notebooks/dataset_exploration.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 초기 데이터셋 분석\n",
    "\n",
    "이 노트북에서 writers.stackexchange.com의 실제 질문 데이터셋을 빠르게 살펴 보겠습니다.\n",
    "이 데이터셋은 원래 [아카이브](https://archive.org/details/stackexchange)에서 받은 것입니다.\n",
    "\n",
    "먼저 데이터를 로드합니다. 다른 csv 파일을 로드한다면 `ml_editor` 파이썬 패키지를 사용해 원시 xml을 전처리해야 합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "from tqdm import tqdm\n",
    "from bs4 import BeautifulSoup\n",
    "import xml.etree.ElementTree as ElT\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.patches import Rectangle\n",
    "import pandas as pd\n",
    "\n",
    "from pathlib import Path\n",
    "import sys\n",
    "sys.path.append(\"..\")\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "df = pd.read_csv(Path('../data/writers.csv'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 데이터 포맷\n",
    "\n",
    "데이터를 어떻게 포맷팅할지 생각해 보죠. 무엇보다도 모델에 어떤 레이블을 제공할지 결정해야 합니다.\n",
    "\n",
    "모델은 질문의 품질을 측정합니다. 이를 위해 다음을 사용할 수 있습니다:\n",
    "- 질문의 업보트(upvote)의 수\n",
    "- 질문의 답변 수, 또는 답변을 받았는지 여부\n",
    "- 채택된 답변이 있는지 여부\n",
    "\n",
    "먼저 데이터셋을 질문과 관련된 답변으로 구성하겠습니다.\n",
    "\n",
    "누락된 값을 채우고 나중에 사용할 두 개의 특성(`text_len`와 `is_question`)을 추가합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 손쉬운 처리를 위해 타입을 바꿉니다\n",
    "df[\"AnswerCount\"] = df[\"AnswerCount\"].fillna(-1)\n",
    "df[\"AnswerCount\"] = df[\"AnswerCount\"].astype(int)\n",
    "df[\"PostTypeId\"] = df[\"PostTypeId\"].astype(int)\n",
    "df[\"Id\"] = df[\"Id\"].astype(int)\n",
    "df.set_index(\"Id\", inplace=True, drop=False)\n",
    "\n",
    "# 포스트의 길이를 추가합니다.\n",
    "df[\"full_text\"] = df[\"Title\"].str.cat(df[\"body_text\"], sep=\" \", na_rep=\"\")\n",
    "df[\"text_len\"] = df[\"full_text\"].str.len()\n",
    "\n",
    "# 질문은 포스트 타입이 1입니다\n",
    "df[\"is_question\"] = df[\"PostTypeId\"] == 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 데이터 품질\n",
    "\n",
    "다음 질문에 답을 하면서 이 데이터셋의 데이터 품질을 조사해 보죠.\n",
    "\n",
    "- 얼마나 많은 데이터가 누락되었나요?\n",
    "- 텍스트의 품질은 무엇인가요?\n",
    "- 대답이 질문과 일치하나요?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 34330 entries, 1 to 42885\n",
      "Data columns (total 26 columns):\n",
      " #   Column                 Non-Null Count  Dtype  \n",
      "---  ------                 --------------  -----  \n",
      " 0   Unnamed: 0             34330 non-null  int64  \n",
      " 1   AcceptedAnswerId       4124 non-null   float64\n",
      " 2   AnswerCount            34330 non-null  int64  \n",
      " 3   Body                   34256 non-null  object \n",
      " 4   ClosedDate             969 non-null    object \n",
      " 5   CommentCount           34330 non-null  int64  \n",
      " 6   CommunityOwnedDate     186 non-null    object \n",
      " 7   CreationDate           34330 non-null  object \n",
      " 8   FavoriteCount          3307 non-null   float64\n",
      " 9   Id                     34330 non-null  int64  \n",
      " 10  LastActivityDate       34330 non-null  object \n",
      " 11  LastEditDate           11201 non-null  object \n",
      " 12  LastEditorDisplayName  614 non-null    object \n",
      " 13  LastEditorUserId       10648 non-null  float64\n",
      " 14  OwnerDisplayName       1976 non-null   object \n",
      " 15  OwnerUserId            32792 non-null  float64\n",
      " 16  ParentId               25679 non-null  float64\n",
      " 17  PostTypeId             34330 non-null  int64  \n",
      " 18  Score                  34330 non-null  int64  \n",
      " 19  Tags                   7971 non-null   object \n",
      " 20  Title                  7971 non-null   object \n",
      " 21  ViewCount              7971 non-null   float64\n",
      " 22  body_text              34256 non-null  object \n",
      " 23  full_text              34330 non-null  object \n",
      " 24  text_len               34330 non-null  int64  \n",
      " 25  is_question            34330 non-null  bool   \n",
      "dtypes: bool(1), float64(6), int64(7), object(12)\n",
      "memory usage: 6.8+ MB\n"
     ]
    }
   ],
   "source": [
    "# 열과 Null이 아닌 개수를 출력합니다\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "질문과 답변으로 구성된 34,000 이상의 포스트가 있습니다.\n",
    "\n",
    "`Body` 열을 보면 `34330 - 34256 = 74` 행이 null인 것 같습니다. 이를 삭제해야 하는지 알아보기 위해 이 행을 출력해 보죠."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "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>Unnamed: 0</th>\n",
       "      <th>AcceptedAnswerId</th>\n",
       "      <th>AnswerCount</th>\n",
       "      <th>Body</th>\n",
       "      <th>ClosedDate</th>\n",
       "      <th>CommentCount</th>\n",
       "      <th>CommunityOwnedDate</th>\n",
       "      <th>CreationDate</th>\n",
       "      <th>FavoriteCount</th>\n",
       "      <th>Id</th>\n",
       "      <th>...</th>\n",
       "      <th>ParentId</th>\n",
       "      <th>PostTypeId</th>\n",
       "      <th>Score</th>\n",
       "      <th>Tags</th>\n",
       "      <th>Title</th>\n",
       "      <th>ViewCount</th>\n",
       "      <th>body_text</th>\n",
       "      <th>full_text</th>\n",
       "      <th>text_len</th>\n",
       "      <th>is_question</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2145</th>\n",
       "      <td>1959</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2011-03-22T19:49:56.600</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2145</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2147</th>\n",
