Dealing with Missing Data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "df8cddd8-ddb5-45e1-ab19-ca78e89ea336",
   "metadata": {},
   "source": [
    "# Data Cleansing Part 2\n",
    "\n",
    "## How to Deal with Missing Data?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "32058647-1718-4ea9-8504-6d3130f1a4cc",
   "metadata": {},
   "source": [
    "### Read the dataset|"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "f68081a4-da84-4629-8e3e-17f59d75f16c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import libraries\n",
    "import pandas as pd\n",
    "import numpy as np\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "f1e81a98-5055-45ba-a92d-b81dc2bc31d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# read the dataset\n",
    "df = pd.read_excel('credit-data-training.xlsx')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "b9858279-168b-4fd2-92da-a9763f85ca45",
   "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>Credit-Application-Result</th>\n",
       "      <th>Account-Balance</th>\n",
       "      <th>Duration-of-Credit-Month</th>\n",
       "      <th>Payment-Status-of-Previous-Credit</th>\n",
       "      <th>Purpose</th>\n",
       "      <th>Credit-Amount</th>\n",
       "      <th>Value-Savings-Stocks</th>\n",
       "      <th>Length-of-current-employment</th>\n",
       "      <th>Instalment-per-cent</th>\n",
       "      <th>Guarantors</th>\n",
       "      <th>Duration-in-Current-address</th>\n",
       "      <th>Most-valuable-available-asset</th>\n",
       "      <th>Age-years</th>\n",
       "      <th>Concurrent-Credits</th>\n",
       "      <th>Type-of-apartment</th>\n",
       "      <th>No-of-Credits-at-this-Bank</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>No-of-dependents</th>\n",
       "      <th>Telephone</th>\n",
       "      <th>Foreign-Worker</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Other</td>\n",
       "      <td>1494</td>\n",
       "      <td>£100-£1000</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>1494</td>\n",
       "      <td>£100-£1000</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1</td>\n",
       "      <td>29.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>No Problems (in this bank)</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>1544</td>\n",
       "      <td>None</td>\n",
       "      <td>1-4 yrs</td>\n",
       "      <td>2</td>\n",
       "      <td>None</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>42.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>More than 1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>No Problems (in this bank)</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>3380</td>\n",
       "      <td>None</td>\n",
       "      <td>1-4 yrs</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>No Account</td>\n",
       "      <td>6</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>343</td>\n",
       "      <td>None</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>4</td>\n",
       "      <td>None</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>27.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Credit-Application-Result Account-Balance  Duration-of-Credit-Month  \\\n",
       "0              Creditworthy    Some Balance                         4   \n",
       "1              Creditworthy    Some Balance                         4   \n",
       "2              Creditworthy    Some Balance                         4   \n",
       "3              Creditworthy    Some Balance                         4   \n",
       "4              Creditworthy      No Account                         6   \n",
       "\n",
       "  Payment-Status-of-Previous-Credit       Purpose  Credit-Amount  \\\n",
       "0                           Paid Up         Other           1494   \n",
       "1                           Paid Up  Home Related           1494   \n",
       "2        No Problems (in this bank)  Home Related           1544   \n",
       "3        No Problems (in this bank)  Home Related           3380   \n",
       "4                           Paid Up  Home Related            343   \n",
       "\n",
       "  Value-Savings-Stocks Length-of-current-employment  Instalment-per-cent  \\\n",
       "0           £100-£1000                        < 1yr                    1   \n",
       "1           £100-£1000                        < 1yr                    1   \n",
       "2                 None                      1-4 yrs                    2   \n",
       "3                 None                      1-4 yrs                    1   \n",
       "4                 None                        < 1yr                    4   \n",
       "\n",
       "  Guarantors  Duration-in-Current-address  Most-valuable-available-asset  \\\n",
