notebooks/dataset_exploration.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analyzing an initial dataset\n",
    "\n",
    "In this notebook, we will quickly explore a real dataset of questions from writers.stackexchange.com. The dataset was initially sourced from the [archive](https://archive.org/details/stackexchange).\n",
    "\n",
    "First, we will load the data. If you are loading a different csv, make sure you have pre-processed the raw xml using the `ml_editor` python package."
   ]
  },
  {
   "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": [
    "## Data format\n",
    "\n",
    "Let's start by thinking through how we would like to format the data. Amongst other decisions, we will need to decide which label we should give our model.\n",
    "\n",
    "We want a model that measures the quality of a question. To that end, we could use:\n",
    "- The number of upvotes a question gets\n",
    "- The number of answers a question gets, or whether they get an answer at all\n",
    "- Whether an answer was marked as accepted or not\n",
    "\n",
    "First, let's format our dataset to reconcile questions and associated answers, and verify that they match up.\n",
    "\n",
    "We will start by filling missing values, as well as adding two features (`text_len` and `is_question`) we will use later."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Start by changing types to make processing easier\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",
    "# Add measure of the length of a post\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",
    "# A question is a post of id 1\n",
    "df[\"is_question\"] = df[\"PostTypeId\"] == 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data quality\n",
    "\n",
    "Let's examine the quality of the data in this dataset, starting by answering the questions below\n",
    "\n",
    "- How much of the data is missing?\n",
    "- What is the quality of the text?\n",
    "- Do the answers match the questions?"
   ]
  },
  {
   "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": [
    "# Start by displaying columns and counts of non-null entries\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We have a little over 34.000 posts which consist of both questions and answers. \n",
    "\n",
    "Looking at the `Body` column, it appears that it is null in `34330 - 34256 = 74` rows. Let's take a look at these rows to see if we should remove them."
   ]
  },
  {
   "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": [
    "All of the null bodys are of  PostTypeId 4 or 5. \n",
    "\n",
    "The readme file that accompanied the archive only mentions PostTypeIds of 1 (questions) and 2 (answers). We will remove all rows not marked `PostTypeId` 1 or 2, since we are only interested in questions and answers."
   ]
  },
  {
   "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": [
    "Now let's look at a few questions and answers and verify that they match, and that the text is readable."
   ]
  },
  {
   "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",
    "# Setting this option allows us to display all the data\n",
    "pd.options.display.max_colwidth = 500\n",
    "q_and_a[[\"body_text\", \"body_text_answer\"]][:3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So far so good, the **questions and answers seem to match**, and the text seems coherent and human readable  except for unescaped newlines characters (\"\\n\").\n",
    "\n",
    "Let's add to each row representing an answer information about the question it was answering."
   ]
  },
  {
   "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": [
    "This allows us to see all answers for a given question like so"
   ]
  },
  {
   "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": [
    "## Data quantity and distribution\n",
    "\n",
    "Now that we have linked questions and answers, and verified data quality, let's produce some summary statistics.\n",
    "\n",
    "- How many questions are in the dataset?\n",
    "- How many questions received answers?\n",
    "- How long do questions tend to be?"
   ]
  },
  {
   "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(\"답변이 없는 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": [
    "Most questions have an answer, and only half of those answers have been accepted. Whether a question received an accepted answer could be a good label.\n",
    "\n",
    "Another potential label would be the number of upvotes a question received, denoted in the `Score` column. This label is helpful because we can either treat it as a binary (whether the score is higher or lower than the median) or continuous depending on how granular we would like our model to be.I've plotted the score distribution below."
   ]
  },
  {
   "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": [
    "Let's dive more into the questions by looking at potential predictive features, such as question length. How long are the questions in our dataset?"
   ]
  },
  {
   "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": [
    "Only a few outliers seem to contain more than two thousand words. Let's zoom in to the shorter questions on the graph."
   ]
  },
  {
   "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": [
    "Now is sentence length predictive of getting many upvotes? \n",
    "\n",
    "In order to start answering this question, we can update this plot to overlay the distributions for both high and low score questions."
   ]
  },
  {
   "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",
    "# We filter out really long questions\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": [
    "It seems that there may be a relatively higher proportion of very short questions amongst unanswered ones. Questions with high scores tend to be longer on average. This makes question length a feature candidate that we should try for our model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Going back to using \"received answers\" as a measure of quality, let's look at the number of answers questions get, and more precisely the distribution of the quantity of answers."
   ]
  },
  {
   "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": [
    "Let's see if questions with high scores tend to get more answers. \n",
    "\n",
    "First, let's plot one as a function of the other, and see whether we can establish a trend."
   ]
  },
  {
   "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": [
    "Scores and number of answers seem correlated. We mentioned above making the score into a binary label by using `is_score_above_median` as the label. Let's see how this label would correlate with the number of answers by plotting histograms of the numbers of answers a question gets for question in each category (score above and below the median)"
   ]
  },
  {
   "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": [
    "It seems like questions that get a high score get more answers. The score of a question thus seems like a good initial label since it is more granular, but the number of answers would be a fine choice as well. \n",
    "\n",
    "While it would make a good label, we cannot use the number of answers as a feature. This is because despite being correlated with the score, as the plot above shows, we will not have access to it at inference time, when we receive input from the users.\n",
    "\n",
    "Since our model needs to work without having access to this input, we can either ignore it completely during training, or use it as a feature during training, and set the value of this feature as a constant in production. The first approach is simpler, but the second approach can help a model separate the effect of the \"number of answers\" feature from other style effects which we do want to capture. I'll let the reader experiment with both approaches.\n",
    "\n",
    "After this summary exploration, we will see how to [split data](https://github.com/hundredblocks/ml-powered-applications/blob/master/notebooks/splitting_data.ipynb)."
   ]
  }
 ],
 "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
}