       "      <td>1961</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2011-03-22T19:51:05.897</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2147</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2215</th>\n",
       "      <td>2029</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2011-03-24T19:35:10.353</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2215</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2218</th>\n",
       "      <td>2032</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2011-03-24T19:41:38.677</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2218</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2225</th>\n",
       "      <td>2039</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2011-03-24T19:58:59.833</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2225</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</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>42223</th>\n",
       "      <td>33706</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2019-02-14T09:42:01.813</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42223</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42549</th>\n",
       "      <td>34006</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2019-02-24T05:18:39.920</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42549</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42551</th>\n",
       "      <td>34008</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2019-02-24T05:26:20.740</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42551</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42629</th>\n",
       "      <td>34085</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2019-02-25T22:45:38.457</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42629</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42787</th>\n",
       "      <td>34237</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2019-03-01T01:51:56.043</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42787</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td></td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>74 rows × 26 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       Unnamed: 0  AcceptedAnswerId  AnswerCount Body ClosedDate  \\\n",
       "Id                                                                 \n",
       "2145         1959               NaN           -1  NaN        NaN   \n",
       "2147         1961               NaN           -1  NaN        NaN   \n",
       "2215         2029               NaN           -1  NaN        NaN   \n",
       "2218         2032               NaN           -1  NaN        NaN   \n",
       "2225         2039               NaN           -1  NaN        NaN   \n",
       "...           ...               ...          ...  ...        ...   \n",
       "42223       33706               NaN           -1  NaN        NaN   \n",
       "42549       34006               NaN           -1  NaN        NaN   \n",
       "42551       34008               NaN           -1  NaN        NaN   \n",
       "42629       34085               NaN           -1  NaN        NaN   \n",
       "42787       34237               NaN           -1  NaN        NaN   \n",
       "\n",
       "       CommentCount CommunityOwnedDate             CreationDate  \\\n",
       "Id                                                                \n",
       "2145              0                NaN  2011-03-22T19:49:56.600   \n",
       "2147              0                NaN  2011-03-22T19:51:05.897   \n",
       "2215              0                NaN  2011-03-24T19:35:10.353   \n",
       "2218              0                NaN  2011-03-24T19:41:38.677   \n",
       "2225              0                NaN  2011-03-24T19:58:59.833   \n",
       "...             ...                ...                      ...   \n",
       "42223             0                NaN  2019-02-14T09:42:01.813   \n",
       "42549             0                NaN  2019-02-24T05:18:39.920   \n",
       "42551             0                NaN  2019-02-24T05:26:20.740   \n",
       "42629             0                NaN  2019-02-25T22:45:38.457   \n",
       "42787             0                NaN  2019-03-01T01:51:56.043   \n",
       "\n",
       "       FavoriteCount     Id  ... ParentId PostTypeId Score  Tags Title  \\\n",
       "Id                           ...                                         \n",
       "2145             NaN   2145  ...      NaN          5     0   NaN   NaN   \n",
       "2147             NaN   2147  ...      NaN          5     0   NaN   NaN   \n",
       "2215             NaN   2215  ...      NaN          5     0   NaN   NaN   \n",
       "2218             NaN   2218  ...      NaN          5     0   NaN   NaN   \n",
       "2225             NaN   2225  ...      NaN          5     0   NaN   NaN   \n",
       "...              ...    ...  ...      ...        ...   ...   ...   ...   \n",
       "42223            NaN  42223  ...      NaN          5     0   NaN   NaN   \n",
       "42549            NaN  42549  ...      NaN          5     0   NaN   NaN   \n",
       "42551            NaN  42551  ...      NaN          5     0   NaN   NaN   \n",
       "42629            NaN  42629  ...      NaN          5     0   NaN   NaN   \n",
       "42787            NaN  42787  ...      NaN          5     0   NaN   NaN   \n",
       "\n",