       "0       None                          2.0                              1   \n",
       "1       None                          2.0                              1   \n",
       "2       None                          1.0                              1   \n",
       "3       None                          1.0                              1   \n",
       "4       None                          1.0                              1   \n",
       "\n",
       "   Age-years Concurrent-Credits  Type-of-apartment No-of-Credits-at-this-Bank  \\\n",
       "0        NaN  Other Banks/Depts                  2                          1   \n",
       "1       29.0  Other Banks/Depts                  2                          1   \n",
       "2       42.0  Other Banks/Depts                  2                More than 1   \n",
       "3       37.0  Other Banks/Depts                  2                          1   \n",
       "4       27.0  Other Banks/Depts                  2                          1   \n",
       "\n",
       "   Occupation  No-of-dependents  Telephone  Foreign-Worker  \n",
       "0           1                 2          1               2  \n",
       "1           1                 2          1               2  \n",
       "2           1                 2          1               1  \n",
       "3           1                 2          1               1  \n",
       "4           1                 1          1               1  "
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# explore the data\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "7693b4ee-a880-4944-8b40-53dc9c0bc907",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(500, 20)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the dimensions of the data\n",
    "df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "298baad7-0b95-4ad8-817f-56becf91fc97",
   "metadata": {},
   "source": [
    "## Let's check and calculate the missing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "f8d67ab1-4353-43d1-bd3b-4aadad3c6906",
   "metadata": {},
   "outputs": [],
   "source": [
    "# let's calculate missing values\n",
    "total = df.isnull().sum().sort_values(ascending=False)\n",
    "# let's calculate the percentage of missing values\n",
    "percent = df.isnull().sum() / df.isnull().count() * 100\n",
    "# round the percentage into 1 digit after decimal\n",
    "percent_rounded = (round(percent, 1)).sort_values(ascending=False)\n",
    "\n",
    "# concatenate the total and percentage \n",
    "missing_data = pd.concat([total, percent_rounded], axis=1, keys=['Total', '%'])\n",
    "\n",
    "# let's get missing data which is total values greater than 0\n",
    "\n",
    "missing_data = missing_data[missing_data['Total']>0]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "6bcf8212-318a-455a-b236-04663b446c82",
   "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>Total</th>\n",
       "      <th>%</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Duration-in-Current-address</th>\n",
       "      <td>344</td>\n",
       "      <td>68.8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Age-years</th>\n",
       "      <td>12</td>\n",
       "      <td>2.4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                             Total     %\n",
       "Duration-in-Current-address    344  68.8\n",
       "Age-years                       12   2.4"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# display the result\n",
    "missing_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "fa920a6c-2181-4ae7-a7eb-00b738a6e539",
   "metadata": {},
   "outputs": [],
   "source": [
    "# drop the duration-in-current-address\n",
    "df.drop(columns='Duration-in-Current-address', axis=1, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "177d7d24-4292-4971-9ad1-55dc2d5a59b5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# compare the mean and median \n",
    "age_mean = df['Age-years'].mean()\n",
    "age_median = df['Age-years'].median()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "0054dcc3-f27a-480e-9b64-de7712e146f0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mean of Age-years: 35.63729508196721\n",
      "The median of Age-years 33.0\n"
     ]
    }
   ],
   "source": [
    "# print the result\n",
    "print(\"The mean of Age-years:\", age_mean)\n",
    "print(\"The median of Age-years\", age_median)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "d2987865-76fb-4461-8fd8-da1dc148b334",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import SimpleImputer Object from the sklearn\n",
    "from sklearn.impute import SimpleImputer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "91944f4f-4614-4287-a4ae-3c6fd189c9f2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# let's impute the missing value of age-years with median as a strategy\n",
