ML_Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "b84b7064",
   "metadata": {},
   "source": [
    "# <center> Customer Segmentation - Hybrid Modelling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "46684339",
   "metadata": {},
   "source": [
    "<b>The objective is to combine unsupervised and supervised learning to create a hybrid model to categorize customers"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7db643b0",
   "metadata": {},
   "source": [
    "## Importing Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0723dd71",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.feature_selection import SelectKBest\n",
    "from sklearn.feature_selection import chi2\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.model_selection import RandomizedSearchCV\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "# Import module for data visualization\n",
    "from plotnine import *\n",
    "import plotnine\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Ignore warnings\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore', category = FutureWarning)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "02ce70c1",
   "metadata": {},
   "source": [
    "## Objective\n",
    "\n",
    "To divide the dataset into k clusters and apply multi-class classification on each cluster"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7911120f",
   "metadata": {},
   "source": [
    "## Reading Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "dc38eecb",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv(\"./data.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "edad1b56",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Ever_Married</th>\n",
       "      <th>Age</th>\n",
       "      <th>Graduated</th>\n",
       "      <th>Profession</th>\n",
       "      <th>Work_Experience</th>\n",
       "      <th>Spending_Score</th>\n",
       "      <th>Family_Size</th>\n",
       "      <th>Var_1</th>\n",
       "      <th>Segmentation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>462809</td>\n",
       "      <td>Male</td>\n",
       "      <td>No</td>\n",
       "      <td>22</td>\n",
       "      <td>No</td>\n",
       "      <td>Healthcare</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Low</td>\n",
       "      <td>4.0</td>\n",
       "      <td>Cat_4</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>462643</td>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>38</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Average</td>\n",
       "      <td>3.0</td>\n",
       "      <td>Cat_4</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>466315</td>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>67</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Low</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Cat_6</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>461735</td>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>67</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Lawyer</td>\n",
       "      <td>0.0</td>\n",
       "      <td>High</td>\n",
       "      <td>2.0</td>\n",
       "      <td>Cat_6</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>462669</td>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>40</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>NaN</td>\n",
       "      <td>High</td>\n",
       "      <td>6.0</td>\n",
       "      <td>Cat_6</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       ID  Gender Ever_Married  Age Graduated     Profession  Work_Experience  \\\n",
       "0  462809    Male           No   22        No     Healthcare              1.0   \n",
       "1  462643  Female          Yes   38       Yes       Engineer              NaN   \n",
       "2  466315  Female          Yes   67       Yes       Engineer              1.0   \n",
       "3  461735    Male          Yes   67       Yes         Lawyer              0.0   \n",
       "4  462669  Female          Yes   40       Yes  Entertainment              NaN   \n",
       "\n",
       "  Spending_Score  Family_Size  Var_1 Segmentation  \n",
       "0            Low          4.0  Cat_4            D  \n",
       "1        Average          3.0  Cat_4            A  \n",
       "2            Low          1.0  Cat_6            B  \n",
       "3           High          2.0  Cat_6            B  \n",
       "4           High          6.0  Cat_6            A  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3918d1da",
   "metadata": {},
   "source": [
    "## Data Preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "01f9d098",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 8068 entries, 0 to 8067\n",
      "Data columns (total 11 columns):\n",
      " #   Column           Non-Null Count  Dtype  \n",
      "---  ------           --------------  -----  \n",
      " 0   ID               8068 non-null   int64  \n",
      " 1   Gender           8068 non-null   object \n",
      " 2   Ever_Married     7928 non-null   object \n",
      " 3   Age              8068 non-null   int64  \n",
      " 4   Graduated        7990 non-null   object \n",
      " 5   Profession       7944 non-null   object \n",
      " 6   Work_Experience  7239 non-null   float64\n",
      " 7   Spending_Score   8068 non-null   object \n",
      " 8   Family_Size      7733 non-null   float64\n",
      " 9   Var_1            7992 non-null   object \n",
      " 10  Segmentation     8068 non-null   object \n",
      "dtypes: float64(2), int64(2), object(7)\n",