       "       ViewCount  body_text  full_text  text_len is_question  \n",
       "Id                                                            \n",
       "2145         NaN        NaN                    1       False  \n",
       "2147         NaN        NaN                    1       False  \n",
       "2215         NaN        NaN                    1       False  \n",
       "2218         NaN        NaN                    1       False  \n",
       "2225         NaN        NaN                    1       False  \n",
       "...          ...        ...        ...       ...         ...  \n",
       "42223        NaN        NaN                    1       False  \n",
       "42549        NaN        NaN                    1       False  \n",
       "42551        NaN        NaN                    1       False  \n",
       "42629        NaN        NaN                    1       False  \n",
       "42787        NaN        NaN                    1       False  \n",
       "\n",
       "[74 rows x 26 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df[\"Body\"].isna()]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "null인 body는 모두 PostTypeId가 4 또는 5입니다.\n",
    "\n",
    "아카이브와 함께  제공된 readme 파일에는 PostTypeId 1(질문)과 2(답변)만 언급되어 있습니다. 질문과 답변만 관심 대상이므로 `PostTypeId`가 1 또는 2가 아닌 모든 행을 삭제합니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 33650 entries, 1 to 42885\n",
      "Data columns (total 26 columns):\n",
      " #   Column                 Non-Null Count  Dtype  \n",
      "---  ------                 --------------  -----  \n",
      " 0   Unnamed: 0             33650 non-null  int64  \n",
      " 1   AcceptedAnswerId       4124 non-null   float64\n",
      " 2   AnswerCount            33650 non-null  int64  \n",
      " 3   Body                   33650 non-null  object \n",
      " 4   ClosedDate             969 non-null    object \n",
      " 5   CommentCount           33650 non-null  int64  \n",
      " 6   CommunityOwnedDate     186 non-null    object \n",
      " 7   CreationDate           33650 non-null  object \n",
      " 8   FavoriteCount          3307 non-null   float64\n",
      " 9   Id                     33650 non-null  int64  \n",
      " 10  LastActivityDate       33650 non-null  object \n",
      " 11  LastEditDate           10521 non-null  object \n",
      " 12  LastEditorDisplayName  606 non-null    object \n",
      " 13  LastEditorUserId       9975 non-null   float64\n",
      " 14  OwnerDisplayName       1971 non-null   object \n",
      " 15  OwnerUserId            32117 non-null  float64\n",
      " 16  ParentId               25679 non-null  float64\n",
      " 17  PostTypeId             33650 non-null  int64  \n",
      " 18  Score                  33650 non-null  int64  \n",
      " 19  Tags                   7971 non-null   object \n",
      " 20  Title                  7971 non-null   object \n",
      " 21  ViewCount              7971 non-null   float64\n",
      " 22  body_text              33650 non-null  object \n",
      " 23  full_text              33650 non-null  object \n",
      " 24  text_len               33650 non-null  int64  \n",
      " 25  is_question            33650 non-null  bool   \n",
      "dtypes: bool(1), float64(6), int64(7), object(12)\n",
      "memory usage: 6.7+ MB\n"
     ]
    }
   ],
   "source": [
    "df = df[df[\"PostTypeId\"].isin([1,2])]\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이제 몇 개의 질문과 답변을 출력하고 두 내용이 일치하는지, 읽을 수 있는지 확인해 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>body_text</th>\n",
       "      <th>body_text_answer</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when re...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>What kind of story is better suited for each point of view? Are there advantages or disadvantages inherent to them?\\nFor example, writing in the first person you are always following a character, while in the third person you can \"jump\" between story lines.\\n</td>\n",
       "      <td>With a story in first person, you are intending the reader to become much more attached to the main character.  Since the reader sees what that character sees and feels what that character feels, the reader will have an emotional investment in that character.  Third person does not have this close tie; a reader can become emotionally invested but it will not be as strong as it will be in first person.\\nContrarily, you cannot have multiple point characters when you use first person without ex...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>I finished my novel, and everyone I've talked to says I need an agent. How do I find one?\\n</td>\n",
       "      <td>Try and find a list of agents who write in your genre. Check out their websites!\\nFind out if they are accepting new clients. If they aren't, then check out another agent. But if they are, try sending them a few chapters from your story, a brief, and a short cover letter asking them to represent you.\\nIn the cover letter mention your previous publication credits. If sent via post, then I suggest you give them a means of reply, whether it be an email or a stamped, addressed envelope.\\nAgents ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                                                                                                                                                                                                                                                                            body_text  \\\n",
