Lecture11_Sentiment_Classifier_Using_Naive_Bayes_With_SKlearn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Author : Jeonghun Yoon"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Naive Bayes with TF_IDF를 이용하여 영화 리뷰를 예측하는 감정 분류기를 구현하라. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Train, test set 구분"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 영화 리뷰를 load한다. 사랑/장르라는 단어를 포함하고 있는 document를 load 한다.\n",
    "reviews = pd.read_csv('./inputs/ratings_train.txt', delimiter='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>id</th>\n",
       "      <th>document</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9976970</td>\n",
       "      <td>아 더빙.. 진짜 짜증나네요 목소리</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3819312</td>\n",
       "      <td>흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10265843</td>\n",
       "      <td>너무재밓었다그래서보는것을추천한다</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9045019</td>\n",
       "      <td>교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6483659</td>\n",
       "      <td>사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5403919</td>\n",
       "      <td>막 걸음마 뗀 3세부터 초등학교 1학년생인 8살용영화.ㅋㅋㅋ...별반개도 아까움.</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7797314</td>\n",
       "      <td>원작의 긴장감을 제대로 살려내지못했다.</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>9443947</td>\n",
       "      <td>별 반개도 아깝다 욕나온다 이응경 길용우 연기생활이몇년인지..정말 발로해도 그것보단...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>7156791</td>\n",
       "      <td>액션이 없는데도 재미 있는 몇안되는 영화</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>5912145</td>\n",
       "      <td>왜케 평점이 낮은건데? 꽤 볼만한데.. 헐리우드식 화려함에만 너무 길들여져 있나?</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id                                           document  label\n",
       "0   9976970                                아 더빙.. 진짜 짜증나네요 목소리      0\n",
       "1   3819312                  흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나      1\n",
       "2  10265843                                  너무재밓었다그래서보는것을추천한다      0\n",
       "3   9045019                      교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정      0\n",
       "4   6483659  사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...      1\n",
       "5   5403919      막 걸음마 뗀 3세부터 초등학교 1학년생인 8살용영화.ㅋㅋㅋ...별반개도 아까움.      0\n",
       "6   7797314                              원작의 긴장감을 제대로 살려내지못했다.      0\n",
       "7   9443947  별 반개도 아깝다 욕나온다 이응경 길용우 연기생활이몇년인지..정말 발로해도 그것보단...      0\n",
       "8   7156791                             액션이 없는데도 재미 있는 몇안되는 영화      1\n",
       "9   5912145      왜케 평점이 낮은건데? 꽤 볼만한데.. 헐리우드식 화려함에만 너무 길들여져 있나?      1"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 데이터 확인\n",
    "reviews.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test set 로드\n",
    "test_reviews = pd.read_csv('./inputs/ratings_test.txt', delimiter='\\t')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>document</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6270596</td>\n",
       "      <td>굳 ㅋ</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9274899</td>\n",
       "      <td>GDNTOPCLASSINTHECLUB</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>8544678</td>\n",