      "memory usage: 693.5+ KB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d4cf4b23",
   "metadata": {},
   "source": [
    "#### ID"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d09db07c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop ID column as its insignificant to our model\n",
    "df.drop('ID', axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd382aca",
   "metadata": {},
   "source": [
    "#### Gender"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ddcfc887",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Male', 'Female'], dtype=object)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Gender.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4e31b675",
   "metadata": {},
   "source": [
    "#### Ever_Married"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "2e97e791",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['No', 'Yes', nan], dtype=object)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Ever_Married.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4eeff595",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filling nan values with \"No\"\n",
    "df.Ever_Married.fillna('No', inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aa040dff",
   "metadata": {},
   "source": [
    "#### Age"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "bc7f161d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([22, 38, 67, 40, 56, 32, 33, 61, 55, 26, 19, 70, 58, 41, 31, 79, 49,\n",
       "       18, 36, 35, 45, 42, 83, 27, 28, 47, 29, 57, 76, 25, 72, 48, 74, 59,\n",
       "       39, 51, 30, 63, 52, 60, 68, 86, 50, 43, 80, 37, 46, 69, 78, 71, 82,\n",
       "       23, 20, 85, 21, 53, 62, 75, 65, 89, 66, 73, 77, 87, 84, 81, 88])"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Age.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eb3aeb8b",
   "metadata": {},
   "source": [
    "#### Graduated"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "e8cd5b8e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['No', 'Yes', nan], dtype=object)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Graduated.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "9f2753cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filling nan values with \"No\"\n",
    "df.Graduated.fillna(\"No\", inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "82e7f4f7",
   "metadata": {},
   "source": [
    "#### Profession"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "d267a921",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Healthcare', 'Engineer', 'Lawyer', 'Entertainment', 'Artist',\n",
       "       'Executive', 'Doctor', 'Homemaker', 'Marketing', nan], dtype=object)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Profession.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2764371b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Assuming nan values are individuals without a job, we fill nan values with 'Homemaker'\n",
    "df.Profession.fillna('Homemaker', inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7149acbb",
   "metadata": {},
   "source": [
    "#### Work Experience"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a9d00e3e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 1., nan,  0.,  4.,  9., 12.,  3., 13.,  5.,  8., 14.,  7.,  2.,\n",
       "        6., 10., 11.])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Work_Experience.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "36ef0adb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# fill nan values with 0.0\n",
    "df.Work_Experience.fillna(0.0, inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e9a2fd43",
   "metadata": {},
   "source": [
    "#### Spending_Score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "8eaa103f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Low', 'Average', 'High'], dtype=object)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Spending_Score.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "eabcdbad",
   "metadata": {},
   "source": [
    "#### Family_Size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "60c6e722",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 4.,  3.,  1.,  2.,  6., nan,  5.,  8.,  7.,  9.])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Family_Size.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "aa847292",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "335"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Family_Size.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "c0240953",
   "metadata": {},
   "outputs": [],
   "source": [
    "# fill na values with 1.0 assuming nan = single sized family\n",
    "df.Family_Size.fillna(1.0, inplace = True)\n",
    "df.Family_Size.fillna(1.0, inplace = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12fb9c8d",
   "metadata": {},
   "source": [
    "#### Var_1\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2a83adf2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Cat_4', 'Cat_6', 'Cat_7', 'Cat_3', 'Cat_1', 'Cat_2', nan, 'Cat_5'],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Var_1.unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "9e9ab220",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "76"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Var_1.isna().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "5290c30e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop rows with na values in Var_1\n",
    "df = df[df['Var_1'].notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "a945e5a5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert this column to numeric\n",