       "Id                                                                                                                                                                                                                                                                                                                      \n",
       "1   I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "2                                                 What kind of story is better suited for each point of view? Are there advantages or disadvantages inherent to them?\\nFor example, writing in the first person you are always following a character, while in the third person you can \"jump\" between story lines.\\n   \n",
       "3                                                                                                                                                                                                                         I finished my novel, and everyone I've talked to says I need an agent. How do I find one?\\n   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       body_text_answer  \n",
       "Id                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       \n",
       "1   When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when re...  \n",
       "2   With a story in first person, you are intending the reader to become much more attached to the main character.  Since the reader sees what that character sees and feels what that character feels, the reader will have an emotional investment in that character.  Third person does not have this close tie; a reader can become emotionally invested but it will not be as strong as it will be in first person.\\nContrarily, you cannot have multiple point characters when you use first person without ex...  \n",
       "3   Try and find a list of agents who write in your genre. Check out their websites!\\nFind out if they are accepting new clients. If they aren't, then check out another agent. But if they are, try sending them a few chapters from your story, a brief, and a short cover letter asking them to represent you.\\nIn the cover letter mention your previous publication credits. If sent via post, then I suggest you give them a means of reply, whether it be an email or a stamped, addressed envelope.\\nAgents ...  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "questions_with_accepted_answers = df[df[\"is_question\"] & ~(df[\"AcceptedAnswerId\"].isna())]\n",
    "q_and_a = questions_with_accepted_answers.join(df[[\"body_text\"]], on=\"AcceptedAnswerId\", how=\"left\", rsuffix=\"_answer\")\n",
    "\n",
    "# 모든 데이터를 출력하기 위해 이 옵션을 설정합니다\n",
    "pd.options.display.max_colwidth = 500\n",
    "q_and_a[[\"body_text\", \"body_text_answer\"]][:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "네 좋습니다. **질문과 답변이 잘 맞는 것 같군요.** 개행문자(\"\\n\")를 제외하면 이 텍스트는 논리적이고 사람이 읽을 수 있습니다.\n",
    "\n",
    "질문이 대답을 받았는지 정보를 각 행에 추가하겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.join(df[[\"Id\", \"Title\", \"body_text\", \"text_len\", \"Score\", \"AcceptedAnswerId\"]], on=\"ParentId\", how=\"left\", rsuffix='_question')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "이렇게 하면 주어진 질문에 대한 모든 답변을 확인할 수 있습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>Unnamed: 0</th>\n",
       "      <th>AcceptedAnswerId</th>\n",
       "      <th>AnswerCount</th>\n",
       "      <th>Body</th>\n",
       "      <th>ClosedDate</th>\n",
       "      <th>CommentCount</th>\n",
       "      <th>CommunityOwnedDate</th>\n",
       "      <th>CreationDate</th>\n",
       "      <th>FavoriteCount</th>\n",
       "      <th>Id</th>\n",
       "      <th>...</th>\n",
       "      <th>body_text</th>\n",
       "      <th>full_text</th>\n",
       "      <th>text_len</th>\n",
       "      <th>is_question</th>\n",
       "      <th>Id_question</th>\n",
       "      <th>Title_question</th>\n",
       "      <th>body_text_question</th>\n",
       "      <th>text_len_question</th>\n",
       "      <th>Score_question</th>\n",
       "      <th>AcceptedAnswerId_question</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>&lt;p&gt;BeginningWriters.com has some good articles for beginning writers.&lt;/p&gt;\\n\\n&lt;p&gt;&lt;a href=\"http://beginningwriters.com/\"&gt;http://beginningwriters.com/&lt;/a&gt;&lt;/p&gt;\\n</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2010-11-18T20:45:45.533</td>\n",
       "      <td>NaN</td>\n",
       "      <td>8</td>\n",
       "      <td>...</td>\n",
       "      <td>BeginningWriters.com has some good articles for beginning writers.\\nhttp://beginningwriters.com/\\n</td>\n",
       "      <td>BeginningWriters.com has some good articles for beginning writers.\\nhttp://beginningwriters.com/\\n</td>\n",
       "      <td>97</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>What are some online guides for starting writers?</td>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>352.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>&lt;p&gt;When I'm thinking about where I learned most how to write, I think that &lt;em&gt;reading&lt;/em&gt; was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I ...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2010-11-18T20:50:16.683</td>\n",
       "      <td>NaN</td>\n",
       "      <td>15</td>\n",
       "      <td>...</td>\n",
       "      <td>When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when re...</td>\n",