    "\n",
    "# create impute object\n",
    "imp = SimpleImputer(missing_values=np.nan, strategy='median')\n",
    "\n",
    "# get the values of Age-years and reshape into 2D array\n",
    "age_t = df['Age-years'].values\n",
    "\n",
    "# transform the age values\n",
    "df['Age-years'] = imp.fit_transform(age_t.reshape(-1, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "1adb24d5-49b6-41a7-a735-c02e5b341704",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Credit-Application-Result            0\n",
       "Account-Balance                      0\n",
       "Duration-of-Credit-Month             0\n",
       "Payment-Status-of-Previous-Credit    0\n",
       "Purpose                              0\n",
       "Credit-Amount                        0\n",
       "Value-Savings-Stocks                 0\n",
       "Length-of-current-employment         0\n",
       "Instalment-per-cent                  0\n",
       "Guarantors                           0\n",
       "Most-valuable-available-asset        0\n",
       "Age-years                            0\n",
       "Concurrent-Credits                   0\n",
       "Type-of-apartment                    0\n",
       "No-of-Credits-at-this-Bank           0\n",
       "Occupation                           0\n",
       "No-of-dependents                     0\n",
       "Telephone                            0\n",
       "Foreign-Worker                       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the missing data again\n",
    "df.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "31293bd0-43cb-4687-9015-936f322396a3",
   "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>Credit-Application-Result</th>\n",
       "      <th>Account-Balance</th>\n",
       "      <th>Duration-of-Credit-Month</th>\n",
       "      <th>Payment-Status-of-Previous-Credit</th>\n",
       "      <th>Purpose</th>\n",
       "      <th>Credit-Amount</th>\n",
       "      <th>Value-Savings-Stocks</th>\n",
       "      <th>Length-of-current-employment</th>\n",
       "      <th>Instalment-per-cent</th>\n",
       "      <th>Guarantors</th>\n",
       "      <th>Most-valuable-available-asset</th>\n",
       "      <th>Age-years</th>\n",
       "      <th>Concurrent-Credits</th>\n",
       "      <th>Type-of-apartment</th>\n",
       "      <th>No-of-Credits-at-this-Bank</th>\n",
       "      <th>Occupation</th>\n",
       "      <th>No-of-dependents</th>\n",
       "      <th>Telephone</th>\n",
       "      <th>Foreign-Worker</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Other</td>\n",
       "      <td>1494</td>\n",
       "      <td>£100-£1000</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>1</td>\n",
       "      <td>33.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>1494</td>\n",
       "      <td>£100-£1000</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>1</td>\n",
       "      <td>29.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>No Problems (in this bank)</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>1544</td>\n",
       "      <td>None</td>\n",
       "      <td>1-4 yrs</td>\n",
       "      <td>2</td>\n",
       "      <td>None</td>\n",
       "      <td>1</td>\n",
       "      <td>42.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>More than 1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>Some Balance</td>\n",
       "      <td>4</td>\n",
       "      <td>No Problems (in this bank)</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>3380</td>\n",
       "      <td>None</td>\n",
       "      <td>1-4 yrs</td>\n",
       "      <td>1</td>\n",
       "      <td>None</td>\n",
       "      <td>1</td>\n",
       "      <td>37.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Creditworthy</td>\n",
       "      <td>No Account</td>\n",
       "      <td>6</td>\n",
       "      <td>Paid Up</td>\n",
       "      <td>Home Related</td>\n",
       "      <td>343</td>\n",
       "      <td>None</td>\n",
       "      <td>&lt; 1yr</td>\n",
       "      <td>4</td>\n",
       "      <td>None</td>\n",
       "      <td>1</td>\n",
       "      <td>27.0</td>\n",
       "      <td>Other Banks/Depts</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Credit-Application-Result Account-Balance  Duration-of-Credit-Month  \\\n",
       "0              Creditworthy    Some Balance                         4   \n",
       "1              Creditworthy    Some Balance                         4   \n",
       "2              Creditworthy    Some Balance                         4   \n",
       "3              Creditworthy    Some Balance                         4   \n",