    "df['Var_1'] = df.Var_1.str.extract(r'(\\d)').astype('float64')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "0dbfe3f6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gender</th>\n",
       "      <th>Ever_Married</th>\n",
       "      <th>Age</th>\n",
       "      <th>Graduated</th>\n",
       "      <th>Profession</th>\n",
       "      <th>Work_Experience</th>\n",
       "      <th>Spending_Score</th>\n",
       "      <th>Family_Size</th>\n",
       "      <th>Var_1</th>\n",
       "      <th>Segmentation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Male</td>\n",
       "      <td>No</td>\n",
       "      <td>22</td>\n",
       "      <td>No</td>\n",
       "      <td>Healthcare</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Low</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>38</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Average</td>\n",
       "      <td>3.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>67</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>1.0</td>\n",
       "      <td>Low</td>\n",
       "      <td>1.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>67</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Lawyer</td>\n",
       "      <td>0.0</td>\n",
       "      <td>High</td>\n",
       "      <td>2.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>40</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>0.0</td>\n",
       "      <td>High</td>\n",
       "      <td>6.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Gender Ever_Married  Age Graduated     Profession  Work_Experience  \\\n",
       "0    Male           No   22        No     Healthcare              1.0   \n",
       "1  Female          Yes   38       Yes       Engineer              0.0   \n",
       "2  Female          Yes   67       Yes       Engineer              1.0   \n",
       "3    Male          Yes   67       Yes         Lawyer              0.0   \n",
       "4  Female          Yes   40       Yes  Entertainment              0.0   \n",
       "\n",
       "  Spending_Score  Family_Size  Var_1 Segmentation  \n",
       "0            Low          4.0    4.0            D  \n",
       "1        Average          3.0    4.0            A  \n",
       "2            Low          1.0    6.0            B  \n",
       "3           High          2.0    6.0            B  \n",
       "4           High          6.0    6.0            A  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0b261d93",
   "metadata": {},
   "source": [
    "## Increasing size of dataset by random sampling "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "2e15b352",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "79920"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "f07c2320",
   "metadata": {},
   "outputs": [],
   "source": [
    "sample = df.sample(frac=0.30)\n",
    "df = df.append(sample)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "ffd121db",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "103900"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.size"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "71bb65bd",
   "metadata": {},
   "source": [
    "## Feature Scaling - Normalizing numeric columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "3805afd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "# Extract numerical column positions\n",
    "numeric_cols = ['Age', 'Work_Experience', 'Family_Size', 'Var_1']\n",
    "scaler = MinMaxScaler().fit(df[numeric_cols])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "4c58e0f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "df[numeric_cols] = scaler.transform(df[numeric_cols])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "118cf0fc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gender</th>\n",
       "      <th>Ever_Married</th>\n",
       "      <th>Age</th>\n",
       "      <th>Graduated</th>\n",
       "      <th>Profession</th>\n",
       "      <th>Work_Experience</th>\n",
       "      <th>Spending_Score</th>\n",
       "      <th>Family_Size</th>\n",
       "      <th>Var_1</th>\n",
       "      <th>Segmentation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Male</td>\n",
       "      <td>No</td>\n",
       "      <td>0.056338</td>\n",
       "      <td>No</td>\n",
       "      <td>Healthcare</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.375</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>D</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.281690</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>Average</td>\n",
       "      <td>0.250</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.690141</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.690141</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Lawyer</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>High</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.309859</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>High</td>\n",
       "      <td>0.625</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Gender Ever_Married       Age Graduated     Profession  Work_Experience  \\\n",
       "0    Male           No  0.056338        No     Healthcare         0.071429   \n",
       "1  Female          Yes  0.281690       Yes       Engineer         0.000000   \n",
       "2  Female          Yes  0.690141       Yes       Engineer         0.071429   \n",
       "3    Male          Yes  0.690141       Yes         Lawyer         0.000000   \n",
