and 75% of the balance data\n",
    "import numpy as np\n",
    "np.quantile(bankData['balance'],[0.25,0.5,0.75])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2309,
     "status": "ok",
     "timestamp": 1573001775620,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "BbcbCowlRCsc",
    "outputId": "6bdb3a5c-42d4-4566-c04a-929bf614db1b"
   },
   "outputs": [],
   "source": [
    "# Creating new features for bank data based on the quantile values\n",
    "\n",
    "bankData['balanceClass'] = 'Quant1'\n",
    "\n",
    "bankData.loc[(bankData['balance'] > 72) & (bankData['balance'] < 448), 'balanceClass'] = 'Quant2'\n",
    "\n",
    "bankData.loc[(bankData['balance'] > 448) & (bankData['balance'] < 1428), 'balanceClass'] = 'Quant3'\n",
    "\n",
    "bankData.loc[bankData['balance'] > 1428, 'balanceClass'] = 'Quant4'\n",
    "\n",
    "bankData.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 173
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2303,
     "status": "ok",
     "timestamp": 1573001775621,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "fGVxqFC3FzTL",
    "outputId": "1ec443fa-32ee-4482-e620-1f8d27091e0a"
   },
   "outputs": [],
   "source": [
    "# Calculating the customers under each quantile \n",
    "balanceTot = bankData.groupby(['balanceClass'])['y'].agg(balanceTot='count').reset_index()\n",
    "balanceTot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 297
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2294,
     "status": "ok",
     "timestamp": 1573001775621,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "vKet2DvBF_p3",
    "outputId": "d56e51b5-ed2e-4d4e-9c97-dcae53e88a03"
   },
   "outputs": [],
   "source": [
    "# Calculating the total customers categorised as per quantile and propensity classification\n",
    "balanceProp = bankData.groupby(['balanceClass', 'y'])['y'].agg(balanceCat='count').reset_index()\n",
    "balanceProp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 297
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2289,
     "status": "ok",
     "timestamp": 1573001775622,
     "user": {
      "displayName": "Anthony So",
      "photoUrl": "https://lh3.googleusercontent.com/a-/AAuE7mCYY-iGjUIqBSnlLoszfZTN7rU7FRNg05Rdt9Ii3A=s64",
      "userId": "11809607246124237079"
     },
     "user_tz": -660
    },
    "id": "ljhfRWOAGHKi",
    "outputId": "f0db8c52-9d68-4a77-d5c0-46f86758fd4c"
   },
   "outputs": [],
   "source": [
    "# Merging both the data frames\n",
    "balanceComb = pd.merge(balanceProp, balanceTot, on = ['balanceClass'])\n",
    "balanceComb['catProp'] = (balanceComb.balanceCat / balanceComb.balanceTot)*100\n",
    "balanceComb"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "Exercise3.03_updated.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": "3.8.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}