data-resampling-and-splitting.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# References\n",
    "- [Handling imbalanced data (kdnuggets)](https://www.kdnuggets.com/2017/06/7-techniques-handle-imbalanced-data.html)\n",
    "- [Resampling strategies for imbalanced datasets (Kaggle)](https://www.kaggle.com/rafjaa/resampling-strategies-for-imbalanced-datasets)\n",
    "- [Handling Imbalanced Data (TensorFlow)](https://www.tensorflow.org/tutorials/structured_data/imbalanced_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For data processing\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "# For data visualization\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sb\n",
    "plt.style.use('fivethirtyeight') \n",
    "\n",
    "# For data splitting\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# For handling imbalanced dataset\n",
    "from imblearn.combine import SMOTETomek\n",
    "from imblearn.combine import SMOTEENN\n",
    "from imblearn.over_sampling import RandomOverSampler\n",
    "from imblearn.under_sampling import RandomUnderSampler"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Splitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>decade_released</th>\n",
       "      <th>genres</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>vote_average</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>283.822</td>\n",
       "      <td>755.450</td>\n",
       "      <td>473.269</td>\n",
       "      <td>356000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Adventure', 'Science Fiction', 'Action']</td>\n",
       "      <td>2.797801e+09</td>\n",
       "      <td>17799</td>\n",
       "      <td>8.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>102.112</td>\n",
       "      <td>108.804</td>\n",
       "      <td>878.826</td>\n",
       "      <td>237000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Action', 'Adventure', 'Fantasy', 'Science Fi...</td>\n",
       "      <td>2.787965e+09</td>\n",
       "      <td>23178</td>\n",
       "      <td>7.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>86.808</td>\n",
       "      <td>201.581</td>\n",
       "      <td>122.649</td>\n",
       "      <td>200000000.0</td>\n",
       "      <td>1990s</td>\n",
       "      <td>['Drama', 'Romance']</td>\n",
       "      <td>2.187464e+09</td>\n",
       "      <td>18982</td>\n",
       "      <td>7.9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>53.930</td>\n",
       "      <td>391.303</td>\n",
       "      <td>235.947</td>\n",
       "      <td>245000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Action', 'Adventure', 'Science Fiction', 'Fa...</td>\n",
       "      <td>2.068224e+09</td>\n",
       "      <td>15675</td>\n",
       "      <td>7.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>299.524</td>\n",
       "      <td>528.712</td>\n",
       "      <td>501.005</td>\n",
       "      <td>300000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Adventure', 'Action', 'Science Fiction']</td>\n",
       "      <td>2.046240e+09</td>\n",
       "      <td>21489</td>\n",
       "      <td>8.3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  casts_popularity_sum  crews_popularity_sum       budget  \\\n",
       "0     283.822               755.450               473.269  356000000.0   \n",
       "1     102.112               108.804               878.826  237000000.0   \n",
       "2      86.808               201.581               122.649  200000000.0   \n",
       "3      53.930               391.303               235.947  245000000.0   \n",
       "4     299.524               528.712               501.005  300000000.0   \n",
       "\n",
       "  decade_released                                             genres  \\\n",
       "0           2010s         ['Adventure', 'Science Fiction', 'Action']   \n",
       "1           2000s  ['Action', 'Adventure', 'Fantasy', 'Science Fi...   \n",
       "2           1990s                               ['Drama', 'Romance']   \n",
       "3           2010s  ['Action', 'Adventure', 'Science Fiction', 'Fa...   \n",
       "4           2010s         ['Adventure', 'Action', 'Science Fiction']   \n",
       "\n",
       "        revenue  vote_count  vote_average  \n",
       "0  2.797801e+09       17799           8.3  \n",
       "1  2.787965e+09       23178           7.5  \n",
       "2  2.187464e+09       18982           7.9  \n",
       "3  2.068224e+09       15675           7.4  \n",
       "4  2.046240e+09       21489           8.3  "
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select several columns\n",
    "cols = ['popularity', 'original_language', 'casts_popularity_sum',\n",
    "        'crews_popularity_sum', 'genres', 'decade_released', \n",
    "        'budget', 'revenue', 'vote_count', 'vote_average']\n",
    "\n",
    "df = pd.read_csv('cleaned-movie-dataset.csv', index_col=0)[cols]\n",
    "\n",
    "# Create a new column 'in_en', 1 if the original language is in English and 0 otherwise\n",
    "df['in_en'] = (df['original_language'] == 'en').astype('int64')\n",
    "\n",
    "df = df.drop(columns=['original_language'])\n",
    "\n",
    "# Rearrange columns\n",
    "df = df[['popularity', 'casts_popularity_sum', 'crews_popularity_sum', \n",
    "         'budget', 'decade_released', 'genres', \n",