       "4  Female          Yes  0.309859       Yes  Entertainment         0.000000   \n",
       "\n",
       "  Spending_Score  Family_Size     Var_1 Segmentation  \n",
       "0            Low        0.375  0.500000            D  \n",
       "1        Average        0.250  0.500000            A  \n",
       "2            Low        0.000  0.833333            B  \n",
       "3           High        0.125  0.833333            B  \n",
       "4           High        0.625  0.833333            A  "
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0695383",
   "metadata": {},
   "source": [
    "## K-prototype clustering"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "69d1df9c",
   "metadata": {},
   "source": [
    "The K-Prototype clustering algorithm in kmodes module needs categorical variables or columns position in the data. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "70faf70c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import module for k-prototype\n",
    "from kmodes.kprototypes import KPrototypes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "7178723f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 10390 entries, 0 to 5977\n",
      "Data columns (total 10 columns):\n",
      " #   Column           Non-Null Count  Dtype  \n",
      "---  ------           --------------  -----  \n",
      " 0   Gender           10390 non-null  object \n",
      " 1   Ever_Married     10390 non-null  object \n",
      " 2   Age              10390 non-null  float64\n",
      " 3   Graduated        10390 non-null  object \n",
      " 4   Profession       10390 non-null  object \n",
      " 5   Work_Experience  10390 non-null  float64\n",
      " 6   Spending_Score   10390 non-null  object \n",
      " 7   Family_Size      10390 non-null  float64\n",
      " 8   Var_1            10390 non-null  float64\n",
      " 9   Segmentation     10390 non-null  object \n",
      "dtypes: float64(4), object(6)\n",
      "memory usage: 892.9+ KB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "bf1efc8d",
   "metadata": {},
   "outputs": [],
   "source": [
    "category_columns_position = [df.columns.get_loc(col) for col in list(df.select_dtypes('O').columns)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "f29263c7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 3, 4, 6, 9]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "category_columns_position"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "28ba5181",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([['Male', 'No', 0.05633802816901412, ..., 0.375, 0.5, 'D'],\n",
       "       ['Female', 'Yes', 0.2816901408450705, ..., 0.25, 0.5, 'A'],\n",
       "       ['Female', 'Yes', 0.6901408450704226, ..., 0.0,\n",
       "        0.8333333333333334, 'B'],\n",
       "       ...,\n",
       "       ['Male', 'Yes', 0.3380281690140845, ..., 0.125,\n",
       "        0.8333333333333334, 'C'],\n",
       "       ['Male', 'No', 0.09859154929577468, ..., 0.75, 0.8333333333333334,\n",
       "        'D'],\n",
       "       ['Male', 'Yes', 0.5211267605633803, ..., 0.125,\n",
       "        0.8333333333333334, 'A']], dtype=object)"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Convert dataframe to matrix\n",
    "df_matrix = df.to_numpy()\n",
    "df_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "00c4eba6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Gender             0\n",
       "Ever_Married       0\n",
       "Age                0\n",
       "Graduated          0\n",
       "Profession         0\n",
       "Work_Experience    0\n",
       "Spending_Score     0\n",
       "Family_Size        0\n",
       "Var_1              0\n",
       "Segmentation       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.isna().sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d8f936d",
   "metadata": {},
   "source": [
    "### Elbow method for finding optimal no. of clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "6f604fd3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 3, 4, 6, 9]"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "category_columns_position"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "fab6c94e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster initiation: 1\n",
      "Cluster initiation: 2\n",
      "Cluster initiation: 3\n",
      "Cluster initiation: 4\n",
      "Cluster initiation: 5\n",
      "Cluster initiation: 6\n",
      "Cluster initiation: 7\n",
      "Cluster initiation: 8\n",
      "Cluster initiation: 9\n"
     ]
    }
   ],
   "source": [
    "cost = []\n",
    "for cluster in range(1,10):\n",
    "    try:\n",
    "        kprototype = KPrototypes(n_jobs = -1, n_clusters = cluster, init = 'Huang', random_state = 0)\n",
    "        kprototype.fit_predict(df_matrix, categorical = category_columns_position)\n",
    "        cost.append(kprototype.cost_)\n",
    "        print('Cluster initiation: {}'.format(cluster))\n",
    "    except:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "7362e3f4",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_cost = pd.DataFrame({'Cluster':range(1, 10), 'Cost':cost})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "f0298d24",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cluster</th>\n",
       "      <th>Cost</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>5719.366755</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>4565.557385</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>4056.569793</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>3747.618776</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>3503.132837</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>6</td>\n",