       "      <td>뭐야 이 평점들은.... 나쁘진 않지만 10점 짜리는 더더욱 아니잖아</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6825595</td>\n",
       "      <td>지루하지는 않은데 완전 막장임... 돈주고 보기에는....</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6723715</td>\n",
       "      <td>3D만 아니었어도 별 다섯 개 줬을텐데.. 왜 3D로 나와서 제 심기를 불편하게 하죠??</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>7898805</td>\n",
       "      <td>음악이 주가 된, 최고의 음악영화</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6315043</td>\n",
       "      <td>진정한 쓰레기</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>6097171</td>\n",
       "      <td>마치 미국애니에서 튀어나온듯한 창의력없는 로봇디자인부터가,고개를 젖게한다</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8932678</td>\n",
       "      <td>갈수록 개판되가는 중국영화 유치하고 내용없음 폼잡다 끝남 말도안되는 무기에 유치한c...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>6242223</td>\n",
       "      <td>이별의 아픔뒤에 찾아오는 새로운 인연의 기쁨 But, 모든 사람이 그렇지는 않네..</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        id                                           document  label\n",
       "0  6270596                                                굳 ㅋ      1\n",
       "1  9274899                               GDNTOPCLASSINTHECLUB      0\n",
       "2  8544678             뭐야 이 평점들은.... 나쁘진 않지만 10점 짜리는 더더욱 아니잖아      0\n",
       "3  6825595                   지루하지는 않은데 완전 막장임... 돈주고 보기에는....      0\n",
       "4  6723715  3D만 아니었어도 별 다섯 개 줬을텐데.. 왜 3D로 나와서 제 심기를 불편하게 하죠??      0\n",
       "5  7898805                                 음악이 주가 된, 최고의 음악영화      1\n",
       "6  6315043                                            진정한 쓰레기      0\n",
       "7  6097171           마치 미국애니에서 튀어나온듯한 창의력없는 로봇디자인부터가,고개를 젖게한다      0\n",
       "8  8932678  갈수록 개판되가는 중국영화 유치하고 내용없음 폼잡다 끝남 말도안되는 무기에 유치한c...      0\n",
       "9  6242223     이별의 아픔뒤에 찾아오는 새로운 인연의 기쁨 But, 모든 사람이 그렇지는 않네..      1"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 데이터 확인\n",
    "test_reviews.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. EDA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train set (150000, 3)\n",
      "Test set (50000, 3)\n"
     ]
    }
   ],
   "source": [
    "# 데이터의 크기\n",
    "print('Train set', reviews.shape)\n",
    "print('Test set', test_reviews.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    75173\n",
       "1    74827\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pos VS Neg dataset 개수 비교\n",
    "reviews.label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1    25173\n",
       "0    24827\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pos VS Neg dataset 개수 비교\n",
    "test_reviews.label.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "문장 길이의 분포를 살펴보자"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "reviews['length'] = reviews['document'].apply(lambda x: len(str(x)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>document</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9976970</td>\n",
       "      <td>아 더빙.. 진짜 짜증나네요 목소리</td>\n",
       "      <td>0</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3819312</td>\n",
       "      <td>흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나</td>\n",
       "      <td>1</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10265843</td>\n",
       "      <td>너무재밓었다그래서보는것을추천한다</td>\n",
