in-vehicle coupon recommendation Data Set_RF_06.05.2022.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "05e0e0cd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(12529, 59) (12529,)\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from numpy.random import RandomState\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm import tqdm\n",
    "\n",
    "seed = 42\n",
    "rng = RandomState(seed)\n",
    "\n",
    "df = pd.read_csv('Model.csv',index_col = False)\n",
    "\n",
    "X, y = df.iloc[:, 1:-1], df.iloc[:, -1]\n",
    "print(X.shape, y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "0255ef64",
   "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>destination_Home</th>\n",
       "      <th>destination_No Urgent Place</th>\n",
       "      <th>destination_Work</th>\n",
       "      <th>passanger_Alone</th>\n",
       "      <th>passanger_Friend(s)</th>\n",
       "      <th>passanger_Kid(s)</th>\n",
       "      <th>passanger_Partner</th>\n",
       "      <th>weather_Rainy</th>\n",
       "      <th>weather_Snowy</th>\n",
       "      <th>weather_Sunny</th>\n",
       "      <th>...</th>\n",
       "      <th>Restaurant20To50</th>\n",
       "      <th>income</th>\n",
       "      <th>education</th>\n",
       "      <th>age</th>\n",
       "      <th>time</th>\n",
       "      <th>expiration</th>\n",
       "      <th>driving_distance</th>\n",
       "      <th>gender</th>\n",
       "      <th>has_children</th>\n",
       "      <th>direction_opp</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12524</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12525</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12526</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12527</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12528</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>12529 rows × 59 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       destination_Home  destination_No Urgent Place  destination_Work  \\\n",
       "0                     0                            1                 0   \n",
       "1                     0                            1                 0   \n",
       "2                     0                            1                 0   \n",
       "3                     0                            1                 0   \n",
       "4                     0                            1                 0   \n",
       "...                 ...                          ...               ...   \n",
       "12524                 1                            0                 0   \n",
       "12525                 0                            0                 1   \n",
       "12526                 0                            0                 1   \n",
       "12527                 0                            0                 1   \n",
       "12528                 0                            0                 1   \n",
       "\n",
       "       passanger_Alone  passanger_Friend(s)  passanger_Kid(s)  \\\n",
       "0                    1                    0                 0   \n",
       "1                    0                    1                 0   \n",
       "2                    0                    1                 0   \n",
       "3                    0                    1                 0   \n",
       "4                    0                    1                 0   \n",
       "...                ...                  ...               ...   \n",
       "12524                0                    0                 0   \n",
       "12525                1                    0                 0   \n",
       "12526                1                    0                 0   \n",
       "12527                1                    0                 0   \n",
       "12528                1                    0                 0   \n",
       "\n",
       "       passanger_Partner  weather_Rainy  weather_Snowy  weather_Sunny  ...  \\\n",
       "0                      0              0              0              1  ...   \n",
       "1                      0              0              0              1  ...   \n",
       "2                      0              0              0              1  ...   \n",
       "3                      0              0              0              1  ...   \n",
       "4                      0              0              0              1  ...   \n",
       "...                  ...            ...            ...            ...  ...   \n",
       "12524                  1              1              0              0  ...   \n",
       "12525                  0              1              0              0  ...   \n",
       "12526                  0              0              1              0  ...   \n",
       "12527                  0              0              1              0  ...   \n",
       "12528                  0              0              0              1  ...   \n",
       "\n",
       "       Restaurant20To50  income  education  age  time  expiration  \\\n",
       "0                     2       3          2    1     1           1   \n",
       "1                     2       3          2    1     0           0   \n",
       "2                     2       3          2    1     0           0   \n",
       "3                     2       3          2    1     1           0   \n",
       "4                     2       3          2    1     1           1   \n",
       "...                 ...     ...        ...  ...   ...         ...   \n",
       "12524                 2       6          4    2     1           1   \n",
       "12525                 2       6          4    2     0           1   \n",
       "12526                 2       6          4    2     0           1   \n",
       "12527                 2       6          4    2     0           1   \n",
       "12528                 2       6          4    2     0           0   \n",
       "\n",
       "       driving_distance  gender  has_children  direction_opp  \n",
       "0                     0       0             1              1  \n",