       "      <td>3332.062187</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7</td>\n",
       "      <td>3233.367810</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>8</td>\n",
       "      <td>3103.224866</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>9</td>\n",
       "      <td>2981.160128</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Cluster         Cost\n",
       "0        1  5719.366755\n",
       "1        2  4565.557385\n",
       "2        3  4056.569793\n",
       "3        4  3747.618776\n",
       "4        5  3503.132837\n",
       "5        6  3332.062187\n",
       "6        7  3233.367810\n",
       "7        8  3103.224866\n",
       "8        9  2981.160128"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_cost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "50a300bd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x480 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<ggplot: (674341150)>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Data viz\n",
    "plotnine.options.figure_size = (8, 4.8)\n",
    "(\n",
    "    ggplot(data = df_cost)+\n",
    "    geom_line(aes(x = 'Cluster',\n",
    "                  y = 'Cost'))+\n",
    "    geom_point(aes(x = 'Cluster',\n",
    "                   y = 'Cost'))+\n",
    "    geom_label(aes(x = 'Cluster',\n",
    "                   y = 'Cost',\n",
    "                   label = 'Cluster'),\n",
    "               size = 10,\n",
    "               nudge_y = 1000) +\n",
    "    labs(title = 'Optimal number of cluster with Elbow Method')+\n",
    "    xlab('Number of Clusters k')+\n",
    "    ylab('Cost')+\n",
    "    theme_minimal()\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8353fd57",
   "metadata": {},
   "source": [
    "<b> Based on the above plot, lets assume k = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "87ba51b3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 2, 2, ..., 2, 1, 2], dtype=uint16)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Fitting the clusters\n",
    "kprototype = KPrototypes(n_jobs = -1, n_clusters = 3, init = 'Huang', random_state = 0)\n",
    "kprototype.fit_predict(df_matrix, categorical = category_columns_position)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "6b4b900b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add clusters \n",
    "df['Cluster'] = kprototype.labels_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "99faae1c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gender</th>\n",
       "      <th>Ever_Married</th>\n",
       "      <th>Age</th>\n",
       "      <th>Graduated</th>\n",
       "      <th>Profession</th>\n",
       "      <th>Work_Experience</th>\n",
       "      <th>Spending_Score</th>\n",
       "      <th>Family_Size</th>\n",
       "      <th>Var_1</th>\n",
       "      <th>Segmentation</th>\n",
       "      <th>Cluster</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1414</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.464789</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6985</th>\n",
       "      <td>Male</td>\n",
       "      <td>No</td>\n",
       "      <td>0.436620</td>\n",
       "      <td>No</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.250</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>D</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7423</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.338028</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Artist</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>Average</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>C</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4963</th>\n",
       "      <td>Male</td>\n",
       "      <td>No</td>\n",
       "      <td>0.098592</td>\n",
       "      <td>No</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>0.142857</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.750</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>D</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5977</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.521127</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Entertainment</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>A</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Gender Ever_Married       Age Graduated     Profession  Work_Experience  \\\n",
       "1414   Male          Yes  0.464789       Yes       Engineer         0.000000   \n",
       "6985   Male           No  0.436620        No  Entertainment         0.071429   \n",
       "7423   Male          Yes  0.338028       Yes         Artist         0.500000   \n",
       "4963   Male           No  0.098592        No  Entertainment         0.142857   \n",
       "5977   Male          Yes  0.521127       Yes  Entertainment         0.071429   \n",
       "\n",
       "     Spending_Score  Family_Size     Var_1 Segmentation  Cluster  \n",
       "1414            Low        0.125  0.500000            A        2  \n",
       "6985            Low        0.250  0.333333            D        1  \n",
       "7423        Average        0.125  0.833333            C        2  \n",
       "4963            Low        0.750  0.833333            D        1  \n",
       "5977            Low        0.125  0.833333            A        2  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "2502d8a5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2    4535\n",