       "      <td>0</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9045019</td>\n",
       "      <td>교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정</td>\n",
       "      <td>0</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6483659</td>\n",
       "      <td>사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...</td>\n",
       "      <td>1</td>\n",
       "      <td>61</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id                                           document  label  length\n",
       "0   9976970                                아 더빙.. 진짜 짜증나네요 목소리      0      19\n",
       "1   3819312                  흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나      1      33\n",
       "2  10265843                                  너무재밓었다그래서보는것을추천한다      0      17\n",
       "3   9045019                      교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정      0      29\n",
       "4   6483659  사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...      1      61"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1114b3898>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYcAAAEKCAYAAAD5MJl4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAFcxJREFUeJzt3X+QXeV93/H3pxAI+EcERlAiQSU3qh1gYpuoQOKOx2NSENiDmImpZXuKYmtG0w6pSZo0htIpjW2m9iQTbE9tXBWwhQcjCHGKxsHGKsbj6RRkJCD8NGELDqxRjDwCgkODLfLtH/dZc9G5q5X2rnSvtO/XzM6e85zn3P3uke797POcc89NVSFJUr9/NOoCJEnjx3CQJHUYDpKkDsNBktRhOEiSOgwHSVKH4SBJ6jAcJEkdhoMkqePQURcwW8ccc0wtWbJk1GVI0gFl69atP6qqhTP1O2DDYcmSJWzZsmXUZUjSASXJX+9JP6eVJEkdhoMkqcNwkCR1GA6SpA7DQZLUYThIkjoMB0lSh+EgSeowHCRJHQfsO6TH2Vc2Pzmw/QOnn7ifK5Gk2XHkIEnqMBwkSR2GgySpw3CQJHUYDpKkDsNBktRhOEiSOgwHSVKH4SBJ6jAcJEkdhoMkqcNwkCR1GA6SpI4ZwyHJtUmeSfJgX9sfJflekvuT/HmSBX3bLk0ykeTRJGf3ta9obRNJLulrX5pkc5LHktyY5LC5/AUlSXtvT0YOXwJW7NK2CTilqn4F+CvgUoAkJwGrgJPbPp9PckiSQ4DPAecAJwHvb30BPgVcWVXLgGeBNUP9RpKkoc0YDlX1HWDHLm3frKqdbfUuYHFbXglsqKqXquoJYAI4rX1NVNXjVfUTYAOwMkmAdwE3t/3XA+cP+TtJkoY0F+ccPgx8vS0vAp7q2zbZ2qZrfwPwXF/QTLUPlGRtki1Jtmzfvn0OSpckDTJUOCS5DNgJXD/VNKBbzaJ9oKpaV1XLq2r5woUL97ZcSdIemvXHhCZZDbwHOLOqpl7QJ4ET+rotBp5uy4PafwQsSHJoGz3095ckjcisRg5JVgAfBc6rqhf7Nm0EViU5PMlSYBnwXeBuYFm7MukweietN7ZQuQN4b9t/NXDL7H4VSdJc2ZNLWW8A7gTelGQyyRrgvwGvAzYluS/JFwCq6iHgJuBh4BvARVX1chsV/DZwG/AIcFPrC72Q+fdJJuidg7hmTn9DSdJem3FaqareP6B52hfwqroCuGJA+63ArQPaH6d3NZMkaUz4DmlJUofhIEnqMBwkSR2GgySpY9bvc9De+8rmJwe2f+D0E/dzJZK0e44cJEkdhoMkqcNwkCR1GA6SpA7DQZLUYThIkjoMB0lSh+EgSeowHCRJHYaDJKnDcJAkdRgOkqQOw0GS1OFdWcfAdHdrBe/YKmk0HDlIkjocOQxhd3/xS9KBzJGDJKljxnBIcm2SZ5I82Nd2dJJNSR5r349q7Uny2SQTSe5PcmrfPqtb/8eSrO5r/9UkD7R9Ppskc/1LSpL2zp6MHL4ErNil7RLg9qpaBtze1gHOAZa1r7XAVdALE+By4HTgNODyqUBpfdb27bfrz5Ik7WczhkNVfQfYsUvzSmB9W14PnN/Xfl313AUsSHI8cDawqap2VNWzwCZgRdv2+qq6s6oKuK7vsSRJIzLbcw7HVdU2gPb92Na+CHiqr99ka9td++SAdknSCM31CelB5wtqFu2DHzxZm2RLki3bt2+fZYmSpJnMNhx+2KaEaN+fae2TwAl9/RYDT8/QvnhA+0BVta6qllfV8oULF86ydEnSTGYbDhuBqSuOVgO39LVf2K5aOgN4vk073QacleSodiL6LOC2tu2FJGe0q5Qu7HssSdKIzPgmuCQ3AO8EjkkySe+qo08CNyVZAzwJXNC63wqcC0wALwIfAqiqHUk+Dtzd+n2sqqZOcv9beldEHQF8vX1JkkZoxnCoqvdPs+nMAX0LuGiax7kWuHZA+xbglJnqkCTtP75DWpLU4b2Vxtx092/ybq2S9iVHDpKkDsNBktRhOEiSOgwHSVKH4SBJ6jAcJEkdhoMkqcNwkCR1GA6SpA7DQZLUYThIkjq8t9IBynsuSdqXHDlIkjoMB0lSh+EgSeowHCRJHYaDJKnDcJAkdRgOkqQOw0GS1GE4SJI6hgqHJL+b5KEkDya5IcnPJ1maZHOSx5LcmOSw1vfwtj7Rti/pe5xLW/ujSc4e7leSJA1r1rfPSLII+AhwUlX9vyQ3AauAc4Erq2pDki8Aa4Cr2vdnq+qXkqwCPgW8L8lJbb+TgV8E/leSf1ZVLw/1m+lVvN2GpL0x7LTSocARSQ4FjgS2Ae8Cbm7b1wPnt+WVbZ22/cwkae0bquqlqnoCmABOG7IuSdIQZh0OVfUD4I+BJ+mFwvPAVuC5qtrZuk0Ci9ryIuCptu/O1v8N/e0D9pEkjcCswyHJUfT+6l9KbzroNcA5A7rW1C7TbJuufdDPXJtkS5It27dv3/uiJUl7ZJhppd8Anqiq7VX1U+CrwK8DC9o0E8Bi4Om2PAmcANC2/wKwo799wD6vUlXrqmp5VS1fuHDhEKVLknZnmHB4EjgjyZHt3MGZwMPAHcB7W5/VwC1teWNbp23/VlVVa1/VrmZaCiwDvjtEXZKkIc36aqWq2pzkZuAeYCdwL7AO+AtgQ5JPtLZr2i7XAF9OMkFvxLCqPc5D7Uqnh9vjXOSVSpI0WkN9ElxVXQ5cvkvz4wy42qiq/h64YJrHuQK4YphaJElzx3dIS5I6DAdJUofhIEnqMBwkSR2GgySpY6irlXTg84Z8kgZx5CBJ6jAcJEkdhoMkqcNwkCR1GA6SpA7DQZLUYThIkjoMB0lSh+EgSeowHCRJHYaDJKnDcJAkdRgOkqQOw0GS1GE4SJI6DAdJUsdQH/aTZAFwNXAKUMCHgUeBG4ElwPeBf1VVzyYJ8BngXOBF4Leq6p72OKuB/9Qe9hNVtX6YujQ8PwRImt+GHTl8BvhGVb0ZeAvwCHAJcHtVLQNub+sA5wDL2tda4CqAJEcDlwOnA6cBlyc5asi6JElDmHU4JHk98A7gGoCq+klVPQesBKb+8l8PnN+WVwLXVc9dwIIkxwNnA5uqakdVPQtsAlbMti5J0vCGmVZ6I7Ad+GKStwBbgYuB46pqG0BVbUtybOu/CHiqb//J1jZdu8aQ003S/DDMtNKhwKnAVVX1NuDveGUKaZAMaKvdtHcfIFmbZEuSLdu3b9/beiVJe2iYcJgEJqtqc1u/mV5Y/LBNF9G+P9PX/4S+/RcDT++mvaOq1lXV8qpavnDhwiFKlyTtzqzDoar+BngqyZta05nAw8BGYHVrWw3c0pY3Ahem5wzg+Tb9dBtwVpKj2onos1qbJGlEhrqUFfh3wPVJDgMeBz5EL3BuSrIGeBK4oPW9ld5lrBP0LmX9EEBV7UjyceDu1u9jVbVjyLokSUNI1cDp/bG3fPny2rJly0hrmO7krF7hiWppvCTZWlXLZ+rnO6QlSR3DTivNC44QJM03jhwkSR2GgySpw3CQJHUYDpKkDsNBktRhOEiSOgwHSVKH4SBJ6jAcJEkdhoMkqcNwkCR1eG8l7VN+rKh0YHLkIEnqMBwkSR2GgySpw3MOffzcBknqceQgSeowHCRJHYaDJKnDcJAkdRgOkqSOocMhySFJ7k3ytba+NMnmJI8luTHJYa398LY+0bYv6XuMS1v7o0nOHrYmSdJw5mLkcDHwSN/6p4Arq2oZ8CywprWvAZ6tql8Crmz9SHISsAo4GVgBfD7JIXNQlyRploYKhySLgXcDV7f1AO8Cbm5d1gPnt+WVbZ22/czWfyWwoapeqqongAngtGHqkiQNZ9iRw6eBPwD+oa2/AXiuqna29UlgUVteBDwF0LY/3/r/rH3APpKkEZh1OCR5D/BMVW3tbx7QtWbYtrt9dv2Za5NsSbJl+/bte1WvJGnPDTNyeDtwXpLvAxvoTSd9GliQZOq2HIuBp9vyJHACQNv+C8CO/vYB+7xKVa2rquVVtXzhwoVDlC5J2p1Zh0NVXVpVi6tqCb0Tyt+qqg8CdwDvbd1WA7e05Y1tnbb9W1VVrX1Vu5ppKbAM+O5s65IkDW9f3Hjvo8CGJJ8A7gWuae3XAF9OMkFvxLAKoKoeSnIT8DCwE7ioql7eB3VJkvbQnIRDVX0b+HZbfpwBVxtV1d8DF0yz/xXAFXNRiyRpeL5DWpLUYThIkjoMB0lSh+EgSeowHCRJHYaDJKnDcJAkdRgOkqQOw0GS1GE4SJI6DAdJUse+uPHe2PvK5idHXYIkjTVHDpKkDsNBktRhOEiSOgwHSVKH4SBJ6jAcJEkd8/JSVo3edJcTf+D0E/dzJZIGceQgSeowHCRJHYaDJKnDcJAkdcw6HJKckOSOJI8keSjJxa396CSbkjzWvh/V2pPks0kmktyf5NS+x1rd+j+WZPXwv5YkaRjDjBx2Ar9XVb8MnAFclOQk4BLg9qpaBtze1gHOAZa1r7XAVdALE+By4HTgNODyqUCRJI3GrMOhqrZV1T1t+QXgEWARsBJY37qtB85vyyuB66rnLmBBkuOBs4FNVbWjqp4FNgErZluXJGl4c/I+hyRLgLcBm4Hjqmob9AIkybGt2yLgqb7dJlvbdO2Dfs5aeqMOTjzR6+ElHTgOtPf2DH1COslrgT8Dfqeq/nZ3XQe01W7au41V66pqeVUtX7hw4d4XK0naI0ONHJL8HL1guL6qvtqaf5jk+DZqOB54prVPAif07b4YeLq1v3OX9m8PU5cOXAfaX1fSwWqYq5UCXAM8UlV/0rdpIzB1xdFq4Ja+9gvbVUtnAM+36afbgLOSHNVORJ/V2iRJIzLMyOHtwL8GHkhyX2v7j8AngZuSrAGeBC5o224FzgUmgBeBDwFU1Y4kHwfubv0+VlU7hqhLkjSkWYdDVf1vBp8vADhzQP8CLprmsa4Frp1tLTr47e5zv51ykuae75CWJHUYDpKkDsNBktThh/3ooOVlsdLsOXKQJHU4ctC844hCmpnhIDWGhvQKp5UkSR2OHKQZOKLQfOTIQZLU4chBB7zd3VpjFD/XEcX8Nqr/j3PNcJDmmKGhg4HhIO0nhoYOJIaDNGJ7Ow1hmGh/MBykg4ijE80Vw0GSZuFgOfE8HcNBmgfm6oXMEcj8YThI2mN7O211MExzHewjhOkYDtIBZhxfrOaqpv0RJuN4/MaR4SBpv9vXYaLhefsMSVKH4SBJ6hibaaUkK4DPAIcAV1fVJ0dckqQx4fTR/jcWI4ckhwCfA84BTgLen+Sk0VYlSfPXWIQDcBowUVWPV9VPgA3AyhHXJEnz1rhMKy0CnupbnwROH1EtkrTfjOu9tcYlHDKgrTqdkrXA2rb64ySP7uXPOQb40V7uMwrWObesc25Z59zaqzo/OPzP+yd70mlcwmESOKFvfTHw9K6dqmodsG62PyTJlqpaPtv99xfrnFvWObesc26Na53jcs7hbmBZkqVJDgNWARtHXJMkzVtjMXKoqp1Jfhu4jd6lrNdW1UMjLkuS5q2xCAeAqroVuHUf/5hZT0ntZ9Y5t6xzblnn3BrLOlPVOe8rSZrnxuWcgyRpjMyLcEiyIsmjSSaSXDLqeqYkOSHJHUkeSfJQkotb+9FJNiV5rH0/atS1Qu+d7EnuTfK1tr40yeZW543tYoJR17ggyc1JvteO66+N4/FM8rvt3/zBJDck+flxOJ5Jrk3yTJIH+9oGHr/0fLY9r+5PcuqI6/yj9u9+f5I/T7Kgb9ulrc5Hk5w9yjr7tv1+kkpyTFsf2fEc5KAPhzG/NcdO4Peq6peBM4CLWm2XALdX1TLg9rY+Di4GHulb/xRwZavzWWDNSKp6tc8A36iqNwNvoVfvWB3PJIuAjwDLq+oUehdhrGI8jueXgBW7tE13/M4BlrWvtcBV+6lGGFznJuCUqvoV4K+ASwHac2oVcHLb5/PtdWFUdZLkBOBfAv3vgBvl8ew46MOBMb41R1Vtq6p72vIL9F7IFtGrb33rth44fzQVviLJYuDdwNVtPcC7gJtbl5HXmeT1wDuAawCq6idV9RxjeDzpXQxyRJJDgSOBbYzB8ayq7wA7dmme7vitBK6rnruABUmOH1WdVfXNqtrZVu+i936pqTo3VNVLVfUEMEHvdWEkdTZXAn/Aq9/sO7LjOch8CIdBt+ZYNKJappVkCfA2YDNwXFVtg16AAMeOrrKf+TS9/8z/0NbfADzX92Qch+P6RmA78MU2/XV1ktcwZsezqn4A/DG9vxq3Ac8DWxm/4zlluuM3zs+tDwNfb8tjVWeS84AfVNVf7rJprOqcD+GwR7fmGKUkrwX+DPidqvrbUdezqyTvAZ6pqq39zQO6jvq4HgqcClxVVW8D/o7xmZL7mTZnvxJYCvwi8Bp6Uwq7GvXxnMk4/h8gyWX0pmyvn2oa0G0kdSY5ErgM+M+DNg9oG9nxnA/hsEe35hiVJD9HLxiur6qvtuYfTg0n2/dnRlVf83bgvCTfpzct9y56I4kFbVoExuO4TgKTVbW5rd9MLyzG7Xj+BvBEVW2vqp8CXwV+nfE7nlOmO35j99xKshp4D/DBeuU6/XGq85/S+6PgL9vzaTFwT5J/zHjVOS/CYWxvzdHm7a8BHqmqP+nbtBFY3ZZXA7fs79r6VdWlVbW4qpbQO37fqqoPAncA723dxqHOvwGeSvKm1nQm8DBjdjzpTSedkeTI9n9gqs6xOp59pjt+G4EL21U2ZwDPT00/jUJ6Hxj2UeC8qnqxb9NGYFWSw5MspXfC97ujqLGqHqiqY6tqSXs+TQKntv+7Y3U8qaqD/gs4l97VC/8XuGzU9fTV9S/oDRvvB+5rX+fSm8+/HXisfT961LX21fxO4Gtt+Y30nmQTwJ8Ch49BfW8FtrRj+j+Bo8bxeAJ/CHwPeBD4MnD4OBxP4AZ650F+Su+Fa810x4/eNMjn2vPqAXpXX42yzgl6c/ZTz6Uv9PW/rNX5KHDOKOvcZfv3gWNGfTwHffkOaUlSx3yYVpIk7SXDQZLUYThIkjoMB0lSh+EgSeowHKQBkvx4HzzmW5Oc27f+X5L8/lz/HGkuGA7S/vNWeu9jkcae4SDNIMl/SHJ3u8f+H7a2Jel9XsT/aJ/L8M0kR7Rt/7z1vbN9xsCD7d35HwPel+S+JO9rD39Skm8neTzJR0b0K0odhoO0G0nOone7hdPo/eX/q0ne0TYvAz5XVScDzwG/2dq/CPybqvo14GXo3T6c3s3Wbqyqt1bVja3vm4Gz2+Nf3u61JY2c4SDt3lnt617gHnov5svatieq6r62vBVY0j597HVV9X9a+1dmePy/qN7nDPyI3g3tjpvT6qVZOnTmLtK8FuC/VtV/f1Vj7/M3Xuprehk4gsG3Xd6dXR/D56TGgiMHafduAz7cPnODJIuSTPthQVX1LPBCu6sm9O5iO+UF4HX7rFJpDhkO0m5U1TfpTQ3dmeQBep8RMdML/BpgXZI76Y0knm/td9A7Ad1/QloaS96VVZpjSV5bVT9uy5cAx1fVxSMuS9orzm9Kc+/dSS6l9/z6a+C3RluOtPccOUiSOjznIEnqMBwkSR2GgySpw3CQJHUYDpKkDsNBktTx/wHIYVvGUsuiiAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1114b38d0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 전체 분포\n",
    "sns.distplot(reviews.length, kde=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x117067390>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1114dcd68>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Positive review에 대한 분포\n",
    "sns.distplot(reviews[reviews.label==1]['length'], kde=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x10c88bda0>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11705f898>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Positive review에 대한 분포\n",
    "sns.distplot(reviews[reviews.label==0]['length'], kde=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "count    150000.000000\n",
       "mean         35.203453\n",
       "std          29.531983\n",
       "min           1.000000\n",
       "25%          16.000000\n",
       "50%          27.000000\n",
       "75%          42.000000\n",
       "max         146.000000\n",