       "1                     0       0             1              1  \n",
       "2                     1       0             1              1  \n",
       "3                     1       0             1              1  \n",
       "4                     1       0             1              1  \n",
       "...                 ...     ...           ...            ...  \n",
       "12524                 0       1             0              0  \n",
       "12525                 0       1             0              1  \n",
       "12526                 0       1             0              0  \n",
       "12527                 2       1             0              1  \n",
       "12528                 0       1             0              0  \n",
       "\n",
       "[12529 rows x 59 columns]"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "c853fac0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        1\n",
       "1        0\n",
       "2        1\n",
       "3        0\n",
       "4        0\n",
       "        ..\n",
       "12524    1\n",
       "12525    1\n",
       "12526    0\n",
       "12527    0\n",
       "12528    0\n",
       "Name: Y, Length: 12529, dtype: int64"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e1e3fc54",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train set:  (10023, 59) (10023,)\n",
      "Test set:  (2506, 59) (2506,)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split, cross_val_score\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y,\n",
    "                                                    test_size=0.2, \n",
    "                                                    random_state=seed)\n",
    "print('Train set: ', X_train.shape, y_train.shape)\n",
    "print('Test set: ', X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "0d307a08",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:26<00:00, 13.11s/it]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "title = 'Change criterion, n_estimators=50'\n",
    "xlabel = 'Criterion'\n",
    "criterions = ['gini', 'entropy']\n",
    "means = []\n",
    "stddevs = []\n",
    "\n",
    "for criterion in tqdm(criterions):\n",
    "    clf = RandomForestClassifier(criterion=criterion, n_estimators=50, \n",
    "                                 random_state=rng)\n",
    "    mean, std = cross_validation_score(clf)\n",
    "    means.append(mean)\n",
    "    stddevs.append(std)\n",
    "    \n",
    "plot_cross_validation(title, criterions, means, stddevs, xlabel)\n",
    "plt.savefig('RF_change_criterion.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "da586993",
   "metadata": {},
   "outputs": [],
   "source": [
    "def cross_validation_score(estimator):\n",
    "    scores = cross_val_score(\n",
    "        estimator, X_train, y_train, cv=10, n_jobs=-1,\n",
    "        scoring='roc_auc'\n",
    "    )\n",
    "    \n",
    "    return scores.mean(), scores.std()\n",
    "\n",
    "\n",
    "def plot_cross_validation(title, X, Y, error, xlabel, ylabel='ROC AUC'):\n",
    "    plt.title(title)\n",
    "    plt.grid()\n",
    "    plt.xlabel(xlabel)\n",
    "    plt.ylabel(ylabel)\n",
    "    \n",
    "    plt.errorbar(X, Y, error, linestyle='None', marker='o')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "de03e4d6",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:53<00:00, 10.61s/it]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "title = 'Change n_estimators, criterion=entropy'\n",
    "xlabel = 'n_estimators'\n",
    "N = [10, 50, 100, 150, 300]\n",
    "means = []\n",
    "stddevs = []\n",
    "\n",
    "for n in tqdm(N):\n",
    "    clf = RandomForestClassifier(criterion='entropy', n_estimators=n, \n",
    "                                 random_state=rng)\n",
    "    mean, std = cross_validation_score(clf)\n",
    "    means.append(mean)\n",
    "    stddevs.append(std)\n",
    "\n",
    "plot_cross_validation(title, N, means, stddevs, xlabel)\n",
    "plt.savefig('RF_change_n_estimators.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "614944f5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test set:  0.7545889864325619\n",
      "ROC AUC on test set:  0.7452392969922865\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import roc_auc_score, accuracy_score\n",
    "\n",
    "rf = RandomForestClassifier(criterion='entropy', n_estimators=100, \n",
    "                            random_state=rng)\n",
    "rf.fit(X_train, y_train)\n",
    "y_pred = rf.predict(X_test)\n",
    "\n",
    "print('Accuracy on test set: ', accuracy_score(y_test, y_pred))\n",
    "print('ROC AUC on test set: ', roc_auc_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "119e7191",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.metrics import plot_confusion_matrix\n",
    "\n",
    "plot_confusion_matrix(rf, X_test, y_test)\n",
    "plt.savefig('RF_confmat.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ba8dbe76",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Visualize the impurity-based importance\n",
    "\n",
    "importances = rf.feature_importances_\n",
    "std = np.std([tree.feature_importances_ for tree in rf.estimators_], \n",
    "             axis=0)\n",
    "\n",
    "# Only inspect top-k most important features\n",
    "k = 10\n",
    "topk = importances.argsort()[-k:][::-1]\n",
    "\n",
    "feature_names = np.array(X.columns)[topk]\n",
    "importances = importances[topk]\n",
    "std = std[topk]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "f48cdfae",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "title = 'Top-%d feature importances using MDI' % (k)\n",
    "ylabel = 'Mean decrease in impurity'\n",
    "\n",
    "plt.title(title)\n",
    "plt.ylabel(ylabel)\n",
    "plt.xticks(rotation=45)\n",
    "plt.grid()\n",
    "\n",
    "plt.errorbar(feature_names, importances, std, linestyle='None', marker='o')\n",
    "plt.savefig('RF_top%d_MDI_importances.png' % (k))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1d590020",
   "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}