Homework - Pandas - Stud results.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "ae6dfc9e",
   "metadata": {},
   "source": [
    "# 0) Подготовка, загрузка данных, первичный осмотр и оценка размера файлов (как по числу строк, так и в Мб)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c9cae31f",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%reset out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "1e5901f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "ea9cd962",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Том в устройстве C имеет метку System\n",
      " Серийный номер тома: 80AD-5924\n",
      "\n",
      " Содержимое папки C:\\Users\\levakin_av\\SkillFactory\\Homework - Pandas - Stud results\\data\n",
      "\n",
      "11.10.2022  18:34    <DIR>          .\n",
      "11.10.2022  18:34    <DIR>          ..\n",
      "03.10.2022  07:58    <DIR>          .ipynb_checkpoints\n",
      "02.10.2022  22:40             9 703 lectures.csv\n",
      "02.10.2022  22:40           296 161 questions.csv\n",
      "03.10.2022  08:09     5 846 760 913 train.csv\n",
      "11.10.2022  04:21     6 656 128 907 train_2.csv\n",
      "11.10.2022  18:34    <DIR>          Математика машинного обучения\n",
      "               4 файлов 12 503 195 684 байт\n",
      "               4 папок  64 472 494 080 байт свободно\n"
     ]
    }
   ],
   "source": [
    "ls data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "17bf6aa7",
   "metadata": {},
   "outputs": [],
   "source": [
    "lect_df = pd.read_csv('data\\lectures.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "24932486",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(418, 4)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "96fc401a",
   "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>lecture_id</th>\n",
       "      <th>tag</th>\n",
       "      <th>part</th>\n",
       "      <th>type_of</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>89</td>\n",
       "      <td>159</td>\n",
       "      <td>5</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>185</td>\n",
       "      <td>45</td>\n",
       "      <td>6</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>192</td>\n",
       "      <td>79</td>\n",
       "      <td>5</td>\n",
       "      <td>solving question</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>317</td>\n",
       "      <td>156</td>\n",
       "      <td>5</td>\n",
       "      <td>solving question</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   lecture_id  tag  part           type_of\n",
       "0          89  159     5           concept\n",
       "1         100   70     1           concept\n",
       "2         185   45     6           concept\n",
       "3         192   79     5  solving question\n",
       "4         317  156     5  solving question"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "b00629bf",
   "metadata": {},
   "outputs": [],
   "source": [
    "quest_df = pd.read_csv('data\\questions.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "ee46ea8a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13523, 5)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "quest_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "5a0d776f",
   "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>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "      <th>tags</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>51 131 162 38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>131 36 81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>131 101 162 92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>131 149 162 29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>131 5 162 38</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   question_id  bundle_id  correct_answer  part            tags\n",
       "0            0          0               0     1   51 131 162 38\n",
       "1            1          1               1     1       131 36 81\n",
       "2            2          2               0     1  131 101 162 92\n",
       "3            3          3               0     1  131 149 162 29\n",
       "4            4          4               3     1    131 5 162 38"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "quest_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "38bab65c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Том в устройстве C имеет метку System\n",
      " Серийный номер тома: 80AD-5924\n",
      "\n",
      " Содержимое папки C:\\Users\\levakin_av\\SkillFactory\\Homework - Pandas - Stud results\\data\n",
      "\n",
      "11.10.2022  18:34    <DIR>          .\n",
      "11.10.2022  18:34    <DIR>          ..\n",
      "03.10.2022  07:58    <DIR>          .ipynb_checkpoints\n",