    "         'revenue', 'vote_count', 'vote_average']]\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the DF into predictors and response\n",
    "X, y = df.iloc[:, :-1], df.iloc[:, -1]\n",
    "\n",
    "# Define good as vote_average >= 7.2 (80% percentile)\n",
    "y = (y >= 7.2).astype('int64')\n",
    "\n",
    "# Split the dataset into 70% train and 30% test set\n",
    "# Set random_state for reproducibility\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.7, random_state=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Good vs Not Good Movies Train Set Proportion\n",
      "1018 : 3931\n"
     ]
    }
   ],
   "source": [
    "print(\"Good vs Not Good Movies Train Set Proportion\")\n",
    "print(y_train.value_counts()[1], ':', y_train.value_counts()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>decade_released</th>\n",
       "      <th>genres</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>is_good</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5527</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Comedy', 'Action', 'Drama']</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3263</th>\n",
       "      <td>18.013</td>\n",
       "      <td>46.018</td>\n",
       "      <td>9.456</td>\n",
       "      <td>26000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Thriller', 'Adventure', 'Mystery', 'Fantasy'...</td>\n",
       "      <td>34592118.0</td>\n",
       "      <td>200</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1022</th>\n",
       "      <td>17.237</td>\n",
       "      <td>182.113</td>\n",
       "      <td>198.986</td>\n",
       "      <td>65000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Drama', 'History', 'Thriller']</td>\n",
       "      <td>163247198.0</td>\n",
       "      <td>1122</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4603</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Action', 'Drama', 'Thriller']</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2733</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Comedy', 'Family']</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "5527       1.680                34.059                 8.671   3300000.0   \n",
       "3263      18.013                46.018                 9.456  26000000.0   \n",
       "1022      17.237               182.113               198.986  65000000.0   \n",
       "4603      11.931               110.585                44.527  20000000.0   \n",
       "2733      14.717                99.835                75.898  51000000.0   \n",
       "\n",
       "     decade_released                                             genres  \\\n",
       "5527           2010s                      ['Comedy', 'Action', 'Drama']   \n",
       "3263           2010s  ['Thriller', 'Adventure', 'Mystery', 'Fantasy'...   \n",
       "1022           2000s                   ['Drama', 'History', 'Thriller']   \n",
       "4603           2000s                    ['Action', 'Drama', 'Thriller']   \n",
       "2733           2000s                               ['Comedy', 'Family']   \n",
       "\n",
       "          revenue  vote_count  is_good  \n",
       "5527   10000000.0           8        0  \n",
       "3263   34592118.0         200        0  \n",
       "1022  163247198.0        1122        0  \n",
       "4603   16605763.0         129        0  \n",
       "2733   47231070.0         405        0  "
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df = pd.concat([X_train, y_train], axis=1).rename(columns={'vote_average': 'is_good'})\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train_df.to_csv('train.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Oversampling on training data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define oversampling strategy (creates 1:2 ratio)\n",
    "over_smp = RandomOverSampler(sampling_strategy=0.5, random_state=20)\n",
    "\n",
    "# Fit and apply the transform\n",
    "X_res, y_res = over_smp.fit_resample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Good vs Not Good Movies Train Set Proportion (After oversampling)\n",
      "1965 : 3931\n"
     ]
    }
   ],
   "source": [
    "print(\"Good vs Not Good Movies Train Set Proportion (After oversampling)\")\n",
    "print(y_res.value_counts()[1], ':', y_res.value_counts()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>decade_released</th>\n",
       "      <th>genres</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>is_good</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Comedy', 'Action', 'Drama']</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>18.013</td>\n",
       "      <td>46.018</td>\n",
       "      <td>9.456</td>\n",
       "      <td>26000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Thriller', 'Adventure', 'Mystery', 'Fantasy'...</td>\n",
       "      <td>34592118.0</td>\n",
       "      <td>200</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>17.237</td>\n",
       "      <td>182.113</td>\n",
       "      <td>198.986</td>\n",
       "      <td>65000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Drama', 'History', 'Thriller']</td>\n",
       "      <td>163247198.0</td>\n",
       "      <td>1122</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Action', 'Drama', 'Thriller']</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Comedy', 'Family']</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "0       1.680                34.059                 8.671   3300000.0   \n",