       "Name: length, dtype: float64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews.length.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\"스폰으로 먹고사는 방송이라 어쩔수 없다고 하지만. 이건 그냥 비현실적인 자동차만;...독일3사&슈퍼카 홍보 프로그램도 아니구.대중적인 자동차 방송으로 이루어 졌으면 합니다. 보는내내 \"\"카탈로그 책자\"\"를 \"\"동영상으로 보여주는 방송\"\" 같아서 씁쓸하네요.!\"'"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews[reviews.length == 146]['document'].iloc[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "문장길이 분포는 긍정/부정에 대한 차이가 크게 없다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Text Pre-Processing\n",
    " - SKlearn에서 제공해주는 라이브러리를 사용할 것이며 해당 *parse* 메소드를 라이브러리에 넘겨줄 것이다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 형태소 분석기\n",
    "import konlpy\n",
    "from konlpy.tag import Okt\n",
    "\n",
    "okt = Okt()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def parse(s):\n",
    "    try:\n",
    "        return okt.nouns(s)\n",
    "    except:\n",
    "        return []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "reviews['parsed_doc'] = reviews.document.apply(parse)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "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>id</th>\n",
       "      <th>document</th>\n",
       "      <th>label</th>\n",
       "      <th>length</th>\n",
       "      <th>parsed_doc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9976970</td>\n",
       "      <td>아 더빙.. 진짜 짜증나네요 목소리</td>\n",
       "      <td>0</td>\n",
       "      <td>19</td>\n",
       "      <td>[더빙, 진짜, 목소리]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3819312</td>\n",
       "      <td>흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나</td>\n",
       "      <td>1</td>\n",
       "      <td>33</td>\n",
       "      <td>[흠, 포스터, 보고, 초딩, 영화, 줄, 오버, 연기]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10265843</td>\n",
       "      <td>너무재밓었다그래서보는것을추천한다</td>\n",
       "      <td>0</td>\n",
       "      <td>17</td>\n",
       "      <td>[무재, 밓었, 다그, 래서, 추천]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9045019</td>\n",
       "      <td>교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정</td>\n",
       "      <td>0</td>\n",
       "      <td>29</td>\n",
       "      <td>[교도소, 이야기, 구먼, 재미, 평점, 조정]</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6483659</td>\n",
       "      <td>사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...</td>\n",
       "      <td>1</td>\n",
       "      <td>61</td>\n",
       "      <td>[몬페, 의, 연기, 영화, 스파이더맨, 커스틴, 던스트]</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         id                                           document  label  length  \\\n",
       "0   9976970                                아 더빙.. 진짜 짜증나네요 목소리      0      19   \n",
       "1   3819312                  흠...포스터보고 초딩영화줄....오버연기조차 가볍지 않구나      1      33   \n",
       "2  10265843                                  너무재밓었다그래서보는것을추천한다      0      17   \n",
       "3   9045019                      교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정      0      29   \n",
       "4   6483659  사이몬페그의 익살스런 연기가 돋보였던 영화!스파이더맨에서 늙어보이기만 했던 커스틴 ...      1      61   \n",
       "\n",
       "                         parsed_doc  \n",
       "0                     [더빙, 진짜, 목소리]  \n",
       "1   [흠, 포스터, 보고, 초딩, 영화, 줄, 오버, 연기]  \n",
       "2              [무재, 밓었, 다그, 래서, 추천]  \n",
       "3        [교도소, 이야기, 구먼, 재미, 평점, 조정]  \n",