      "02.10.2022  22:40             9 703 lectures.csv\n",
      "02.10.2022  22:40           296 161 questions.csv\n",
      "03.10.2022  08:09     5 846 760 913 train.csv\n",
      "11.10.2022  04:21     6 656 128 907 train_2.csv\n",
      "11.10.2022  18:34    <DIR>          Математика машинного обучения\n",
      "               4 файлов 12 503 195 684 байт\n",
      "               4 папок  64 472 481 792 байт свободно\n"
     ]
    }
   ],
   "source": [
    "ls data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "135b0562",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "101230333\n"
     ]
    }
   ],
   "source": [
    "big_file = open(r'data\\train.csv', 'rb') # r перед строкой пути - обязательно. Без него не работает.\n",
    "line_count = 0\n",
    "for line in big_file:\n",
    "    line_count += 1\n",
    "print(line_count)\n",
    "# 101 230 333"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "3e551db9",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = pd.read_csv(r\"data\\train.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "51a0b28e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(101230332, 10)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.shape\n",
    "# (101230332, 10) - ура, загрузился весь обхем данных\n",
    "# все данные загрузились в фрейм, но глядя на монитор потребления ресурсов делаю вывод, что фрейм хранится на жетском диске, а не в опер. памяти.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "ca42f76d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101230332 entries, 0 to 101230331\n",
      "Data columns (total 10 columns):\n",
      " #   Column                          Dtype  \n",
      "---  ------                          -----  \n",
      " 0   row_id                          int64  \n",
      " 1   timestamp                       int64  \n",
      " 2   user_id                         int64  \n",
      " 3   content_id                      int64  \n",
      " 4   content_type_id                 int64  \n",
      " 5   task_container_id               int64  \n",
      " 6   user_answer                     int64  \n",
      " 7   answered_correctly              int64  \n",
      " 8   prior_question_elapsed_time     float64\n",
      " 9   prior_question_had_explanation  object \n",
      "dtypes: float64(1), int64(8), object(1)\n",
      "memory usage: 7.5+ GB\n"
     ]
    }
   ],
   "source": [
    "#оценим размер фрейма\n",
    "train_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "bd28fad6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# интересно, что фрейм занимает более 7.5 Гб, хотя сам файл занимает 5.4 Гб\n",
    "# Так же заметим, что числовые типы кодированы int64, хотя из постановки задачи мы знаем, что они другие:\n",
    "    #row_id: (int64)\n",
    "    #timestamp: (int64)\n",
    "    #user_id: (int32)\n",
    "    #content_id: (int16)\n",
    "    #content_type_id: (int8)\n",
    "    #task_container_id: (int16)\n",
    "    #user_answer: (int8)\n",
    "    #answered_correctly: (int8)\n",
    "    #prior_question_elapsed_time: (float32)\n",
    "    #prior_question_had_explanation: (bool)\n",
    "# Это поможет нам сократить размер датафрейма. Но попробуем исследовать оптимальны ли значения в постановке или можно выбрать еще\n",
    "# более экономные типы данных для int-колонок"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3fb337b1",
   "metadata": {},
   "source": [
    "# 1) Сокращение объемов фрейма train.csv ( и не только) за счет изменения типов  данных"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f738feaa",
   "metadata": {},
   "source": [
    "## 1.1) Столбец row_id (ID code for the row)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "7b46dddf",
   "metadata": {},
   "outputs": [],
   "source": [
    "row_id_ser = pd.Series(train_df.row_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6021757c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 101230331 101230332\n"
     ]
    }
   ],
   "source": [
    "print(row_id_ser.min(), row_id_ser.max(), row_id_ser.nunique())\n",
    "# 0 101230331 101230332\n",
    "# максимальное значение помещается в диапазон int32\t (-2147483648 to 2147483647)\\\n",
    "# ТО ЕСТЬ В ПОСТАНОВКЕ НАС ОБМАНЫВАЮТ. \n",
    "# проверим, что изменение типа не приведет к проблемам"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e733fd1d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 101230331 101230332\n"
     ]
    }
   ],
   "source": [
    "row_id_ser = row_id_ser.astype('int32')\n",
    "print(row_id_ser.min(), row_id_ser.max(), row_id_ser.nunique())\n",
    "# 0 101230331 101230332\n",
    "# ничего не изменилось, но на всякий случай сравним поэлементно."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "7f69d81e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True    101230332\n",