       "4              Creditworthy      No Account                         6   \n",
       "\n",
       "  Payment-Status-of-Previous-Credit       Purpose  Credit-Amount  \\\n",
       "0                           Paid Up         Other           1494   \n",
       "1                           Paid Up  Home Related           1494   \n",
       "2        No Problems (in this bank)  Home Related           1544   \n",
       "3        No Problems (in this bank)  Home Related           3380   \n",
       "4                           Paid Up  Home Related            343   \n",
       "\n",
       "  Value-Savings-Stocks Length-of-current-employment  Instalment-per-cent  \\\n",
       "0           £100-£1000                        < 1yr                    1   \n",
       "1           £100-£1000                        < 1yr                    1   \n",
       "2                 None                      1-4 yrs                    2   \n",
       "3                 None                      1-4 yrs                    1   \n",
       "4                 None                        < 1yr                    4   \n",
       "\n",
       "  Guarantors  Most-valuable-available-asset  Age-years Concurrent-Credits  \\\n",
       "0       None                              1       33.0  Other Banks/Depts   \n",
       "1       None                              1       29.0  Other Banks/Depts   \n",
       "2       None                              1       42.0  Other Banks/Depts   \n",
       "3       None                              1       37.0  Other Banks/Depts   \n",
       "4       None                              1       27.0  Other Banks/Depts   \n",
       "\n",
       "   Type-of-apartment No-of-Credits-at-this-Bank  Occupation  No-of-dependents  \\\n",
       "0                  2                          1           1                 2   \n",
       "1                  2                          1           1                 2   \n",
       "2                  2                More than 1           1                 2   \n",
       "3                  2                          1           1                 2   \n",
       "4                  2                          1           1                 1   \n",
       "\n",
       "   Telephone  Foreign-Worker  \n",
       "0          1               2  \n",
       "1          1               2  \n",
       "2          1               1  \n",
       "3          1               1  \n",
       "4          1               1  "
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c6954a2-2be4-4f71-90cc-cb45d4be00bb",
   "metadata": {},
   "source": [
    "----"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0681441e-a007-4003-b686-3bc6a0c90b30",
   "metadata": {},
   "source": [
    "# How to Detect and Remove Outliers using Boxplot"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f54880c-b213-4a69-91b9-4ffad985526e",
   "metadata": {},
   "source": [
    "What is Outlier?\n",
    "> an outlier is a data point that differs significantly from other observations.\n",
    "\n",
    "> An outlier may be due to variability in the measurement or it may indicate experimental error\n",
    "\n",
    "> Outliers can occur by chance in any distribution, but they often indicate either measurement error or that the population has a heavy-tailed distribution."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "00386bad-13f3-48f0-8e55-10987d48f8f3",
   "metadata": {},
   "source": [
    "## Let's read and explore the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "3fe65176-ad36-4304-b54e-379da6101aa3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import some manipulation libraries\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "import seaborn as sns\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "1edbf6cc-d133-43e7-a816-72e2932ee0e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# read the dataset\n",
    "train_data = pd.read_csv(\"train.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "11e223bb-1215-423f-a420-9fccfe553938",
   "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>2010 Census Population</th>\n",
       "      <th>City</th>\n",
       "      <th>Land Area</th>\n",
       "      <th>Households with Under 18</th>\n",
       "      <th>Population Density</th>\n",
       "      <th>Total Families</th>\n",
       "      <th>Total Pawdacity Sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4585</td>\n",
       "      <td>Buffalo</td>\n",
       "      <td>3115.50750</td>\n",
       "      <td>746</td>\n",
       "      <td>1.55</td>\n",
       "      <td>1819.50</td>\n",
       "      <td>185328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>35316</td>\n",
       "      <td>Casper</td>\n",
