TunedModels.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true,
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import glob\n",
    "import os\n",
    "from tqdm import tqdm\n",
    "from pandas_profiling import ProfileReport\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import spacy\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "outputs": [],
   "source": [
    "df=pd.read_csv(r\"C:/Users/ch14b/Desktop/Machine Learning/Project/bert.csv\")\n",
    "test=pd.read_csv(r\"C:\\Users\\ch14b\\Desktop\\Machine Learning\\Project\\2021-01\\bertwave2.csv\")\n",
    "df=pd.concat([df,test],axis=0)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "outputs": [
    {
     "data": {
      "text/plain": "   00  000  000000  10  100  11  12  13  14  15  ...  years  yes  yesterday  \\\n0   0    0       0   0    0   0   0   0   0   0  ...      0    0          0   \n1   0    0       0   0    0   0   0   0   0   0  ...      0    0          0   \n2   0    0       0   0    0   0   0   1   0   0  ...      0    0          0   \n3   0    0       0   0    0   0   0   0   0   0  ...      0    0          0   \n4   0    0       0   0    0   0   0   0   0   0  ...      0    0          0   \n\n   yet  york  you  young  your  yourself  zero  \n0    0     0    0      0     0         0     0  \n1    0     0    1      0     0         0     0  \n2    0     0    0      0     0         0     0  \n3    0     0    0      0     0         0     0  \n4    0     0    0      0     0         0     0  \n\n[5 rows x 800 columns]",
      "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>00</th>\n      <th>000</th>\n      <th>000000</th>\n      <th>10</th>\n      <th>100</th>\n      <th>11</th>\n      <th>12</th>\n      <th>13</th>\n      <th>14</th>\n      <th>15</th>\n      <th>...</th>\n      <th>years</th>\n      <th>yes</th>\n      <th>yesterday</th>\n      <th>yet</th>\n      <th>york</th>\n      <th>you</th>\n      <th>young</th>\n      <th>your</th>\n      <th>yourself</th>\n      <th>zero</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</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>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0</td>\n      <td>0</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>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n<p>5 rows × 800 columns</p>\n</div>"
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "bow=CountVectorizer(max_features=800)\n",
    "bow.fit(df['full_text'])\n",
    "X =bow.transform(df['full_text'])\n",
    "data_bow = pd.DataFrame(X.toarray(),columns=bow.get_feature_names())\n",
    "data_bow.head()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "outputs": [
    {
     "data": {
      "text/plain": "        00  000  000000  10  100  11  12  13  14  15  ...  yes  yesterday  \\\n0        0    0       0   0    0   0   0   0   0   0  ...    0          0   \n1        0    0       0   0    0   0   0   0   0   0  ...    0          0   \n2        0    0       0   0    0   0   0   1   0   0  ...    0          0   \n3        0    0       0   0    0   0   0   0   0   0  ...    0          0   \n4        0    0       0   0    0   0   0   0   0   0  ...    0          0   \n...     ..  ...     ...  ..  ...  ..  ..  ..  ..  ..  ...  ...        ...   \n660694   0    0       0   0    0   0   0   0   0   0  ...    0          0   \n660695   0    0       0   0    0   0   0   0   0   0  ...    0          0   \n660696   0    0       0   0    0   0   0   0   0   0  ...    0          0   \n660697   0    0       0   0    0   0   0   0   0   0  ...    0          0   \n660698   0    0       0   0    0   0   0   0   0   0  ...    0          0   \n\n        yet  york  you  young  your  yourself  zero  sentiment  \n0         0     0    0      0     0         0     0          0  \n1         0     0    1      0     0         0     0          2  \n2         0     0    0      0     0         0     0          1  \n3         0     0    0      0     0         0     0          1  \n4         0     0    0      0     0         0     0          0  \n...     ...   ...  ...    ...   ...       ...   ...        ...  \n660694    0     0    1      0     0         0     0          1  \n660695    0     0    0      0     0         0     0          0  \n660696    0     0    0      0     0         0     0          1  \n660697    0     0    0      0     0         0     0          0  \n660698    0     0    0      0     0         0     0          1  \n\n[660699 rows x 801 columns]",
      "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>00</th>\n      <th>000</th>\n      <th>000000</th>\n      <th>10</th>\n      <th>100</th>\n      <th>11</th>\n      <th>12</th>\n      <th>13</th>\n      <th>14</th>\n      <th>15</th>\n      <th>...</th>\n      <th>yes</th>\n      <th>yesterday</th>\n      <th>yet</th>\n      <th>york</th>\n      <th>you</th>\n      <th>young</th>\n      <th>your</th>\n      <th>yourself</th>\n      <th>zero</th>\n      <th>sentiment</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</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>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>0</td>\n      <td>0</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>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</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    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</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    </tr>\n    <tr>\n      <th>4</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</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>660694</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</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>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>660695</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>660696</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</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    </tr>\n    <tr>\n      <th>660697</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>660698</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</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    </tr>\n  </tbody>\n</table>\n<p>660699 rows × 801 columns</p>\n</div>"
