Textual_BigData - INFOMSO3.8 - Projet 7.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import spacy\n",
    "import en_core_web_md\n",
    "from time import time\n",
    "import re\n",
    "import numpy as np\n",
    "import random\n",
    "import seaborn as sns\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "from sklearn.metrics import confusion_matrix, accuracy_score, precision_score, recall_score, f1_score, precision_recall_curve, roc_curve, roc_auc_score, classification_report\n",
    "from sklearn.model_selection import train_test_split, StratifiedShuffleSplit, cross_val_predict, GridSearchCV\n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier, ExtraTreesClassifier, GradientBoostingClassifier\n",
    "from sklearn.linear_model import SGDClassifier\n",
    "from sklearn.naive_bayes import GaussianNB, MultinomialNB\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.svm import LinearSVC, SVC\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.pipeline import Pipeline\n",
    "\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer, TfidfTransformer\n",
    "\n",
    "from xgboost import XGBClassifier\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import operator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# !python -m spacy download en_core_web_md"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Présentation et objectifs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Le but de ce projet est d'appliquer des méthodes de Natural Language Processing à un problème de classification de textes. Pour cela, nous allons prendre exemple sur un dataset de synopsis films labellisés par leur genre. Le but sera alors de prédire le genre d'un film connaissant son synopsis.\n",
    "\n",
    "Les membres du projet sont ADIM Mehdi, BABINET Pierre, DARTIGUENAVE Aliénor et KUOCH Thomas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Description des données"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0, 0.5, 'proportion'),\n",
       " Text(0.5, 0, 'categorie'),\n",
       " Text(0.5, 1.0, 'Proportion des catégories de films dans le dataset')]"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "df = pd.read_csv('data.csv')\n",
    "ax = df['Genre1'].value_counts(normalize=True).plot(kind='bar')\n",
    "ax.set(xlabel=\"categorie\", ylabel=\"proportion\", title=\"Proportion des catégories de films dans le dataset\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Il y a 4000 films dans le dataset\n"
     ]
    }
   ],
   "source": [
    "print(\"Il y a {} films dans le dataset\".format(len(df)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "On ne sélectionne seulement les catégories qui ont plus de 100 films."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[Text(0, 0.5, 'proportion'),\n",
       " Text(0.5, 0, 'categorie'),\n",
       " Text(0.5, 1.0, 'Proportion des catégories de films dans le dataset après sélection')]"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "# Select only genres which have more than 100 synopsis\n",
    "genres = pd.DataFrame(df['Genre1'].value_counts())\n",
    "to_select = genres.loc[genres['Genre1'] > 100]\n",
    "to_select = list(to_select.index.values)\n",
    "df = df[(df.Genre1.isin(to_select))]\n",
    "df = df.reset_index()\n",
    "\n",
    "ax = df['Genre1'].value_counts(normalize=True).plot(kind='bar')\n",
    "ax.set(xlabel=\"categorie\", ylabel=\"proportion\", title=\"Proportion des catégories de films dans le dataset après sélection\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Lemmatisation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "On procède à la lemmatisation des synopsis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "nlp = spacy.load('en', disable=['ner', 'parser'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "def cleaning(doc):\n",
    "    txt = [token.lemma_ for token in doc if not token.is_stop]\n",
    "    if len(txt) > 2:\n",
    "        return ' '.join(txt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "brief_cleaning = (re.sub(\"[^A-Za-z']+\", ' ', str(row)).lower() for row in df['Synopsis'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time to clean up everything: 0.71 mins\n"