       "4  [몬페, 의, 연기, 영화, 스파이더맨, 커스틴, 던스트]  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Vectorization (Bag-of-Words)\n",
    "- Bag of words count를 vector로 만든다.\n",
    "- Term frequency를 사용한다.\n",
    "- ex) 전체 corpus가 `cat, love, i, do, like, him, you`라고 하자. 문장이 `i love you only you` 이면, 이 문장의 bag of word vector는 `(0, 1, 1, 0, 0, 0, 2)` 이다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "bow_transformer = CountVectorizer(analyzer=parse) \\\n",
    "    .fit(reviews.document)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "38648"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Corpus\n",
    "len(bow_transformer.vocabulary_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'교도소 이야기구먼 ..솔직히 재미는 없다..평점 조정'"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample = reviews.document.iloc[3]\n",
    "sample"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "sample_bow = bow_transformer.transform([sample])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  (0, 2509)\t1\n",
      "  (0, 2630)\t1\n",
      "  (0, 26019)\t1\n",
      "  (0, 27667)\t1\n",
      "  (0, 29259)\t1\n",
      "  (0, 35255)\t1\n"
     ]
    }
   ],
   "source": [
    "print(sample_bow)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "교도소\n",
      "구먼\n",
      "이야기\n"
     ]
    }
   ],
   "source": [
    "print(bow_transformer.get_feature_names()[2509])\n",
    "print(bow_transformer.get_feature_names()[2630])\n",
    "print(bow_transformer.get_feature_names()[26019])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "reviews_bow = bow_transformer.transform(reviews.document)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(150000, 38648)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Sparse matrix의 크기\n",
    "reviews_bow.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "847810"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Non-zero의 발생 횟수\n",
    "reviews_bow.nnz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.015\n"
     ]
    }
   ],
   "source": [
    "# Sparsity\n",
    "sparsity = 100.0 * \\\n",
    "    reviews_bow.nnz / (reviews_bow.shape[0] * reviews_bow.shape[1])\n",
    "print(round(sparsity, 3))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Normalization of vector (TF-IDF)\n",
    "\n",
    "### 4.1. TF (Term Frequency)\n",
    "Term Frequency, which measures how frequently a term occurs in a document. Since every document is different in length, it is possible that a term would appear much more times in long documents than shorter ones. Thus, the term frequency is often divided by the document length as a way of normalization:\n",
    "\n",
    "$$TF(t)=\\frac{\\text{The number of times term t appears in a document}}{\\text{The total number of terms in the document}}$$\n",
    "\n",
    "### 4.2. IDF (Inverse Document Frequency)\n",
    "Inverse Document Frequency, which measures how important a term is. While computing TF, all terms are considered equally important. However it is known that certain terms, such as \"is\", \"of\", and \"that\", may appear a lot of times but have little importance. Thus we need to weigh down the frequent terms while scale up the rare ones, by computing the following:\n",
    "\n",