       "1    2937\n",
       "0    2918\n",
       "Name: Cluster, dtype: int64"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['Cluster'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "c3b73eec",
   "metadata": {},
   "outputs": [],
   "source": [
    "df1 = df[df.Cluster == 0]\n",
    "df2 = df[df.Cluster == 1]\n",
    "df3 = df[df.Cluster == 2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "39b8dff1",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/664932745.py:1: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/664932745.py:2: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/664932745.py:3: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n"
     ]
    }
   ],
   "source": [
    "df1.drop(['Cluster'], axis = 1, inplace = True)\n",
    "df2.drop(['Cluster'], axis = 1, inplace = True)\n",
    "df3.drop(['Cluster'], axis = 1, inplace = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "70fd512d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "A    1366\n",
       "B     709\n",
       "D     533\n",
       "C     310\n",
       "Name: Segmentation, dtype: int64"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1['Segmentation'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "9d90022f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "D    2025\n",
       "A     334\n",
       "B     293\n",
       "C     285\n",
       "Name: Segmentation, dtype: int64"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2['Segmentation'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "3c64d6e9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "C    1942\n",
       "B    1390\n",
       "A     844\n",
       "D     359\n",
       "Name: Segmentation, dtype: int64"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df3['Segmentation'].value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b3da6493",
   "metadata": {},
   "source": [
    "## Multi-Class Classification on each cluster"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0ce67c0e",
   "metadata": {},
   "source": [
    "<b> Now, we have 3 subsets of the data from clustering. We perform classification individually on each subset of data and compute the mean accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f2516dc",
   "metadata": {},
   "source": [
    "From the above distributions, we can see that the classes are imbalanced, and this will cause the model to be biased towards a particular class. Therefore, we need to oversample to address the class imbalance problem."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "09847a32",
   "metadata": {},
   "source": [
    "## Decision Tree Classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4dcaaf13",
   "metadata": {},
   "source": [
    "#### Label Encoding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "47a9f619",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0, 1, 3, 4, 6, 9]"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "category_columns_position"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "f88a6d8f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Gender</th>\n",
       "      <th>Ever_Married</th>\n",
       "      <th>Age</th>\n",
       "      <th>Graduated</th>\n",
       "      <th>Profession</th>\n",
       "      <th>Work_Experience</th>\n",
       "      <th>Spending_Score</th>\n",
       "      <th>Family_Size</th>\n",
       "      <th>Var_1</th>\n",
       "      <th>Segmentation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Female</td>\n",
       "      <td>No</td>\n",
       "      <td>0.112676</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Engineer</td>\n",
       "      <td>0.071429</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.250</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>Female</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.563380</td>\n",
       "      <td>No</td>\n",
       "      <td>Doctor</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.436620</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Homemaker</td>\n",
       "      <td>0.857143</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>Male</td>\n",
       "      <td>Yes</td>\n",
       "      <td>0.211268</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Artist</td>\n",
       "      <td>0.928571</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.333333</td>\n",
       "      <td>A</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>Female</td>\n",
       "      <td>No</td>\n",
       "      <td>0.253521</td>\n",
       "      <td>Yes</td>\n",
       "      <td>Artist</td>\n",
       "      <td>0.357143</td>\n",
       "      <td>Low</td>\n",
       "      <td>0.125</td>\n",
       "      <td>0.833333</td>\n",
       "      <td>B</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    Gender Ever_Married       Age Graduated Profession  Work_Experience  \\\n",
       "10  Female           No  0.112676       Yes   Engineer         0.071429   \n",
       "14  Female          Yes  0.563380        No     Doctor         0.000000   \n",
       "20    Male          Yes  0.436620       Yes  Homemaker         0.857143   \n",
       "22    Male          Yes  0.211268       Yes     Artist         0.928571   \n",
       "23  Female           No  0.253521       Yes     Artist         0.357143   \n",
       "\n",
       "   Spending_Score  Family_Size     Var_1 Segmentation  \n",
       "10            Low        0.250  0.833333            A  \n",