     ]
    }
   ],
   "source": [
    "t = time()\n",
    "\n",
    "txt = [cleaning(doc) for doc in nlp.pipe(brief_cleaning, batch_size=5000, n_threads=-1)]\n",
    "\n",
    "print('Time to clean up everything: {} mins'.format(round((time() - t) / 60, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_clean = pd.DataFrame({'clean': txt})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Rearrange columns\n",
    "df['Synopsis_clean'] = df_clean\n",
    "cols = ['Title', 'Synopsis', 'Synopsis_clean', 'Genre1', 'Genre2', 'Genre3']\n",
    "df = df[cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Average of count words in Synopsis_clean\n",
    "\n",
    "mean = []\n",
    "std = []\n",
    "\n",
    "for column in ['Synopsis', 'Synopsis_clean']:\n",
    "    count_of_words = []\n",
    "    for index, row in df.iterrows():\n",
    "        phrase = row[column]\n",
    "        try:\n",
    "            count_of_words.append(len(phrase.split(\" \")))\n",
    "        except:\n",
    "            pass\n",
    "    mean.append(np.mean(count_of_words))\n",
    "    std.append(np.std(count_of_words))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "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": [
    "# data to plot\n",
    "n_groups = 2\n",
    "\n",
    "# create plot\n",
    "fig, ax = plt.subplots()\n",
    "index = np.arange(n_groups)\n",
    "bar_width = 0.35\n",
    "opacity = 0.8\n",
    " \n",
    "rects1 = plt.bar(index, mean, bar_width,\n",
    "alpha=opacity,\n",
    "color='b',\n",
    "label='Mean')\n",
    " \n",
    "rects2 = plt.bar(index + bar_width, std, bar_width,\n",
    "alpha=opacity,\n",
    "color='g',\n",
    "label='Standard Deviation')\n",
    " \n",
    "plt.title('Count of Words in Synopsis')\n",
    "plt.xticks(index + bar_width/2, ('Before Lemmatisation', 'After Lemmatisation'))\n",
    "plt.legend()\n",
    " \n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Construction du modèle de prédiction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Stratification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Le dataset étant déséquilibré, il faut créer des jeux d'entraînement et de test qui tiennent compte de ces proportions. On crée ce que l'on appelle des jeux stratifiés."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "split = StratifiedShuffleSplit(n_splits=1, test_size=0.2, random_state=1)\n",
    "for train_index, test_index in split.split(df, df[\"Genre1\"]):\n",
    "    strat_train_set = df.loc[train_index]\n",
    "    strat_test_set = df.loc[test_index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "weights = dict(strat_train_set['Genre1'].value_counts(normalize=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_XY(df):\n",
    "    columns = [str(i) for i in range (0, 128)]\n",
    "    X = np.empty((0,128))\n",
    "    k = 0\n",
    "\n",
    "    for index, row in df.iterrows():\n",
    "        phrase = row['Synopsis_clean']\n",
    "        embedded = nlp(phrase).vector\n",
    "        X = np.vstack((X, embedded))\n",
    "        \n",
    "    Y = df['Genre1'].values\n",
    "    return X, Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "Xtrain, Ytrain = get_XY(strat_train_set)\n",
    "Xtest, Ytest = get_XY(strat_test_set)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "sc = StandardScaler()\n",
    "sc.fit(Xtrain)\n",
    "Xtrain_norm = sc.transform(Xtrain)\n",
    "Xtest_norm = sc.transform(Xtest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pca = PCA(0.75) #On veut k tel qu'il permet de garder 75% de la variance\n",
    "# pca.fit(Xtrain)\n",
    "# Xtrain = pca.transform(Xtrain)\n",
    "# Xtest = pca.transform(Xtest)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "La PCA ne donne pas de meilleurs résultats."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ML Classifiers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "clfs = {\n",
    "    'CART': DecisionTreeClassifier(class_weight=weights),\n",
    "    'RF': RandomForestClassifier(n_estimators=300, max_depth=1, class_weight=weights),\n",