       "      <td>3894.30910</td>\n",
       "      <td>7788</td>\n",
       "      <td>11.16</td>\n",
       "      <td>8756.32</td>\n",
       "      <td>317736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>59466</td>\n",
       "      <td>Cheyenne</td>\n",
       "      <td>1500.17840</td>\n",
       "      <td>7158</td>\n",
       "      <td>20.34</td>\n",
       "      <td>14612.64</td>\n",
       "      <td>917892</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9520</td>\n",
       "      <td>Cody</td>\n",
       "      <td>2998.95696</td>\n",
       "      <td>1403</td>\n",
       "      <td>1.82</td>\n",
       "      <td>3515.62</td>\n",
       "      <td>218376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6120</td>\n",
       "      <td>Douglas</td>\n",
       "      <td>1829.46510</td>\n",
       "      <td>832</td>\n",
       "      <td>1.46</td>\n",
       "      <td>1744.08</td>\n",
       "      <td>208008</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   2010 Census Population      City   Land Area  Households with Under 18  \\\n",
       "0                    4585   Buffalo  3115.50750                       746   \n",
       "1                   35316    Casper  3894.30910                      7788   \n",
       "2                   59466  Cheyenne  1500.17840                      7158   \n",
       "3                    9520      Cody  2998.95696                      1403   \n",
       "4                    6120   Douglas  1829.46510                       832   \n",
       "\n",
       "   Population Density  Total Families  Total Pawdacity Sales  \n",
       "0                1.55         1819.50                 185328  \n",
       "1               11.16         8756.32                 317736  \n",
       "2               20.34        14612.64                 917892  \n",
       "3                1.82         3515.62                 218376  \n",
       "4                1.46         1744.08                 208008  "
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# explore the dataset\n",
    "train_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "23801ed5-9568-495b-910d-bb411913b817",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2010 Census Population      0\n",
       "City                        0\n",
       "Land Area                   0\n",
       "Households with Under 18    0\n",
       "Population Density          0\n",
       "Total Families              0\n",
       "Total Pawdacity Sales       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the missing values\n",
    "train_data.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "6f275d04-9e4d-45f2-81d9-8fbd587b8770",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10, 7)"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# check the shape of the data\n",
    "train_data.shape\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "0f7842e6-363a-40c6-8ad4-e62201defe72",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['2010 Census Population', 'Land Area', 'Households with Under 18',\n",
       "       'Population Density', 'Total Families', 'Total Pawdacity Sales'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# get numerical columns\n",
    "num_cols = train_data.describe(exclude='object')\n",
    "col_names = num_cols.columns\n",
    "col_names"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "91ab0254-50e4-4ff4-b28c-fd36825301f4",
   "metadata": {},
   "source": [
    "## Let's Plot the boxplot to visualize the distribution of the data\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "8f4685b1-8d4e-46e2-801e-3983ed655e81",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1000x700 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\"\"\"\n",
    "solution\n",
    "\"\"\"\n",
    "fig = plt.figure(figsize=(10, 7))\n",
    "\n",
    "for i in range(6):\n",
    "    fig.tight_layout()\n",
    "    \n",
    "    grid_col = \"32{}\".format(i+1)\n",
    "    \n",
    "    ax = fig.add_subplot(int(grid_col))\n",
    "    \n",
    "    sns.boxplot(data=train_data, x=col_names[i], ax=ax)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "169dbcc1-2cfd-4e5a-a757-be72522ca75d",
   "metadata": {},
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed6df225-d686-40d8-8170-9b8212e036d6",
   "metadata": {},
   "source": [
    "# How to Detect and Remove Outliers using Interquantile Range(IQR)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9718949-da88-4c22-a106-35034caa1a47",
   "metadata": {},
   "source": [
    "## Read and Explore the dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "d91a127c-f657-4734-bdb7-b1adee1f6006",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import some manipulation libraries\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "8683e2ad-8152-47d1-abda-a864ec6973cb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# read the dataset\n",