       "1      18.013                46.018                 9.456  26000000.0   \n",
       "2      17.237               182.113               198.986  65000000.0   \n",
       "3      11.931               110.585                44.527  20000000.0   \n",
       "4      14.717                99.835                75.898  51000000.0   \n",
       "\n",
       "  decade_released                                             genres  \\\n",
       "0           2010s                      ['Comedy', 'Action', 'Drama']   \n",
       "1           2010s  ['Thriller', 'Adventure', 'Mystery', 'Fantasy'...   \n",
       "2           2000s                   ['Drama', 'History', 'Thriller']   \n",
       "3           2000s                    ['Action', 'Drama', 'Thriller']   \n",
       "4           2000s                               ['Comedy', 'Family']   \n",
       "\n",
       "       revenue  vote_count  is_good  \n",
       "0   10000000.0           8        0  \n",
       "1   34592118.0         200        0  \n",
       "2  163247198.0        1122        0  \n",
       "3   16605763.0         129        0  \n",
       "4   47231070.0         405        0  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Combine the X and y to a single DF and rename column\n",
    "oversampled_train_df = pd.concat([X_res, y_res], axis=1).rename(columns={'vote_average': 'is_good'})\n",
    "oversampled_train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save as csv\n",
    "# oversampled_train_df.to_csv('oversampled-train.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SMOTETomek Resampling on training data\n",
    "This only applies for numerical variables."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5527</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3263</th>\n",
       "      <td>18.013</td>\n",
       "      <td>46.018</td>\n",
       "      <td>9.456</td>\n",
       "      <td>26000000.0</td>\n",
       "      <td>34592118.0</td>\n",
       "      <td>200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1022</th>\n",
       "      <td>17.237</td>\n",
       "      <td>182.113</td>\n",
       "      <td>198.986</td>\n",
       "      <td>65000000.0</td>\n",
       "      <td>163247198.0</td>\n",
       "      <td>1122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4603</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2733</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "5527       1.680                34.059                 8.671   3300000.0   \n",
       "3263      18.013                46.018                 9.456  26000000.0   \n",
       "1022      17.237               182.113               198.986  65000000.0   \n",
       "4603      11.931               110.585                44.527  20000000.0   \n",
       "2733      14.717                99.835                75.898  51000000.0   \n",
       "\n",
       "          revenue  vote_count  \n",
       "5527   10000000.0           8  \n",
       "3263   34592118.0         200  \n",
       "1022  163247198.0        1122  \n",
       "4603   16605763.0         129  \n",
       "2733   47231070.0         405  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select numerical variables only\n",
    "X_train = X_train[['popularity', 'casts_popularity_sum',\n",
    "                   'crews_popularity_sum', 'budget',\n",
    "                   'revenue', 'vote_count']]\n",
    "X_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set random_state for reproducibility\n",
    "smt = SMOTETomek(random_state=20)\n",
    "X_res, y_res = smt.fit_resample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Good vs Not Good Movies Train Set Proportion (After SMOTETomek resampling)\n",
      "3431 : 3431\n"
     ]
    }
   ],
   "source": [
    "print(\"Good vs Not Good Movies Train Set Proportion (After SMOTETomek resampling)\")\n",
    "print(y_res.value_counts()[1], ':', y_res.value_counts()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>is_good</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>18.013</td>\n",
       "      <td>46.018</td>\n",
       "      <td>9.456</td>\n",
       "      <td>26000000.0</td>\n",
       "      <td>34592118.0</td>\n",
       "      <td>200</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>23.357</td>\n",
       "      <td>99.896</td>\n",
       "      <td>62.278</td>\n",
       "      <td>32000000.0</td>\n",
       "      <td>83615414.0</td>\n",
       "      <td>897</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "0       1.680                34.059                 8.671   3300000.0   \n",
       "1      18.013                46.018                 9.456  26000000.0   \n",
       "2      11.931               110.585                44.527  20000000.0   \n",
       "3      14.717                99.835                75.898  51000000.0   \n",
       "4      23.357                99.896                62.278  32000000.0   \n",
       "\n",
       "      revenue  vote_count  is_good  \n",
       "0  10000000.0           8        0  \n",
       "1  34592118.0         200        0  \n",
       "2  16605763.0         129        0  \n",
       "3  47231070.0         405        0  \n",