       "14            Low        0.000  0.333333            B  \n",
       "20            Low        0.000  0.333333            A  \n",
       "22            Low        0.125  0.333333            A  \n",
       "23            Low        0.125  0.833333            B  "
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "ab986ae2",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:4: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:5: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:6: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:7: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:8: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:9: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:11: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:12: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:13: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:14: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:15: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:16: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:18: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:19: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:20: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:21: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:22: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "/var/folders/c9/s8djz0q939q5zk6br0gymtm80000gn/T/ipykernel_89714/1268981441.py:23: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n"
     ]
    }
   ],
   "source": [
    "from sklearn.preprocessing import LabelEncoder\n",
    "\n",
    "le = LabelEncoder()\n",
    "df1.Gender = le.fit_transform(df1['Gender'])\n",
    "df1.Ever_Married = le.fit_transform(df1['Ever_Married'])\n",
    "df1.Graduated = le.fit_transform(df1['Graduated'])\n",
    "df1.Profession = le.fit_transform(df1['Profession'])\n",
    "df1.Spending_Score = le.fit_transform(df1['Spending_Score'])\n",
    "df1.Segmentation = le.fit_transform(df1['Segmentation'])\n",
    "\n",
    "df2.Gender = le.fit_transform(df2['Gender'])\n",
    "df2.Ever_Married = le.fit_transform(df2['Ever_Married'])\n",
    "df2.Graduated = le.fit_transform(df2['Graduated'])\n",
    "df2.Profession = le.fit_transform(df2['Profession'])\n",
    "df2.Spending_Score = le.fit_transform(df2['Spending_Score'])\n",
    "df2.Segmentation = le.fit_transform(df2['Segmentation'])\n",
    "\n",
    "df3.Gender = le.fit_transform(df3['Gender'])\n",
    "df3.Ever_Married = le.fit_transform(df3['Ever_Married'])\n",
    "df3.Graduated = le.fit_transform(df3['Graduated'])\n",
    "df3.Profession = le.fit_transform(df3['Profession'])\n",
    "df3.Spending_Score = le.fit_transform(df3['Spending_Score'])\n",
    "df3.Segmentation = le.fit_transform(df3['Segmentation'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "80f15e5c",
   "metadata": {},
   "source": [
    "### Cluster 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2375a7f2",
   "metadata": {},
   "source": [
    "### Addressing Class Imbalance Problem"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "f54b3128",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    1366\n",
       "1     709\n",
       "3     533\n",
       "2     310\n",
       "Name: Segmentation, dtype: int64"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1['Segmentation'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "7de1d0ac",
   "metadata": {},
   "outputs": [],
   "source": [
    "from imblearn.over_sampling import SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "91764573",
   "metadata": {},
   "outputs": [],
   "source": [
    "# X is all columns except the segmentation, y is segmentation\n",
    "X = df1.drop(['Segmentation'], axis = 1)\n",
    "y = df1['Segmentation']\n",
    "\n",
    "oversample = SMOTE()\n",
    "X, y = oversample.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "aeb1107c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    1366\n",
       "1    1366\n",
       "2    1366\n",
       "3    1366\n",
       "Name: Segmentation, dtype: int64"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "298b2e1f",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "48c7c490",
   "metadata": {},
   "source": [
    "### Hyperparameter Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "a7543220",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "def dtree_grid_search(X,y,nfolds):\n",
    "    #create a dictionary of all values we want to test\n",
    "    param_grid = { 'criterion':['gini','entropy'],'max_depth': np.arange(3, 20)}\n",
    "    # decision tree model\n",
    "    dtree_model=DecisionTreeClassifier()\n",
    "    #use gridsearch to test all values\n",
    "    dtree_gscv = GridSearchCV(dtree_model, param_grid, cv=nfolds)\n",
    "    #fit model to data\n",
    "    dtree_gscv.fit(X, y)\n",
    "    return dtree_gscv.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "9f0427b1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'criterion': 'gini', 'max_depth': 15}\n"
     ]
    }
   ],
   "source": [
    "best = dtree_grid_search(X_train, y_train, 10)\n",
    "print(best)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "abcd3a01",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.tree import DecisionTreeClassifier\n",
    "dtree = DecisionTreeClassifier(criterion = best['criterion'], \n",
    "                               max_depth = best['max_depth'], \n",