    "train = pd.read_csv('../train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "decf14de-743c-4e6d-afa3-17ade7819639",
   "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>2010 Census Population</th>\n",
       "      <th>City</th>\n",
       "      <th>Land Area</th>\n",
       "      <th>Households with Under 18</th>\n",
       "      <th>Population Density</th>\n",
       "      <th>Total Families</th>\n",
       "      <th>Total Pawdacity Sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>4585</td>\n",
       "      <td>Buffalo</td>\n",
       "      <td>3115.50750</td>\n",
       "      <td>746</td>\n",
       "      <td>1.55</td>\n",
       "      <td>1819.50</td>\n",
       "      <td>185328</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>35316</td>\n",
       "      <td>Casper</td>\n",
       "      <td>3894.30910</td>\n",
       "      <td>7788</td>\n",
       "      <td>11.16</td>\n",
       "      <td>8756.32</td>\n",
       "      <td>317736</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>59466</td>\n",
       "      <td>Cheyenne</td>\n",
       "      <td>1500.17840</td>\n",
       "      <td>7158</td>\n",
       "      <td>20.34</td>\n",
       "      <td>14612.64</td>\n",
       "      <td>917892</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9520</td>\n",
       "      <td>Cody</td>\n",
       "      <td>2998.95696</td>\n",
       "      <td>1403</td>\n",
       "      <td>1.82</td>\n",
       "      <td>3515.62</td>\n",
       "      <td>218376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6120</td>\n",
       "      <td>Douglas</td>\n",
       "      <td>1829.46510</td>\n",
       "      <td>832</td>\n",
       "      <td>1.46</td>\n",
       "      <td>1744.08</td>\n",
       "      <td>208008</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   2010 Census Population      City   Land Area  Households with Under 18  \\\n",
       "0                    4585   Buffalo  3115.50750                       746   \n",
       "1                   35316    Casper  3894.30910                      7788   \n",
       "2                   59466  Cheyenne  1500.17840                      7158   \n",
       "3                    9520      Cody  2998.95696                      1403   \n",
       "4                    6120   Douglas  1829.46510                       832   \n",
       "\n",
       "   Population Density  Total Families  Total Pawdacity Sales  \n",
       "0                1.55         1819.50                 185328  \n",
       "1               11.16         8756.32                 317736  \n",
       "2               20.34        14612.64                 917892  \n",
       "3                1.82         3515.62                 218376  \n",
       "4                1.46         1744.08                 208008  "
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Explore the dataset\n",
    "train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "f589b880-e5b1-4ec6-b24c-b7a5fb22eb6f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2010 Census Population      0\n",
       "City                        0\n",
       "Land Area                   0\n",
       "Households with Under 18    0\n",
       "Population Density          0\n",
       "Total Families              0\n",
       "Total Pawdacity Sales       0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check missing values\n",
    "train.isnull().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "1adad5f4-acfd-4a19-ba13-c25286b3edf7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10, 7)"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check the shape of the dataset\n",
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "d92a79e3-d54c-41de-a0b3-76c58fdc8d16",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['2010 Census Population', 'Land Area', 'Households with Under 18',\n",
       "       'Population Density', 'Total Families', 'Total Pawdacity Sales'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the numerical columns\n",
    "num_cols = train.describe(exclude=\"object\")\n",
    "col_names = num_cols.columns\n",
    "col_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "fa980011-6cf6-474a-b6bd-d738c55c5a2f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the function to detect outlier\n",
    "def detect_outlier(df, cols):\n",
    "    \n",
    "    # create an empty dict\n",
    "    outlier_dict = {}\n",
    "    \n",
    "    for col in cols:\n",
    "        \n",
    "        # claculate q1, q3, iqr\n",
    "        \n",
    "        q1 = np.quantile(df[col], 0.25)\n",