       "4  83615414.0         897        0  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Combine the X and y to a single DF and rename column\n",
    "smotetomek_train_df = pd.concat([X_res, y_res], axis=1).rename(columns={'vote_average': 'is_good'})\n",
    "smotetomek_train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save as csv\n",
    "# smotetomek_train_df.to_csv('smotetomek-train.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SMOTEENN Resampling on training data\n",
    "Similar with SMOTETomek resampling, this only applies for numerical variables."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5527</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3263</th>\n",
       "      <td>18.013</td>\n",
       "      <td>46.018</td>\n",
       "      <td>9.456</td>\n",
       "      <td>26000000.0</td>\n",
       "      <td>34592118.0</td>\n",
       "      <td>200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1022</th>\n",
       "      <td>17.237</td>\n",
       "      <td>182.113</td>\n",
       "      <td>198.986</td>\n",
       "      <td>65000000.0</td>\n",
       "      <td>163247198.0</td>\n",
       "      <td>1122</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4603</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2733</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "5527       1.680                34.059                 8.671   3300000.0   \n",
       "3263      18.013                46.018                 9.456  26000000.0   \n",
       "1022      17.237               182.113               198.986  65000000.0   \n",
       "4603      11.931               110.585                44.527  20000000.0   \n",
       "2733      14.717                99.835                75.898  51000000.0   \n",
       "\n",
       "          revenue  vote_count  \n",
       "5527   10000000.0           8  \n",
       "3263   34592118.0         200  \n",
       "1022  163247198.0        1122  \n",
       "4603   16605763.0         129  \n",
       "2733   47231070.0         405  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train = X_train[['popularity', 'casts_popularity_sum',\n",
    "                   'crews_popularity_sum', 'budget',\n",
    "                   'revenue', 'vote_count']]\n",
    "X_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set random_state for reproducibility\n",
    "sme = SMOTEENN(random_state=20)\n",
    "X_res, y_res = sme.fit_resample(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Good vs Not Good Movies Train Set Proportion (After SMOTEENN resampling)\n",
      "1939 : 1756\n"
     ]
    }
   ],
   "source": [
    "print(\"Good vs Not Good Movies Train Set Proportion (After SMOTEENN resampling)\")\n",
    "print(y_res.value_counts()[1], ':', y_res.value_counts()[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>is_good</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.680</td>\n",
       "      <td>34.059</td>\n",
       "      <td>8.671</td>\n",
       "      <td>3300000.0</td>\n",
       "      <td>10000000.0</td>\n",
       "      <td>8</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>11.931</td>\n",
       "      <td>110.585</td>\n",
       "      <td>44.527</td>\n",
       "      <td>20000000.0</td>\n",
       "      <td>16605763.0</td>\n",
       "      <td>129</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>14.717</td>\n",
       "      <td>99.835</td>\n",
       "      <td>75.898</td>\n",
       "      <td>51000000.0</td>\n",
       "      <td>47231070.0</td>\n",
       "      <td>405</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>12.050</td>\n",
       "      <td>33.433</td>\n",
       "      <td>10.263</td>\n",
       "      <td>17500000.0</td>\n",
       "      <td>25024919.0</td>\n",
       "      <td>380</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>7.503</td>\n",
       "      <td>36.672</td>\n",
       "      <td>13.520</td>\n",
       "      <td>160000.0</td>\n",
       "      <td>4000000.0</td>\n",
       "      <td>51</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   popularity  casts_popularity_sum  crews_popularity_sum      budget  \\\n",
       "0       1.680                34.059                 8.671   3300000.0   \n",
       "1      11.931               110.585                44.527  20000000.0   \n",
       "2      14.717                99.835                75.898  51000000.0   \n",
       "3      12.050                33.433                10.263  17500000.0   \n",
       "4       7.503                36.672                13.520    160000.0   \n",
       "\n",
       "      revenue  vote_count  is_good  \n",
       "0  10000000.0           8        0  \n",
       "1  16605763.0         129        0  \n",
       "2  47231070.0         405        0  \n",
       "3  25024919.0         380        0  \n",
       "4   4000000.0          51        0  "
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Combine the X and y to a single DF and rename column\n",
    "smoteenn_train_df = pd.concat([X_res, y_res], axis=1).rename(columns={'vote_average': 'is_good'})\n",
    "smoteenn_train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save as csv\n",
    "# smoteenn_train_df.to_csv('smoteenn-train.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save test data as csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>popularity</th>\n",
       "      <th>casts_popularity_sum</th>\n",