    "    'SGD' : SGDClassifier(loss='hinge', penalty='l2',alpha=1e-3, random_state=42, max_iter=5, tol=None),\n",
    "    'svc' : LinearSVC()\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "genres = pd.DataFrame(df['Genre1'].value_counts())\n",
    "genres = list(genres.index.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_classifier(clfs, Xtrain, Ytrain):\n",
    "    for i in clfs:\n",
    "        \n",
    "        clf = clfs[i]\n",
    "        clf.fit(Xtrain, Ytrain)\n",
    "\n",
    "        Ypred = cross_val_predict(clf, Xtrain, Ytrain, cv=10)\n",
    "\n",
    "        print('Classification report for classifier', i)\n",
    "        print('Global Accuracy %s' % accuracy_score(Ypred, Ytrain))\n",
    "        print(classification_report(Ytrain, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for classifier CART\n",
      "Global Accuracy 0.20025632809996796\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.20      0.21      0.20       550\n",
      "       Drama       0.04      0.04      0.04       162\n",
      "      Action       0.10      0.09      0.09       190\n",
      "       Crime       0.05      0.06      0.05       143\n",
      "   Animation       0.33      0.32      0.32       951\n",
      " Documentary       0.09      0.10      0.09       226\n",
      "   Adventure       0.12      0.12      0.12       183\n",
      "   Biography       0.22      0.21      0.22       619\n",
      "      Horror       0.02      0.02      0.02        97\n",
      "\n",
      "   micro avg       0.20      0.20      0.20      3121\n",
      "   macro avg       0.13      0.13      0.13      3121\n",
      "weighted avg       0.20      0.20      0.20      3121\n",
      "\n",
      "Classification report for classifier RF\n",
      "Global Accuracy 0.30471002883691123\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.00      0.00      0.00       550\n",
      "       Drama       0.00      0.00      0.00       162\n",
      "      Action       0.00      0.00      0.00       190\n",
      "       Crime       0.00      0.00      0.00       143\n",
      "   Animation       0.30      1.00      0.47       951\n",
      " Documentary       0.00      0.00      0.00       226\n",
      "   Adventure       0.00      0.00      0.00       183\n",
      "   Biography       0.00      0.00      0.00       619\n",
      "      Horror       0.00      0.00      0.00        97\n",
      "\n",
      "   micro avg       0.30      0.30      0.30      3121\n",
      "   macro avg       0.03      0.11      0.05      3121\n",
      "weighted avg       0.09      0.30      0.14      3121\n",
      "\n",
      "Classification report for classifier SGD\n",
      "Global Accuracy 0.28740788208907403\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.26      0.33      0.29       550\n",
      "       Drama       0.09      0.02      0.03       162\n",
      "      Action       0.14      0.04      0.06       190\n",
      "       Crime       0.16      0.03      0.05       143\n",
      "   Animation       0.35      0.52      0.42       951\n",
      " Documentary       0.11      0.05      0.07       226\n",
      "   Adventure       0.37      0.14      0.20       183\n",
      "   Biography       0.23      0.27      0.25       619\n",
      "      Horror       0.13      0.02      0.04        97\n",
      "\n",
      "   micro avg       0.29      0.29      0.29      3121\n",
      "   macro avg       0.20      0.16      0.16      3121\n",
      "weighted avg       0.25      0.29      0.25      3121\n",
      "\n",
      "Classification report for classifier svc\n",
      "Global Accuracy 0.3431592438321051\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.30      0.34      0.32       550\n",
      "       Drama       0.14      0.01      0.01       162\n",
      "      Action       0.04      0.01      0.01       190\n",
      "       Crime       0.22      0.04      0.07       143\n",
      "   Animation       0.38      0.68      0.49       951\n",
      " Documentary       0.21      0.03      0.05       226\n",
      "   Adventure       0.28      0.22      0.25       183\n",
      "   Biography       0.31      0.29      0.30       619\n",