       "Name: row_id, dtype: int64"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(row_id_ser == pd.Series(train_df.row_id)).value_counts()\n",
    "#True    101230332\n",
    "#Name: row_id, dtype: int64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2ea3cff8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101230332 entries, 0 to 101230331\n",
      "Data columns (total 10 columns):\n",
      " #   Column                          Dtype  \n",
      "---  ------                          -----  \n",
      " 0   row_id                          int32  \n",
      " 1   timestamp                       int64  \n",
      " 2   user_id                         int64  \n",
      " 3   content_id                      int64  \n",
      " 4   content_type_id                 int64  \n",
      " 5   task_container_id               int64  \n",
      " 6   user_answer                     int64  \n",
      " 7   answered_correctly              int64  \n",
      " 8   prior_question_elapsed_time     float64\n",
      " 9   prior_question_had_explanation  object \n",
      "dtypes: float64(1), int32(1), int64(7), object(1)\n",
      "memory usage: 7.2+ GB\n"
     ]
    }
   ],
   "source": [
    "# итак, моножно заменить тип на Int32\n",
    "train_df.row_id = train_df.row_id.astype('int32')\n",
    "train_df.info()\n",
    "# row_id                          int32\n",
    "# usage: 7.2+ GB\n",
    "# Заметим так же что мы могли бы использовать uint32, так как у нас строго положительные значения\n",
    "# Но uint32 занимает столько же места как и int32"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "838a63f5",
   "metadata": {},
   "source": [
    "## 1.2) Числовые столбцы кроме row_id "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "a6d084e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "## Займемся остальными столбцами с числовыми значениями. Будем исследовать их по той же схеме, но все сразу.\n",
    "# для этого нам пригодится инфомрация:\n",
    "#int8\tByte (-128 to 127)\n",
    "#int16\tInteger (-32768 to 32767)\n",
    "#int32\tInteger (-2147483648 to 2147483647)\n",
    "#int64\tInteger (-9223372036854775808 to 9223372036854775807)\n",
    "#uint8\tUnsigned integer (0 to 255)\n",
    "#uint16\tUnsigned integer (0 to 65535)\n",
    "#uint32\tUnsigned integer (0 to 4294967295)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "f4f1cf7d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['row_id', 'timestamp', 'user_id', 'content_id', 'content_type_id',\n",
       "       'task_container_id', 'user_answer', 'answered_correctly',\n",
       "       'prior_question_elapsed_time', 'prior_question_had_explanation'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "23190edb",
   "metadata": {},
   "outputs": [],
   "source": [
    "# сохраним результаты descibe в служебный фрейм, чтобы в последствии контролировать наши преобразования типов, сравниваия результаты\n",
    "numcols = ['timestamp', 'user_id', 'content_id', 'content_type_id',\n",
    "       'task_container_id', 'user_answer', 'answered_correctly',\n",
    "       'prior_question_elapsed_time']\n",
    "control_desc = train_df[numcols].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "54877f2b",
   "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>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>prior_question_elapsed_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>1.012303e+08</td>\n",
       "      <td>9.887879e+07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>7.703644e+09</td>\n",
       "      <td>1.076732e+09</td>\n",
       "      <td>5.219605e+03</td>\n",
       "      <td>1.935222e-02</td>\n",
       "      <td>9.040624e+02</td>\n",
       "      <td>1.376123e+00</td>\n",
       "      <td>6.251644e-01</td>\n",
       "      <td>2.542381e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.159266e+10</td>\n",
       "      <td>6.197163e+08</td>\n",
       "      <td>3.866359e+03</td>\n",
       "      <td>1.377596e-01</td>\n",
       "      <td>1.358302e+03</td>\n",
       "      <td>1.192896e+00</td>\n",
       "      <td>5.225307e-01</td>\n",
       "      <td>1.994815e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.150000e+02</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>-1.000000e+00</td>\n",
       "      <td>-1.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>5.243436e+08</td>\n",
       "      <td>5.408116e+08</td>\n",
       "      <td>2.063000e+03</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.040000e+02</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.600000e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>2.674234e+09</td>\n",