       "      <th>crews_popularity_sum</th>\n",
       "      <th>budget</th>\n",
       "      <th>decade_released</th>\n",
       "      <th>genres</th>\n",
       "      <th>revenue</th>\n",
       "      <th>vote_count</th>\n",
       "      <th>is_good</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>6183</th>\n",
       "      <td>10.639</td>\n",
       "      <td>53.101</td>\n",
       "      <td>21.680</td>\n",
       "      <td>5800000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Action', 'Adventure', 'Drama']</td>\n",
       "      <td>7000000.0</td>\n",
       "      <td>28</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>786</th>\n",
       "      <td>48.052</td>\n",
       "      <td>100.519</td>\n",
       "      <td>97.495</td>\n",
       "      <td>60000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Action', 'Thriller']</td>\n",
       "      <td>205754447.0</td>\n",
       "      <td>3573</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>37.487</td>\n",
       "      <td>255.221</td>\n",
       "      <td>181.062</td>\n",
       "      <td>170000000.0</td>\n",
       "      <td>2010s</td>\n",
       "      <td>['Action', 'Science Fiction', 'Adventure']</td>\n",
       "      <td>772776600.0</td>\n",
       "      <td>22271</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4763</th>\n",
       "      <td>19.589</td>\n",
       "      <td>83.531</td>\n",
       "      <td>99.984</td>\n",
       "      <td>80000000.0</td>\n",
       "      <td>2000s</td>\n",
       "      <td>['Science Fiction', 'Thriller', 'Mystery']</td>\n",
       "      <td>15071514.0</td>\n",
       "      <td>989</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>429</th>\n",
       "      <td>27.201</td>\n",
       "      <td>121.827</td>\n",
       "      <td>136.950</td>\n",
       "      <td>40000000.0</td>\n",
       "      <td>1980s</td>\n",
       "      <td>['Adventure', 'Comedy', 'Family', 'Science Fic...</td>\n",
       "      <td>332000000.0</td>\n",
       "      <td>9474</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      popularity  casts_popularity_sum  crews_popularity_sum       budget  \\\n",
       "6183      10.639                53.101                21.680    5800000.0   \n",
       "786       48.052               100.519                97.495   60000000.0   \n",
       "100       37.487               255.221               181.062  170000000.0   \n",
       "4763      19.589                83.531                99.984   80000000.0   \n",
       "429       27.201               121.827               136.950   40000000.0   \n",
       "\n",
       "     decade_released                                             genres  \\\n",
       "6183           2000s                   ['Action', 'Adventure', 'Drama']   \n",
       "786            2010s                             ['Action', 'Thriller']   \n",
       "100            2010s         ['Action', 'Science Fiction', 'Adventure']   \n",
       "4763           2000s         ['Science Fiction', 'Thriller', 'Mystery']   \n",
       "429            1980s  ['Adventure', 'Comedy', 'Family', 'Science Fic...   \n",
       "\n",
       "          revenue  vote_count  is_good  \n",
       "6183    7000000.0          28        0  \n",
       "786   205754447.0        3573        0  \n",
       "100   772776600.0       22271        1  \n",
       "4763   15071514.0         989        0  \n",
       "429   332000000.0        9474        1  "
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Combine the X and y to a single DF and rename column\n",
    "test_df = pd.concat([X_test, y_test], axis=1).rename(columns={'vote_average': 'is_good'})\n",
    "test_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save as csv\n",
    "# test_df.to_csv('test.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualisation of resampled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "oversampled_train_df = pd.read_csv('oversampled-train.csv',index_col= 0)\n",
    "smoteenn_train_df = pd.read_csv('smoteenn-train.csv',index_col= 0)\n",
    "smotetomek_train_df = pd.read_csv('smotetomek-train.csv',index_col= 0)\n",
    "train_df = pd.read_csv('train.csv',index_col=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0, 'is_good - SMOTEENN resampling')"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1728x576 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes =plt.subplots(1,4,figsize=(24,8))\n",
    "plt.setp(axes,ylim=(0,4000))\n",
    "sb.countplot(x=train_df['is_good'], ax=axes[0])\n",
    "ax = axes[0]\n",
    "ax.set_xlabel(\"is_good - Original\",fontsize=15)\n",
    "sb.countplot(x=oversampled_train_df['is_good'], ax=axes[1])\n",
    "ax = axes[1]\n",
    "ax.set_xlabel(\"is_good - Oversampling\",fontsize=15)\n",
    "sb.countplot(x=smotetomek_train_df['is_good'], ax=axes[2])\n",
    "ax = axes[2]\n",
    "ax.set_xlabel(\"is_good - SMOTETomek resampling\",fontsize=15)\n",
    "sb.countplot(x=smoteenn_train_df['is_good'], ax=axes[3])\n",
    "ax = axes[3]\n",
    "ax.set_xlabel(\"is_good - SMOTEENN resampling\",fontsize=15)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Therefore, we can now see that the ratio between good vs not good movies is more balanced with the resampled data vs the original data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}