      "      Horror       0.00      0.00      0.00        97\n",
      "\n",
      "   micro avg       0.34      0.34      0.34      3121\n",
      "   macro avg       0.21      0.18      0.17      3121\n",
      "weighted avg       0.28      0.34      0.29      3121\n",
      "\n"
     ]
    }
   ],
   "source": [
    "run_classifier(clfs, Xtrain_norm, Ytrain)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cette première approche ne donne pas de résultats satisfaisants."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Méthode TFIDF"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Il s'agit maintenant d'essayer la méthode TFIDF pour préparer nos données avant d'essayer nos modèles."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "Xtrain, Ytrain = strat_train_set['Synopsis_clean'], strat_train_set['Genre1']\n",
    "Xtest, Ytest = strat_test_set['Synopsis_clean'], strat_test_set['Genre1']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_pipeline(clfs, Xtrain, Ytrain):\n",
    "    for i in clfs:\n",
    "        clf = clfs[i]\n",
    "        \n",
    "        pipeline = Pipeline([('vect', CountVectorizer()),\n",
    "                       ('tfidf', TfidfTransformer()),\n",
    "                       ('clf', clf),\n",
    "                      ])\n",
    "        \n",
    "        pipeline.fit(Xtrain, Ytrain)\n",
    "        \n",
    "        Ypred = cross_val_predict(pipeline, Xtrain, Ytrain, cv=5)\n",
    "        print('Classification report for classifier', i)\n",
    "        print('Global Accuracy %s' % accuracy_score(Ypred, Ytrain))\n",
    "        print(classification_report(Ytrain, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for classifier CART\n",
      "Global Accuracy 0.30278756808715157\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.35      0.40      0.38       550\n",
      "       Drama       0.06      0.05      0.06       162\n",
      "      Action       0.15      0.11      0.12       190\n",
      "       Crime       0.09      0.06      0.07       143\n",
      "   Animation       0.41      0.48      0.44       951\n",
      " Documentary       0.13      0.09      0.11       226\n",
      "   Adventure       0.41      0.33      0.37       183\n",
      "   Biography       0.24      0.24      0.24       619\n",
      "      Horror       0.06      0.05      0.05        97\n",
      "\n",
      "   micro avg       0.30      0.30      0.30      3121\n",
      "   macro avg       0.21      0.20      0.20      3121\n",
      "weighted avg       0.29      0.30      0.29      3121\n",
      "\n",
      "Classification report for classifier RF\n",
      "Global Accuracy 0.30471002883691123\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.00      0.00      0.00       550\n",
      "       Drama       0.00      0.00      0.00       162\n",
      "      Action       0.00      0.00      0.00       190\n",
      "       Crime       0.00      0.00      0.00       143\n",
      "   Animation       0.30      1.00      0.47       951\n",
      " Documentary       0.00      0.00      0.00       226\n",
      "   Adventure       0.00      0.00      0.00       183\n",
      "   Biography       0.00      0.00      0.00       619\n",
      "      Horror       0.00      0.00      0.00        97\n",
      "\n",
      "   micro avg       0.30      0.30      0.30      3121\n",
      "   macro avg       0.03      0.11      0.05      3121\n",
      "weighted avg       0.09      0.30      0.14      3121\n",
      "\n",
      "Classification report for classifier SGD\n",
      "Global Accuracy 0.49471323293816083\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.52      0.71      0.60       550\n",
      "       Drama       0.34      0.07      0.12       162\n",
      "      Action       0.61      0.20      0.30       190\n",
      "       Crime       0.30      0.05      0.08       143\n",
      "   Animation       0.50      0.79      0.61       951\n",
      " Documentary       0.35      0.15      0.21       226\n",
      "   Adventure       0.65      0.58      0.61       183\n",
      "   Biography       0.42      0.31      0.36       619\n",