    "$$IDF(t)=log_{e}(\\frac{\\text{The total number of documents}}{\\text{The number of documents with term t in it}})$$"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(예시)\n",
    "100개의 단어를 포함하고 있는 document를 생각해보자. \"cat\"이라는 단어가 그 document에 3번 나온다고 가정하자. 10,000,000의 문서 중에서 1,000의 문서에서만 \"cat\"이 나온다고 가정하자.\n",
    "- TF : $3/100 = 0.03$\n",
    "- IDF : $log_e(10,000,000 / 1,000)=4$\n",
    "- TF-IDF : $0.03 \\times 4 = 0.12$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "tfidf_transformer = TfidfTransformer().fit(reviews_bow)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "  (0, 35255)\t0.2229955053854873\n",
      "  (0, 29259)\t0.48387295041594\n",
      "  (0, 27667)\t0.24505937911070386\n",
      "  (0, 26019)\t0.27807555746194873\n",
      "  (0, 2630)\t0.5469932440848155\n",
      "  (0, 2509)\t0.5287326814292047\n"
     ]
    }
   ],
   "source": [
    "sample_tfidf = tfidf_transformer.transform(sample_bow)\n",
    "print(sample_tfidf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10.08518389566662\n",
      "5.304085093270837\n"
     ]
    }
   ],
   "source": [
    "print(tfidf_transformer.idf_[bow_transformer.vocabulary_['교도소']])\n",
    "print(tfidf_transformer.idf_[bow_transformer.vocabulary_['이야기']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(150000, 38648)"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "reviews_tfidf = tfidf_transformer.transform(reviews_bow)\n",
    "reviews_tfidf.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "모델을 학습한다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "sentiment_detect_model = MultinomialNB().fit(reviews_tfidf, reviews['label'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "샘플데이터에 대하여 모델을 테스트해보자."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 4번째 샘플에 대하여 예측값을 구해보자.\n",
    "sentiment_detect_model.predict(sample_tfidf)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 4번째 샘플의 label\n",
    "reviews['label'].iloc[3]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 1, 0, 1, 0, 0, 0, 1, 1])"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 학습 데이터에 대하여 정확도를 체크해보자.\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "train_preds = sentiment_detect_model.predict(reviews_tfidf)\n",
    "train_preds[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_targets = reviews['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.81092"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(train_targets, train_preds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test set에 대한 bag of words vertors\n",
    "test_reviews_bow = bow_transformer.transform(test_reviews.document)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test set에 대한 tf-idf vectors\n",
    "test_reviews_tfidf = tfidf_transformer.transform(test_reviews_bow)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "# NB 모델에서 예측\n",
    "test_preds = sentiment_detect_model.predict(test_reviews_tfidf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Test 데이터 참 값\n",
    "test_targets = test_reviews['label'].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.76656"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Test data에서의 성능\n",
    "accuracy_score(test_targets, test_preds)"
   ]
  }
 ],
 "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": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}