    "        \n",
    "        q3 = np.quantile(df[col], 0.75)\n",
    "        \n",
    "        iqr = q3 - q1\n",
    "        \n",
    "        # find upper fence and lower fence\n",
    "        upper_fence = q3 + (1.5 * iqr)\n",
    "        \n",
    "        lower_fence = q1 - (1.5*iqr)\n",
    "        \n",
    "        \n",
    "        outlier = df.loc[((df[col]<=lower_fence)|(df[col]>=upper_fence))][[col]]\n",
    "        \n",
    "        outlier_dict[col] = [q1, q3, iqr, upper_fence, lower_fence, outlier.values]\n",
    "    \n",
    "    \n",
    "    return outlier_dict\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "203593b6-dc71-4545-9728-f02996cff10f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# call the function and create dataframe\n",
    "\n",
    "outlier = detect_outlier(train, col_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "56d0fb27-1e0d-43c1-9654-3341a0bbdd12",
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.DataFrame(outlier).set_index(pd.Index(['q1', 'q3', 'iqr', 'upper_fence', 'lower fence','outlier']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "e3f6857f-d103-4368-b6ab-b47b30f40df6",
   "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>2010 Census Population</th>\n",
       "      <th>Land Area</th>\n",
       "      <th>Households with Under 18</th>\n",
       "      <th>Population Density</th>\n",
       "      <th>Total Families</th>\n",
       "      <th>Total Pawdacity Sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>q1</th>\n",
       "      <td>7115.50</td>\n",
       "      <td>1845.593087</td>\n",
       "      <td>1289.00</td>\n",
       "      <td>1.67000</td>\n",
       "      <td>2818.02500</td>\n",
       "      <td>222264.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>q3</th>\n",
       "      <td>21638.00</td>\n",
       "      <td>3699.608700</td>\n",
       "      <td>3686.50</td>\n",
       "      <td>7.97250</td>\n",
       "      <td>7189.06250</td>\n",
       "      <td>306990.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>iqr</th>\n",
       "      <td>14522.50</td>\n",
       "      <td>1854.015613</td>\n",
       "      <td>2397.50</td>\n",
       "      <td>6.30250</td>\n",
       "      <td>4371.03750</td>\n",
       "      <td>84726.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>upper_fence</th>\n",
       "      <td>43421.75</td>\n",
       "      <td>6480.632120</td>\n",
       "      <td>7282.75</td>\n",
       "      <td>17.42625</td>\n",
       "      <td>13745.61875</td>\n",
       "      <td>434079.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>lower fence</th>\n",
       "      <td>-14668.25</td>\n",
       "      <td>-935.430332</td>\n",
       "      <td>-2307.25</td>\n",
       "      <td>-7.78375</td>\n",
       "      <td>-3738.53125</td>\n",
       "      <td>95175.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>outlier</th>\n",
       "      <td>[[59466]]</td>\n",
       "      <td>[[6620.201916]]</td>\n",
       "      <td>[[7788]]</td>\n",
       "      <td>[[20.34]]</td>\n",
       "      <td>[[14612.64]]</td>\n",
       "      <td>[[917892]]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            2010 Census Population        Land Area Households with Under 18  \\\n",
       "q1                         7115.50      1845.593087                  1289.00   \n",
       "q3                        21638.00      3699.608700                  3686.50   \n",
       "iqr                       14522.50      1854.015613                  2397.50   \n",
       "upper_fence               43421.75      6480.632120                  7282.75   \n",
       "lower fence              -14668.25      -935.430332                 -2307.25   \n",
       "outlier                  [[59466]]  [[6620.201916]]                 [[7788]]   \n",
       "\n",
       "            Population Density Total Families Total Pawdacity Sales  \n",
       "q1                     1.67000     2818.02500              222264.0  \n",
       "q3                     7.97250     7189.06250              306990.0  \n",
       "iqr                    6.30250     4371.03750               84726.0  \n",
       "upper_fence           17.42625    13745.61875              434079.0  \n",
       "lower fence           -7.78375    -3738.53125               95175.0  \n",
       "outlier              [[20.34]]   [[14612.64]]            [[917892]]  "
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e13a5f47-51cc-4012-a662-e8c14c760772",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.9.10"
  },
  "widgets": {
   "application/vnd.jupyter.widget-state+json": {
    "state": {},
    "version_major": 2,
    "version_minor": 0
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}