      "      Horror       0.45      0.13      0.21        97\n",
      "\n",
      "   micro avg       0.49      0.49      0.49      3121\n",
      "   macro avg       0.46      0.33      0.34      3121\n",
      "weighted avg       0.47      0.49      0.45      3121\n",
      "\n",
      "Classification report for classifier svc\n",
      "Global Accuracy 0.4793335469400833\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.51      0.65      0.57       550\n",
      "       Drama       0.28      0.09      0.14       162\n",
      "      Action       0.49      0.21      0.29       190\n",
      "       Crime       0.37      0.08      0.13       143\n",
      "   Animation       0.52      0.70      0.60       951\n",
      " Documentary       0.35      0.17      0.23       226\n",
      "   Adventure       0.72      0.56      0.63       183\n",
      "   Biography       0.36      0.41      0.38       619\n",
      "      Horror       0.58      0.14      0.23        97\n",
      "\n",
      "   micro avg       0.48      0.48      0.48      3121\n",
      "   macro avg       0.46      0.33      0.35      3121\n",
      "weighted avg       0.47      0.48      0.45      3121\n",
      "\n"
     ]
    }
   ],
   "source": [
    "run_pipeline(clfs, Xtrain, Ytrain)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Le meilleur classifier est le SGDGradient et le LinearSVC"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classifieur SGD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "cv = CountVectorizer()\n",
    "Xtrain_cv = cv.fit_transform(Xtrain)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for classifier SGD\n",
      "Global Accuracy 0.49086831143864146\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.51      0.72      0.60       550\n",
      "       Drama       0.43      0.07      0.13       162\n",
      "      Action       0.66      0.23      0.34       190\n",
      "       Crime       0.42      0.08      0.13       143\n",
      "   Animation       0.49      0.80      0.61       951\n",
      " Documentary       0.36      0.11      0.17       226\n",
      "   Adventure       0.66      0.56      0.61       183\n",
      "   Biography       0.40      0.27      0.32       619\n",
      "      Horror       0.53      0.10      0.17        97\n",
      "\n",
      "   micro avg       0.49      0.49      0.49      3121\n",
      "   macro avg       0.49      0.33      0.34      3121\n",
      "weighted avg       0.48      0.49      0.44      3121\n",
      "\n"
     ]
    }
   ],
   "source": [
    "sgd = SGDClassifier(loss='hinge', penalty='l2',alpha=1e-3, random_state=None, max_iter=5, tol=None, class_weight = None)\n",
    "\n",
    "tfidf = TfidfVectorizer()\n",
    "tfidf.fit(Xtrain, Ytrain)\n",
    "Xtrain_tfidf = tfidf.transform(Xtrain)\n",
    "\n",
    "Ypred = cross_val_predict(sgd, Xtrain_tfidf, Ytrain, cv=10)      \n",
    "\n",
    "print('Classification report for classifier SGD')\n",
    "print('Global Accuracy %s' % accuracy_score(Ypred, Ytrain))\n",
    "print(classification_report(Ytrain, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for classifier SGD\n",
      "Global Accuracy 0.5236875800256082\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.56      0.74      0.64       138\n",
      "       Drama       0.67      0.10      0.17        41\n",
      "      Action       0.67      0.30      0.41        47\n",
      "       Crime       0.43      0.08      0.14        36\n",
      "   Animation       0.50      0.82      0.62       238\n",
      " Documentary       0.42      0.14      0.21        56\n",
      "   Adventure       0.64      0.61      0.62        46\n",
      "   Biography       0.45      0.30      0.36       155\n",
      "      Horror       0.90      0.38      0.53        24\n",
      "\n",
      "   micro avg       0.52      0.52      0.52       781\n",
      "   macro avg       0.58      0.38      0.41       781\n",
      "weighted avg       0.53      0.52      0.48       781\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Evaluation du Classifier\n",
    "sgd.fit(Xtrain_tfidf, Ytrain)\n",
    "\n",
    "Xtest_tfidf = tfidf.transform(Xtest)\n",