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_bow['sentiment']=df['sentiment_score'].to_numpy()\n",
    "data_bow"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "train, test = train_test_split(data_bow, test_size = 0.2)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "outputs": [],
   "source": [
    "trainX=train.loc[:,train.columns!='sentiment']\n",
    "trainY=train['sentiment']\n",
    "testX=test.loc[:,test.columns!='sentiment']\n",
    "testY=test['sentiment']"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB\n",
    "classifier = MultinomialNB(alpha=1)\n",
    "classifier.fit(trainX,trainY)\n",
    "y_pred=classifier.predict(testX)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.6414635992129559\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.63      0.68      0.66     52823\n",
      "           1       0.68      0.66      0.67     63446\n",
      "           2       0.50      0.44      0.47     15871\n",
      "\n",
      "    accuracy                           0.64    132140\n",
      "   macro avg       0.61      0.59      0.60    132140\n",
      "weighted avg       0.64      0.64      0.64    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "outputs": [
    {
     "data": {
      "text/plain": "Text(0.5, 1.0, 'Total Impurity vs effective alpha for training set')"
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": "<Figure size 640x480 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn import tree\n",
    "classifier=tree.DecisionTreeClassifier(max_features='sqrt', ccp_alpha=0.0001)\n",
    "path = classifier.cost_complexity_pruning_path(trainX, trainY)\n",
    "ccp_alphas, impurities = path.ccp_alphas, path.impurities\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "ax.plot(ccp_alphas[:-1], impurities[:-1], marker=\"o\", drawstyle=\"steps-post\")\n",
    "ax.set_xlabel(\"effective alpha\")\n",
    "ax.set_ylabel(\"total impurity of leaves\")\n",
    "ax.set_title(\"Total Impurity vs effective alpha for training set\")"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "outputs": [],
   "source": [
    "classifier=tree.DecisionTreeClassifier(ccp_alpha=0.0001)\n",
    "classifier.fit(trainX,trainY)\n",
    "y_pred=classifier.predict(testX)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.6006735280762827\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.60      0.56      0.58     52823\n",
      "           1       0.60      0.73      0.66     63446\n",
      "           2       0.62      0.22      0.33     15871\n",
      "\n",
      "    accuracy                           0.60    132140\n",
      "   macro avg       0.61      0.50      0.52    132140\n",
      "weighted avg       0.60      0.60      0.59    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "classifier = RandomForestClassifier(max_depth=10, max_features='sqrt')\n",
    "classifier.fit(trainX,trainY)\n",
    "y_pred=classifier.predict(testX)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.5893824731345543\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.69      0.41      0.51     52823\n",
      "           1       0.56      0.89      0.68     63446\n",
      "           2       0.95      0.02      0.03     15871\n",
      "\n",
      "    accuracy                           0.59    132140\n",
      "   macro avg       0.73      0.44      0.41    132140\n",
      "weighted avg       0.66      0.59      0.54    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "classifier = RandomForestClassifier(max_features='sqrt')\n",
    "classifier.fit(trainX,trainY)\n",
    "y_pred=classifier.predict(testX)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.7502043287422431\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.74      0.77      0.76     52823\n",
      "           1       0.75      0.81      0.78     63446\n",
      "           2       0.82      0.43      0.56     15871\n",
      "\n",
      "    accuracy                           0.75    132140\n",
      "   macro avg       0.77      0.67      0.70    132140\n",
      "weighted avg       0.75      0.75      0.74    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "outputs": [],
   "source": [
    "\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "classifier = MLPClassifier(alpha=1, max_iter=1000, hidden_layer_sizes=(10,))\n",
    "classifier.fit(trainX,trainY)\n",
    "y_pred=classifier.predict(testX)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.6669517178749811\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.68      0.65      0.67     52823\n",
      "           1       0.66      0.78      0.71     63446\n",
      "           2       0.68      0.29      0.41     15871\n",
      "\n",
      "    accuracy                           0.67    132140\n",
      "   macro avg       0.67      0.57      0.60    132140\n",
      "weighted avg       0.67      0.67      0.66    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "outputs": [],
   "source": [],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  0.6657635840774936\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.68      0.66      0.67     52823\n",
      "           1       0.66      0.77      0.71     63446\n",
      "           2       0.68      0.29      0.41     15871\n",
      "\n",
      "    accuracy                           0.67    132140\n",
      "   macro avg       0.67      0.57      0.59    132140\n",
      "weighted avg       0.67      0.67      0.66    132140\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy = \",metrics.accuracy_score(testY,y_pred))\n",
    "print(\"Classification Report\")\n",
    "print(metrics.classification_report(testY,y_pred))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}