       "      <td>When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when r...</td>\n",
       "      <td>962</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>What are some online guides for starting writers?</td>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>352.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>65</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>&lt;p&gt;I particularly like &lt;a href=\"http://www.advancedfictionwriting.com/art/snowflake.php\" rel=\"nofollow\" rel=\"nofollow\"&gt;The Snowflake Method&lt;/a&gt; and most of what Randy writes, including his book.&lt;/p&gt;\\n</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2010-11-18T23:19:33.157</td>\n",
       "      <td>NaN</td>\n",
       "      <td>79</td>\n",
       "      <td>...</td>\n",
       "      <td>I particularly like The Snowflake Method and most of what Randy writes, including his book.\\n</td>\n",
       "      <td>I particularly like The Snowflake Method and most of what Randy writes, including his book.\\n</td>\n",
       "      <td>93</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>What are some online guides for starting writers?</td>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>352.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>106</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>&lt;p&gt;Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?&lt;/p&gt;\\n\\n&lt;p&gt;&lt;strong&gt;&lt;a href=\"http://www.writingexcuses.com/\" rel=\"nofollow\"&gt;Writing Excuses&lt;/a&gt;&lt;/strong&gt; got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on &lt;a href=\"http://mgowen.com/2010/11/12/writing-excuses/\" rel=\"nofollow\"&gt;my blog&lt;/a&gt;)&lt;/p&gt;\\n</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2010-11-19T04:25:03.407</td>\n",
       "      <td>NaN</td>\n",
       "      <td>123</td>\n",
       "      <td>...</td>\n",
       "      <td>Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?\\nWriting Excuses got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on my blog)\\n</td>\n",
       "      <td>Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?\\nWriting Excuses got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on my blog)\\n</td>\n",
       "      <td>321</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>What are some online guides for starting writers?</td>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>352.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>254</th>\n",
       "      <td>231</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "      <td>&lt;p&gt;&lt;a href=\"http://www.writersdigest.com/GeneralMenu/\" rel=\"nofollow\"&gt;Writer's Digest&lt;/a&gt; has many good articles.&lt;/p&gt;\\n</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2010-11-19T20:09:56.347</td>\n",
       "      <td>NaN</td>\n",
       "      <td>254</td>\n",
       "      <td>...</td>\n",
       "      <td>Writer's Digest has many good articles.\\n</td>\n",
       "      <td>Writer's Digest has many good articles.\\n</td>\n",
       "      <td>41</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>What are some online guides for starting writers?</td>\n",
       "      <td>I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n</td>\n",
       "      <td>352.0</td>\n",
       "      <td>32.0</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 32 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     Unnamed: 0  AcceptedAnswerId  AnswerCount  \\\n",
       "Id                                               \n",
       "8             5               NaN           -1   \n",
       "15           10               NaN           -1   \n",
       "79           65               NaN           -1   \n",
       "123         106               NaN           -1   \n",
       "254         231               NaN           -1   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    Body  \\\n",
       "Id                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         \n",
       "8                                                                                                                                                                                                                                                                                                                                                          <p>BeginningWriters.com has some good articles for beginning writers.</p>\\n\\n<p><a href=\"http://beginningwriters.com/\">http://beginningwriters.com/</a></p>\\n   \n",
       "15   <p>When I'm thinking about where I learned most how to write, I think that <em>reading</em> was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I ...   \n",
       "79                                                                                                                                                                                                                                                                                                              <p>I particularly like <a href=\"http://www.advancedfictionwriting.com/art/snowflake.php\" rel=\"nofollow\" rel=\"nofollow\">The Snowflake Method</a> and most of what Randy writes, including his book.</p>\\n   \n",
       "123           <p>Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?</p>\\n\\n<p><strong><a href=\"http://www.writingexcuses.com/\" rel=\"nofollow\">Writing Excuses</a></strong> got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on <a href=\"http://mgowen.com/2010/11/12/writing-excuses/\" rel=\"nofollow\">my blog</a>)</p>\\n   \n",
