7a). Modélisation_1 (CNN Aléatoire).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "01a4b76d-59d4-4f19-9608-86bc9e720c98",
   "metadata": {
    "id": "01a4b76d-59d4-4f19-9608-86bc9e720c98"
   },
   "source": [
    "**1. IMPORTER LES LIBRAIRIES**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "0b3ebafe-aaa4-4553-8e91-9e911284678c",
   "metadata": {
    "id": "0b3ebafe-aaa4-4553-8e91-9e911284678c",
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# pip install tensorflow==2.12.0 & keras==2.12.0\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import re\n",
    "\n",
    "# Train_Test_Split\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "# Deep Learning\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.utils import pad_sequences\n",
    "from keras.initializers import Constant\n",
    "from keras.models import Model\n",
    "from keras.layers import *\n",
    "from tensorflow.python.keras import regularizers\n",
    "\n",
    "# Callbacks\n",
    "from tensorflow.keras.callbacks import ReduceLROnPlateau, EarlyStopping\n",
    "\n",
    "# Performances\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score\n",
    "\n",
    "# Enregistrer le modèle\n",
    "from sklearn import model_selection\n",
    "import joblib\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c67b5971-6763-4ff1-b875-485191e73032",
   "metadata": {},
   "source": [
    "**2. IMPORTER LE FICHIER**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "2b66b02d-8a3a-452c-bea7-59931d5cb015",
   "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>commentaires</th>\n",
       "      <th>notes</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>commandee simple et rapide commande simple et ...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>rapide et simple tres bon service</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Honorable bon suivi !</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Cdiscount cautionne l'escroquerie en ce… Cdisc...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Simple Simple, rapide, prix raisonnables et €c...</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                        commentaires  notes\n",
       "0  commandee simple et rapide commande simple et ...      4\n",
       "1                  rapide et simple tres bon service      4\n",
       "2                              Honorable bon suivi !      4\n",
       "3  Cdiscount cautionne l'escroquerie en ce… Cdisc...      0\n",
       "4  Simple Simple, rapide, prix raisonnables et €c...      4"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(\"2. Pre-processing & Feature engineering_1.csv\")\n",
    "df = df[['commentaires', 'notes']]\n",
    "\n",
    "# Remplace les notes 1,2,3,4,5 par 0,1,2,3,4\n",
    "df['notes'] = df['notes'].replace((1, 2, 3, 4, 5), (0, 1, 2, 3, 4))\n",
    "\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b2d95529-39bd-49c5-9dee-1b497d98f831",
   "metadata": {
    "id": "b2d95529-39bd-49c5-9dee-1b497d98f831"
   },
   "source": [
    "**3. PRE-PROCESSING & FEATURE ENGINEERING_2**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "14d6a7a1-0a7f-4903-8eb9-8d38bd437142",
   "metadata": {},
   "outputs": [],
   "source": [
    "#############################\n",
    "### Nettoyage des données ###\n",
    "#############################\n",
    "def clean_str(commentaire):\n",
    "    commentaire = str(commentaire)\n",
    "    commentaire = re.sub(r'([^\\s\\w]|_)+', '', commentaire)\n",
    "    return commentaire.strip().lower()\n",
    "\n",
    "df['text'] = df['commentaires'].apply(clean_str)\n",
    "\n",
    "##########################\n",
    "### Ré-échantillonnage ###\n",
    "##########################\n",
    "# df.notes.value_counts()\n",
    "# >>> 4:43209, 3:13751, 0:10773, 2:4122, 1:1991\n",
    "# Notre jeu de données est déséquilibré\n",
    "df_0 = df[df['notes'] == 0].sample(frac=1)\n",
    "df_1 = df[df['notes'] == 1].sample(frac=1)\n",
    "df_2 = df[df['notes'] == 2].sample(frac=1)\n",
    "df_3 = df[df['notes'] == 3].sample(frac=1)\n",