    "                               random_state = 1).fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "bdfce47b",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = dtree.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "b6296351",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.62      0.63      0.62       462\n",
      "           1       0.64      0.65      0.64       431\n",
      "           2       0.73      0.70      0.72       464\n",
      "           3       0.74      0.75      0.74       447\n",
      "\n",
      "    accuracy                           0.68      1804\n",
      "   macro avg       0.68      0.68      0.68      1804\n",
      "weighted avg       0.68      0.68      0.68      1804\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "d93188ff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.6829268292682927\n"
     ]
    }
   ],
   "source": [
    "acc1 = accuracy_score(y_test, predictions)\n",
    "print(acc1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "05ce5bb0",
   "metadata": {},
   "source": [
    "### Cluster 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "d74a9733",
   "metadata": {},
   "outputs": [],
   "source": [
    "# X is all columns except the segmentation, y is segmentation\n",
    "X = df2.drop(['Segmentation'], axis = 1)\n",
    "y = df2['Segmentation']\n",
    "oversample = SMOTE()\n",
    "X, y = oversample.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "bbb28a85",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "dc3c9e15",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'criterion': 'entropy', 'max_depth': 19}\n"
     ]
    }
   ],
   "source": [
    "best = dtree_grid_search(X_train, y_train, 10)\n",
    "print(best)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "71a5e471",
   "metadata": {},
   "outputs": [],
   "source": [
    "dtree = DecisionTreeClassifier(criterion = best['criterion'], \n",
    "                               max_depth = best['max_depth'], \n",
    "                               random_state = 1).fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "b8cf7818",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = dtree.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "658f52d9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.84      0.84      0.84       654\n",
      "           1       0.82      0.81      0.81       717\n",
      "           2       0.78      0.81      0.79       628\n",
      "           3       0.82      0.79      0.81       674\n",
      "\n",
      "    accuracy                           0.81      2673\n",
      "   macro avg       0.81      0.81      0.81      2673\n",
      "weighted avg       0.81      0.81      0.81      2673\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "57e90a30",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8114478114478114\n"
     ]
    }
   ],
   "source": [
    "acc2 = accuracy_score(y_test, predictions)\n",
    "print(acc2)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "563090b4",
   "metadata": {},
   "source": [
    "### Cluster 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "36c2a6bd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# X is all columns except the segmentation, y is segmentation\n",
    "X = df3.drop(['Segmentation'], axis = 1)\n",
    "y = df3['Segmentation']\n",
    "oversample = SMOTE()\n",
    "X, y = oversample.fit_resample(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "fdd83737",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "3dee8e9f",
   "metadata": {},
   "outputs": [],
   "source": [
    "best = dtree_grid_search(X_train, y_train, 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "6521088f",
   "metadata": {},
   "outputs": [],
   "source": [
    "dtree = DecisionTreeClassifier(criterion = best['criterion'], \n",
    "                               max_depth = best['max_depth'], \n",
    "                               random_state = 1).fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "3af6ecdc",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = dtree.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "8b4b7cc8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.62      0.64      0.63       633\n",
      "           1       0.57      0.55      0.56       647\n",
      "           2       0.63      0.67      0.65       647\n",
      "           3       0.80      0.76      0.78       637\n",
      "\n",
      "    accuracy                           0.65      2564\n",
      "   macro avg       0.66      0.65      0.65      2564\n",
      "weighted avg       0.66      0.65      0.65      2564\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test, predictions))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "0f8c8d6b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.6540561622464899\n"
     ]
    }
   ],
   "source": [
    "acc3 = accuracy_score(y_test, predictions)\n",
    "print(acc3)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a1f40d4",
   "metadata": {},
   "source": [
    "## Computing Final Accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2cd64eef",
   "metadata": {},
   "source": [
    "We take the mean accuracy from all the 3 clusters to compute the final accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "367fbd6c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Final accuracy = 71.61 %\n"
     ]
    }
   ],
   "source": [
    "print(f'Final accuracy = {round((acc1+acc2+acc3)*100/3, 2)} %')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.7"
  },
  "vscode": {
   "interpreter": {
    "hash": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}