       "254                                                                                                                                                                                                                                                                                                                                                                                              <p><a href=\"http://www.writersdigest.com/GeneralMenu/\" rel=\"nofollow\">Writer's Digest</a> has many good articles.</p>\\n   \n",
       "\n",
       "    ClosedDate  CommentCount CommunityOwnedDate             CreationDate  \\\n",
       "Id                                                                         \n",
       "8          NaN             0                NaN  2010-11-18T20:45:45.533   \n",
       "15         NaN             3                NaN  2010-11-18T20:50:16.683   \n",
       "79         NaN             0                NaN  2010-11-18T23:19:33.157   \n",
       "123        NaN             0                NaN  2010-11-19T04:25:03.407   \n",
       "254        NaN             0                NaN  2010-11-19T20:09:56.347   \n",
       "\n",
       "     FavoriteCount   Id  ...  \\\n",
       "Id                       ...   \n",
       "8              NaN    8  ...   \n",
       "15             NaN   15  ...   \n",
       "79             NaN   79  ...   \n",
       "123            NaN  123  ...   \n",
       "254            NaN  254  ...   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               body_text  \\\n",
       "Id                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         \n",
       "8                                                                                                                                                                                                                                                                                                                                                                                                                     BeginningWriters.com has some good articles for beginning writers.\\nhttp://beginningwriters.com/\\n   \n",
       "15   When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when re...   \n",
       "79                                                                                                                                                                                                                                                                                                                                                                                                                         I particularly like The Snowflake Method and most of what Randy writes, including his book.\\n   \n",
       "123                                                                                                                                                                                   Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?\\nWriting Excuses got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on my blog)\\n   \n",
       "254                                                                                                                                                                                                                                                                                                                                                                                                                                                                            Writer's Digest has many good articles.\\n   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               full_text  \\\n",
       "Id                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         \n",
       "8                                                                                                                                                                                                                                                                                                                                                                                                                     BeginningWriters.com has some good articles for beginning writers.\\nhttp://beginningwriters.com/\\n   \n",
       "15    When I'm thinking about where I learned most how to write, I think that reading was the most important guide to me. This may sound silly, but by reading good written newspaper articles (facts, opinions, scientific articles and most of all, criticisms of films and music), I learned how others did the job, what works and what doesn't. In my own writing, I try to mimic other people's styles that I liked. Moreover, I learn new things by reading, giving me a broader background that I need when r...   \n",
       "79                                                                                                                                                                                                                                                                                                                                                                                                                         I particularly like The Snowflake Method and most of what Randy writes, including his book.\\n   \n",