    "df_4 = df[df['notes'] == 4].sample(frac=1)\n",
    "\n",
    "sample_size = len(df[df['notes'] == 0]) # Nombre d'échantillons minimum\n",
    "data = pd.concat([df_0.head(sample_size), df_1.head(sample_size), df_2.head(sample_size), df_3.head(sample_size), df_4.head(sample_size)]).sample(frac=1)\n",
    "\n",
    "####################\n",
    "### Tokenization ###\n",
    "####################\n",
    "sequence_length = 100\n",
    "max_features = 40000\n",
    "\n",
    "tokenizer = Tokenizer(num_words=max_features, split=' ', oov_token='<unw>')\n",
    "tokenizer.fit_on_texts(data['commentaires'].values)\n",
    "X = tokenizer.texts_to_sequences(data['commentaires'].values)\n",
    "\n",
    "###############\n",
    "### PADDING ###\n",
    "###############\n",
    "X = pad_sequences(X, sequence_length)\n",
    "\n",
    "#######################\n",
    "### DICHOTOMISATION ###\n",
    "#######################\n",
    "y = pd.get_dummies(data['notes']).values"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "32099e2f-1856-4297-a38e-8e97189b803b",
   "metadata": {},
   "source": [
    "**4. TRAIN_TEST_SPLIT**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "4612cf9d-8a48-48ee-9e2e-bd61ecd0ae23",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train = (30745, 100)\n",
      "X_test = (7687, 100)\n",
      "y_train = (30745, 5)\n",
      "y_test = (7687, 5)\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = model_selection.train_test_split(X, y, test_size=0.2, random_state=42)\n",
    "\n",
    "# Vérification\n",
    "print(\"X_train =\", X_train.shape)\n",
    "print(\"X_test =\", X_test.shape)\n",
    "print(\"y_train =\", y_train.shape)\n",
    "print(\"y_test =\", y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "533730c8-86dc-4d64-8b95-85c278efc08c",
   "metadata": {},
   "source": [
    "**5. MODELISATION 1 : CNN aléatoire**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "ebba8b5d-b15b-4b21-bb67-69d56040e8f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_dim = 300\n",
    "num_filters = 100\n",
    "batch_size = 128\n",
    "epochs = 20"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "0b4cdd16-f975-40cc-b29e-ea60629ea1d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "reduce_learning_rate = ReduceLROnPlateau(monitor = 'val_loss',\t\t\t\t\t\n",
    "                                         min_delta = 0.01 ,\t\t\t\t\n",
    "                                         patience = 3,\t\t\t\t\t\n",
    "                                         factor = 0.1, \t\t\t\t\n",
    "                                         cooldown = 4,\t\t\t\t\n",
    "                                         verbose = 1)\n",
    "\n",
    "early_stopping = EarlyStopping(monitor = 'val_loss',\n",
    "                               min_delta = 0.01,\n",
    "                               patience = 5,\n",
    "                               verbose = 1,\n",
    "                               mode='min',\n",
    "                               restore_best_weights = True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "12d1cd58-b72e-4cdd-91d0-6ed1c43401c1",
   "metadata": {},
   "source": [
    "*1) Modélisation*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "5e158b56-d626-4843-a922-3e1e22967f10",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "193/193 [==============================] - 33s 169ms/step - loss: 1.8400 - accuracy: 0.2885 - val_loss: 1.4703 - val_accuracy: 0.2822\n",
      "Epoch 2/20\n",
      "193/193 [==============================] - 33s 172ms/step - loss: 1.4756 - accuracy: 0.3262 - val_loss: 1.5228 - val_accuracy: 0.3635\n",
      "Epoch 3/20\n",
      "193/193 [==============================] - 34s 175ms/step - loss: 1.4786 - accuracy: 0.4144 - val_loss: 1.4649 - val_accuracy: 0.4594\n",
      "Epoch 4/20\n",
      "193/193 [==============================] - 32s 166ms/step - loss: 1.4451 - accuracy: 0.4668 - val_loss: 1.4333 - val_accuracy: 0.4680\n",
      "Epoch 5/20\n",
      "193/193 [==============================] - 31s 162ms/step - loss: 1.4122 - accuracy: 0.4826 - val_loss: 1.4277 - val_accuracy: 0.4786\n",