    "Ypred = sgd.predict(Xtest_tfidf)\n",
    "print('Classification report for classifier SGD')\n",
    "print('Global Accuracy %s' % accuracy_score(Ypred, Ytest))\n",
    "print(classification_report(Ytest, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classifieur SVM "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for LinearSVM\n",
      "Global Accuracy 0.47965395706504327\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.52      0.65      0.58       550\n",
      "       Drama       0.33      0.10      0.15       162\n",
      "      Action       0.56      0.21      0.31       190\n",
      "       Crime       0.39      0.08      0.13       143\n",
      "   Animation       0.50      0.71      0.59       951\n",
      " Documentary       0.41      0.18      0.25       226\n",
      "   Adventure       0.73      0.50      0.60       183\n",
      "   Biography       0.36      0.40      0.38       619\n",
      "      Horror       0.60      0.12      0.21        97\n",
      "\n",
      "   micro avg       0.48      0.48      0.48      3121\n",
      "   macro avg       0.49      0.33      0.35      3121\n",
      "weighted avg       0.48      0.48      0.45      3121\n",
      "\n"
     ]
    }
   ],
   "source": [
    "svc = LinearSVC()\n",
    "\n",
    "tfidf = TfidfVectorizer()\n",
    "tfidf.fit(Xtrain, Ytrain)\n",
    "Xtrain_tfidf = tfidf.transform(Xtrain)\n",
    "\n",
    "Ypred = cross_val_predict(svc, Xtrain_tfidf, Ytrain, cv=10)      \n",
    "\n",
    "print('Classification report for LinearSVM')\n",
    "print('Global Accuracy %s' % accuracy_score(Ypred, Ytrain))\n",
    "print(classification_report(Ytrain, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report for classifier LinearSVC\n",
      "Global Accuracy 0.5198463508322664\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "      Comedy       0.58      0.70      0.63       138\n",
      "       Drama       0.50      0.15      0.23        41\n",
      "      Action       0.58      0.32      0.41        47\n",
      "       Crime       0.50      0.17      0.25        36\n",
      "   Animation       0.52      0.71      0.60       238\n",
      " Documentary       0.34      0.18      0.24        56\n",
      "   Adventure       0.75      0.59      0.66        46\n",
      "   Biography       0.42      0.45      0.43       155\n",
      "      Horror       0.88      0.29      0.44        24\n",
      "\n",
      "   micro avg       0.52      0.52      0.52       781\n",
      "   macro avg       0.56      0.39      0.43       781\n",
      "weighted avg       0.52      0.52      0.50       781\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#Evaluation du Classifier\n",
    "svc.fit(Xtrain_tfidf, Ytrain)\n",
    "\n",
    "Xtest_tfidf = tfidf.transform(Xtest)\n",
    "Ypred = svc.predict(Xtest_tfidf)\n",
    "print('Classification report for classifier LinearSVC')\n",
    "print('Global Accuracy %s' % accuracy_score(Ypred, Ytest))\n",
    "print(classification_report(Ytest, Ypred, target_names=genres))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Matrice de confusion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "cm = confusion_matrix(Ypred, Ytest, labels=genres)\n",
    "cm = cm / cm.astype(np.float).sum(axis=0)\n",
    "cm = cm.round(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x20fd17f53c8>"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "ax = sns.heatmap(cm, annot=True)\n",
    "\n",
    "ax.set_xticklabels(genres)\n",
    "ax.set_yticklabels(genres)\n",
    "plt.xlabel('True')\n",
    "plt.ylabel('Predicted')\n",
    "plt.title(\"Matrice de Confusion du modèle SVM linéaire\")\n",
    "plt.xticks(rotation=90)\n",
    "plt.yticks(rotation=0)\n",
    "ax"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Malgré nos efforts et nos essais de différents modèles et approches, le problème reste difficile et nous n'avions pas réussi à avoir mieux qu'une précision de 50%. La meilleure méthode a été de lemmatiser les synopsis, de construire la matrice TFIDF et d'entraîner notre modèle sur cette matrice. "
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}