       "123                                                                                                                                                                                   Reading a lot and loving to read are most important of course, but have you tried writing-related podcasts?\\nWriting Excuses got me into writing fiction again. It's a short podcast about how to write fiction well and how to \"break in\" to the industry, from 3 guys who do and did. (I wrote a bit more about it on my blog)\\n   \n",
       "254                                                                                                                                                                                                                                                                                                                                                                                                                                                                            Writer's Digest has many good articles.\\n   \n",
       "\n",
       "    text_len  is_question Id_question  \\\n",
       "Id                                      \n",
       "8         97        False         1.0   \n",
       "15       962        False         1.0   \n",
       "79        93        False         1.0   \n",
       "123      321        False         1.0   \n",
       "254       41        False         1.0   \n",
       "\n",
       "                                        Title_question  \\\n",
       "Id                                                       \n",
       "8    What are some online guides for starting writers?   \n",
       "15   What are some online guides for starting writers?   \n",
       "79   What are some online guides for starting writers?   \n",
       "123  What are some online guides for starting writers?   \n",
       "254  What are some online guides for starting writers?   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                    body_text_question  \\\n",
       "Id                                                                                                                                                                                                                                                                                                                       \n",
       "8    I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "15   I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "79   I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "123  I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "254  I've always wanted to start writing (in a totally amateur way), but whenever I want to start something I instantly get blocked having a lot of questions and doubts.\\nAre there some resources on how to start becoming a writer?\\nI'm thinking something with tips and easy exercises to get the ball rolling.\\n   \n",
       "\n",
       "     text_len_question  Score_question AcceptedAnswerId_question  \n",
       "Id                                                                \n",
       "8                352.0            32.0                      15.0  \n",
       "15               352.0            32.0                      15.0  \n",
       "79               352.0            32.0                      15.0  \n",
       "123              352.0            32.0                      15.0  \n",
       "254              352.0            32.0                      15.0  \n",
       "\n",
       "[5 rows x 32 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df[\"Id_question\"] ==1].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 데이터 품질과 분포\n",
    "\n",
    "이제 질문과 답변을 연결했고 데이터 품질을 검증했으므로 요약 통계를 만들어 보죠.\n",
    "\n",
    "- 데이터셋에 얼마나 많은 질문이 있나요?\n",
    "- 얼마나 많은 질문이 답변을 받았나요?\n",
    "- 질문은 얼마나 긴가요?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "답변이 없는 질문: 7971개, 1개 이상의 답변을 가진 질문: 7827개, 답변이 채택된 질문: 4124개\n"
     ]
    }
   ],
   "source": [
    "has_accepted_answer = df[df[\"is_question\"] & ~(df[\"AcceptedAnswerId\"].isna())]\n",
    "received_answers = df[df[\"is_question\"] & (df[\"AnswerCount\"]!=0)]\n",
    "no_answers = df[df[\"is_question\"] & (df[\"AcceptedAnswerId\"].isna()) & (df[\"AnswerCount\"]==0)]\n",
    "\n",
    "print(\"총 질문: %s개, 1개 이상의 답변을 가진 질문: %s개, 답변이 채택된 질문: %s개\" % (\n",
    "    len(df[df[\"is_question\"]]),\n",
    "    len(received_answers),\n",
    "    len(has_accepted_answer)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "대부분의 질문은 답변을 받았습니다. 이 답변 중 절반만 채택되었습니다. 질문에 채택된 답변이 있는지 여부는 좋은 레이블이 될 수 있습니다.\n",
    "\n",
    "또 다른 가능성있는 레이블은 `Score` 열에 있는 질문이 받은 업보트 수입니다. 모델을 얼마나 상세화할지에 따라 이 레이블을 이진 특성(예를 들어, 이 점수가 중앙값보다 큰지 작은지) 또는 연속적으로 다룰 수 있기 때문에 유용합니다. 이 점수의 분포를 다음에 그려 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(16,10))\n",
    "fig.suptitle(\"Distribution of question scores\")\n",
    "plt.xlabel(\"Question scores\")\n",
    "plt.ylabel(\"Number of questions\")\n",
    "df[\"Score\"].hist(bins=200);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "질문 길이 같은 예측에 좋은 특성을 살펴 보는 식으로 질문에 조금 더 자세히 분석해 보겠습니다. 데이터셋에 있는 질문의 길이는 얼마나 될까요?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(16,10))\n",
    "fig.suptitle(\"Distribution of question length for sentences\")\n",
    "plt.xlabel(\"Words per question\")\n",
    "plt.ylabel(\"Number of questions\")\n",