      "Epoch 6/20\n",
      "193/193 [==============================] - 33s 171ms/step - loss: 1.3998 - accuracy: 0.4865 - val_loss: 1.3999 - val_accuracy: 0.4594\n",
      "Epoch 7/20\n",
      "193/193 [==============================] - 31s 160ms/step - loss: 1.3725 - accuracy: 0.5010 - val_loss: 1.4637 - val_accuracy: 0.4620\n",
      "Epoch 8/20\n",
      "193/193 [==============================] - 32s 166ms/step - loss: 1.3747 - accuracy: 0.4983 - val_loss: 1.4129 - val_accuracy: 0.4944\n",
      "Epoch 9/20\n",
      "193/193 [==============================] - 32s 164ms/step - loss: 1.3417 - accuracy: 0.5030 - val_loss: 1.3482 - val_accuracy: 0.5020\n",
      "Epoch 10/20\n",
      "193/193 [==============================] - 32s 166ms/step - loss: 1.3531 - accuracy: 0.5052 - val_loss: 1.3604 - val_accuracy: 0.4972\n",
      "Epoch 11/20\n",
      "193/193 [==============================] - 32s 163ms/step - loss: 1.3203 - accuracy: 0.5114 - val_loss: 1.3686 - val_accuracy: 0.4941\n",
      "Epoch 12/20\n",
      "193/193 [==============================] - 31s 160ms/step - loss: 1.3142 - accuracy: 0.5146 - val_loss: 1.3303 - val_accuracy: 0.5087\n",
      "Epoch 13/20\n",
      "193/193 [==============================] - 32s 165ms/step - loss: 1.2932 - accuracy: 0.5212 - val_loss: 1.3883 - val_accuracy: 0.4973\n",
      "Epoch 14/20\n",
      "193/193 [==============================] - 32s 168ms/step - loss: 1.3064 - accuracy: 0.5187 - val_loss: 1.3936 - val_accuracy: 0.4981\n",
      "Epoch 15/20\n",
      "193/193 [==============================] - 32s 163ms/step - loss: 1.2836 - accuracy: 0.5218 - val_loss: 1.4455 - val_accuracy: 0.4864\n",
      "Epoch 16/20\n",
      "193/193 [==============================] - 31s 162ms/step - loss: 1.2877 - accuracy: 0.5245 - val_loss: 1.3401 - val_accuracy: 0.5071\n",
      "Epoch 17/20\n",
      "193/193 [==============================] - 31s 161ms/step - loss: 1.2790 - accuracy: 0.5285 - val_loss: 1.3968 - val_accuracy: 0.4861\n",
      "Epoch 18/20\n",
      "193/193 [==============================] - 32s 167ms/step - loss: 1.2699 - accuracy: 0.5293 - val_loss: 1.3164 - val_accuracy: 0.5061\n",
      "Epoch 19/20\n",
      "193/193 [==============================] - 32s 167ms/step - loss: 1.2641 - accuracy: 0.5280 - val_loss: 1.3234 - val_accuracy: 0.4671\n",
      "Epoch 20/20\n",
      "193/193 [==============================] - 32s 166ms/step - loss: 1.2753 - accuracy: 0.5301 - val_loss: 1.2919 - val_accuracy: 0.5111\n"
     ]
    }
   ],
   "source": [
    "########################\n",
    "### Architecture CNN ###\n",
    "########################\n",
    "inputs = Input(shape=(sequence_length,), dtype='int32')\n",
    "embedding_layer = Embedding(input_dim=max_features, output_dim=embedding_dim, input_length=sequence_length)(inputs)\n",
    "reshape = Reshape((sequence_length, embedding_dim, 1))(embedding_layer)\n",
    "conv_0 = Conv2D(num_filters, kernel_size=(3, embedding_dim), activation='relu', kernel_regularizer=regularizers.l2(3))(reshape)\n",
    "conv_1 = Conv2D(num_filters, kernel_size=(4, embedding_dim), activation='relu', kernel_regularizer=regularizers.l2(3))(reshape)\n",
    "conv_2 = Conv2D(num_filters, kernel_size=(5, embedding_dim), activation='relu', kernel_regularizer=regularizers.l2(3))(reshape)\n",
    "maxpool_0 = MaxPool2D(pool_size=(sequence_length - 3 + 1, 1), strides=(1,1), padding='valid')(conv_0)\n",
    "maxpool_1 = MaxPool2D(pool_size=(sequence_length - 4 + 1, 1), strides=(1,1), padding='valid')(conv_1)\n",
    "maxpool_2 = MaxPool2D(pool_size=(sequence_length - 5 + 1, 1), strides=(1,1), padding='valid')(conv_2)\n",
    "concatenated_tensor = Concatenate(axis=1)([maxpool_0, maxpool_1, maxpool_2])\n",
    "flatten = Flatten()(concatenated_tensor)\n",
    "dropout = Dropout(0.5)(flatten)\n",
    "output = Dense(units=5, activation='softmax')(dropout)\n",
    "\n",
    "###################\n",
    "### Compilation ###\n",
    "###################\n",