    "q_len = plt.hist(df[\"text_len\"],  bins=2000,log=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "몇 개의 이상치만이 이 천개 이상의 단어를 포함하고 있습니다. 그래프에서 짧은 질문을 크게 확대해 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(16,10))\n",
    "fig.suptitle(\n",
    "    \"Distribution of question length for sentences shorter than 2000 words\")\n",
    "plt.xlabel(\"Words per question\")\n",
    "plt.ylabel(\"Number of questions\")\n",
    "q_len_trunc = plt.hist(df[df[\"text_len\"]<2000][\"text_len\"], bins=200, log=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "질문의 길이로 많은 업보트를 얻는 것을 예측할 수 있을까요?\n",
    "\n",
    "이 질문에 대답하기 위해 높은 점수의 질문과 낮은 점수의 질문에 대한 두 분포를 겹쳐서 그려 볼 수 있습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "high_score = df[\"Score\"] > df[\"Score\"].median()\n",
    "# 아주 긴 질문을 제외힙니다\n",
    "normal_length = df[\"text_len\"] < 2000\n",
    "\n",
    "ax = df[df[\"is_question\"] & high_score & normal_length][\"text_len\"].hist(\n",
    "    bins=60,\n",
    "    density=True,\n",
    "    histtype=\"step\",\n",
    "    color=\"orange\",\n",
    "    linewidth=3,\n",
    "    grid=False,\n",
    "    figsize=(16, 10),\n",
    ")\n",
    "\n",
    "df[df[\"is_question\"] & ~high_score & normal_length][\"text_len\"].hist(\n",
    "    bins=60,\n",
    "    density=True,\n",
    "    histtype=\"step\",\n",
    "    color=\"purple\",\n",
    "    linewidth=3,\n",
    "    grid=False,\n",
    ")\n",
    "\n",
    "handles = [\n",
    "    Rectangle((0, 0), 1, 1, color=c, ec=\"k\") for c in [\"orange\", \"purple\"]\n",
    "]\n",
    "labels = [\"High score\", \"Low score\"]\n",
    "plt.legend(handles, labels)\n",
    "ax.set_xlabel(\"Sentence length (characters)\")\n",
    "ax.set_ylabel(\"Percentage of sentences\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "높은 점수의 질문은 평균적으로 길이가 더 깁니다. 즉 질문의 길이는 모델에 적용할 수 있는 후보 특성입니다."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "품질 측정 방법으로 질문이 받은 답변 수의 분포를 자세히 살펴 보겠습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(16,10))\n",
    "fig.suptitle(\"Distribution of number of answers\")\n",
    "plt.xlabel(\"Number of answers\")\n",
    "plt.ylabel(\"Number of questions\")\n",
    "hist = df[df[\"is_question\"]][\"AnswerCount\"].hist(bins=40)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "높은 점수의 질문이 더 많은 답변을 받았는지 확인해 보죠.\n",
    "\n",
    "먼저 한 변수에 대한 함수로 다른 한 변수를 그리면 트렌드를 파악할 수 있습니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "scatter = df[df[\"is_question\"]][[\"Score\", \"AnswerCount\"]].plot(x=\"Score\", y=\"AnswerCount\", \n",
    "                                                               kind=\"scatter\",\n",
    "                                                              figsize=(16, 10))\n",
    "ax.set_xlabel(\"Score\")\n",
    "ax.set_ylabel(\"Num answers\")\n",
    "scatter.set_title(\"Answer counts as a function of question score\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "점수와 답변의 개수는 상관관계가 있는 것 같군요. 점수를 `is_score_above_median`를 사용해 이진 레이블로 만드는 것에 대해 언급했습니다. 이 레이블이 답변의 개수와 상관관계가 있는지 알아 보겠습니다. 이를 위해 카테고리(중앙값보다 큰 점수와 작은 점수)마다 질문이 받은 답변의 개수를 히스토그램으로 그립니다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1152x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(16,10))\n",
    "fig.suptitle(\"Distribution of number of answers for high and low score questions\")\n",
    "plt.xlim(-5,80)\n",
    "\n",
    "ax = df[df[\"is_question\"] &\n",
    "        (df[\"Score\"] > df[\"Score\"].median())][\"AnswerCount\"].hist(bins=60,\n",
    "                                                          density=True,\n",
    "                                                          histtype=\"step\",\n",
    "                                                          color=\"orange\",\n",
    "                                                          linewidth=3,\n",
    "                                                          grid=False,\n",
    "                                                          figsize=(16, 10))\n",
    "\n",
    "df[df[\"is_question\"] &\n",
    "   ~(df[\"Score\"] > df[\"Score\"].median())][\"AnswerCount\"].hist(bins=60,\n",
    "                                                     density=True,\n",
    "                                                     histtype=\"step\",\n",
    "                                                     color=\"purple\",\n",
    "                                                     linewidth=3,\n",
    "                                                     grid=False)\n",
    "\n",
    "handles = [Rectangle((0, 0), 1, 1, color=c, ec=\"k\") for c in\n",
    "           [\"orange\", \"purple\"]]\n",
    "labels = [\"High score\", \"Low score\"]\n",
    "plt.legend(handles, labels)\n",
    "ax.set_xlabel(\"Num answers\")\n",
    "ax.set_ylabel(\"Percentage of sentences\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "높은 점수를 얻은 질문이 더 많은 답변을 받은 것 같습니다. 질문의 점수는 상세 수준으로 다룰 수 있기 때문에 좋은 레이블로 보입니다. 하지만 답변의 개수도 좋은 선택입니다.\n",
    "\n",
    "답변의 개수가 좋은 레이블이지만 특성으로 사용할 수는 없습니다. 그래프에서 보듯이 점수와 상관관계가 있음에도 사용자가 질문을 입력하는 추론시에 이를 얻을 수 없기 때문입니다.\n",
    "\n",
    "모델이 이 정보 없이 동작해야 하므로 훈련하는 동안 이를 완전히 무시하거나 훈련에 사용하고 제품 환경에서는 특성 값을 상수로 설정할 수 있습니다. 첫 번째 방법이 간단합니다. 하지만 두 번째 방법은 감지하려는 다른 스타일의 영향에서 답변 개수 특성의 영향을 분리하는데 도움이 됩니다. 두 방법에 대해 실험을 독자에게 남겨 놓겠습니다.\n",
    "\n",
    "데이터 탐색을 마친 후 [데이터 분할](https://github.com/hundredblocks/ml-powered-applications/blob/master/notebooks/splitting_data.ipynb) 방법을 알아 보겠습니다."
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}