    "model = Model(inputs=inputs, outputs=output)\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "# print(model.summary())\n",
    "\n",
    "##############################\n",
    "### Entraînement du modèle ###\n",
    "##############################\n",
    "history = model.fit(X_train, y_train, \n",
    "                    epochs=epochs, \n",
    "                    batch_size=batch_size, \n",
    "                    verbose=1, \n",
    "                    validation_split=0.2, \n",
    "                    shuffle=True)\n",
    "                    #callbacks = [reduce_learning_rate, early_stopping])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bdcc00d8-4abe-4f8a-b310-2982e475e2f6",
   "metadata": {},
   "source": [
    "*2) Performances du modèle*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "b06095ea-57fd-4aed-b8d8-0e1d8be9d161",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x300 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12,3))\n",
    "plt.subplot(121)\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "\n",
    "plt.subplot(122)\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'validation'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "7a539ac6-6e17-455f-8be2-86edde2acb2d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "241/241 [==============================] - 2s 9ms/step\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.63      0.84      0.72      2216\n",
      "           1       0.00      0.00      0.00       405\n",
      "           2       0.00      0.00      0.00       828\n",
      "           3       0.35      0.26      0.30      2108\n",
      "           4       0.48      0.70      0.57      2130\n",
      "\n",
      "    accuracy                           0.51      7687\n",
      "   macro avg       0.29      0.36      0.32      7687\n",
      "weighted avg       0.41      0.51      0.45      7687\n",
      "\n"
     ]
    },
    {
     "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>Classe prédite</th>\n",
       "      <th>0</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Classe réelle</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1871</td>\n",
       "      <td>218</td>\n",
       "      <td>127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>260</td>\n",
       "      <td>98</td>\n",
       "      <td>47</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>330</td>\n",
       "      <td>309</td>\n",
       "      <td>189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>295</td>\n",
       "      <td>556</td>\n",
       "      <td>1257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>214</td>\n",
       "      <td>423</td>\n",
       "      <td>1493</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Classe prédite     0    3     4\n",
       "Classe réelle                  \n",
       "0               1871  218   127\n",
       "1                260   98    47\n",
       "2                330  309   189\n",
       "3                295  556  1257\n",
       "4                214  423  1493"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = model.predict(X_test)\n",
    "accuracy_score(list(map(lambda x: np.argmax(x), y_test)), list(map(lambda x: np.argmax(x), y_pred)))\n",
    "\n",
    "print(classification_report(np.argmax(y_test, axis=1), np.argmax(y_pred, axis=1)))\n",
    "\n",
    "confusion_matrix = pd.crosstab(np.argmax(y_test, axis=1), np.argmax(y_pred, axis=1), rownames=['Classe réelle'], colnames=['Classe prédite'])\n",
    "display(confusion_matrix)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b5d7467b-5a49-4b34-aa0f-0a73d1f025f6",
   "metadata": {},
   "source": [
    "**6. ENREGISTREMENT DU MODELE**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "657ec036-7146-4b28-a60b-1921709966e2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['7.a) Deep_Learning_1.pkl']"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "joblib.dump(model, '7.a) Deep_Learning_1.pkl')"
   ]
  }
 ],
 "metadata": {
  "colab": {
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}