       "      <td>1.071781e+09</td>\n",
       "      <td>5.026000e+03</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>3.820000e+02</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>2.100000e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>9.924551e+09</td>\n",
       "      <td>1.615742e+09</td>\n",
       "      <td>7.425000e+03</td>\n",
       "      <td>0.000000e+00</td>\n",
       "      <td>1.094000e+03</td>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>2.966600e+04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>8.742577e+10</td>\n",
       "      <td>2.147483e+09</td>\n",
       "      <td>3.273600e+04</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>9.999000e+03</td>\n",
       "      <td>3.000000e+00</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>3.000000e+05</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          timestamp       user_id    content_id  content_type_id  \\\n",
       "count  1.012303e+08  1.012303e+08  1.012303e+08     1.012303e+08   \n",
       "mean   7.703644e+09  1.076732e+09  5.219605e+03     1.935222e-02   \n",
       "std    1.159266e+10  6.197163e+08  3.866359e+03     1.377596e-01   \n",
       "min    0.000000e+00  1.150000e+02  0.000000e+00     0.000000e+00   \n",
       "25%    5.243436e+08  5.408116e+08  2.063000e+03     0.000000e+00   \n",
       "50%    2.674234e+09  1.071781e+09  5.026000e+03     0.000000e+00   \n",
       "75%    9.924551e+09  1.615742e+09  7.425000e+03     0.000000e+00   \n",
       "max    8.742577e+10  2.147483e+09  3.273600e+04     1.000000e+00   \n",
       "\n",
       "       task_container_id   user_answer  answered_correctly  \\\n",
       "count       1.012303e+08  1.012303e+08        1.012303e+08   \n",
       "mean        9.040624e+02  1.376123e+00        6.251644e-01   \n",
       "std         1.358302e+03  1.192896e+00        5.225307e-01   \n",
       "min         0.000000e+00 -1.000000e+00       -1.000000e+00   \n",
       "25%         1.040000e+02  0.000000e+00        0.000000e+00   \n",
       "50%         3.820000e+02  1.000000e+00        1.000000e+00   \n",
       "75%         1.094000e+03  3.000000e+00        1.000000e+00   \n",
       "max         9.999000e+03  3.000000e+00        1.000000e+00   \n",
       "\n",
       "       prior_question_elapsed_time  \n",
       "count                 9.887879e+07  \n",
       "mean                  2.542381e+04  \n",
       "std                   1.994815e+04  \n",
       "min                   0.000000e+00  \n",
       "25%                   1.600000e+04  \n",
       "50%                   2.100000e+04  \n",
       "75%                   2.966600e+04  \n",
       "max                   3.000000e+05  "
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "control_desc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "38197e2e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101230332 entries, 0 to 101230331\n",
      "Data columns (total 10 columns):\n",
      " #   Column                          Dtype  \n",
      "---  ------                          -----  \n",
      " 0   row_id                          int32  \n",
      " 1   timestamp                       int64  \n",
      " 2   user_id                         int64  \n",
      " 3   content_id                      int16  \n",
      " 4   content_type_id                 int8   \n",
      " 5   task_container_id               int16  \n",
      " 6   user_answer                     int8   \n",
      " 7   answered_correctly              int8   \n",
      " 8   prior_question_elapsed_time     float64\n",
      " 9   prior_question_had_explanation  object \n",
      "dtypes: float64(1), int16(2), int32(1), int64(2), int8(3), object(1)\n",
      "memory usage: 4.1+ GB\n"
     ]
    }
   ],
   "source": [
    "# видим, что сразу можно сделать такие преобразованиея:\n",
    "# user_answer -> int8\n",
    "# answered_correctly -> int8\n",
    "# task_container_id -> int16\n",
    "# content_type_id -> int8\n",
    "# content_id -> uint16\n",
    "# сделаем эти преобразования:\n",
    "train_df.user_answer = train_df.user_answer.astype('int8')\n",
    "train_df.answered_correctly = train_df.answered_correctly.astype('int8')\n",
    "train_df.task_container_id = train_df.task_container_id.astype('int16')\n",
    "train_df.content_type_id = train_df.content_type_id.astype('int8')\n",
    "train_df.content_id = train_df.content_id.astype('int16')\n",
    "train_df.info()\n",
    "# memory usage: 4.1+ GB - отличный результат!\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "70f60afe",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "False\n"
     ]
    }
   ],
   "source": [
    "# сравниваем describe до и после изменения типов\n",
    "curr_desc = train_df[numcols].describe()\n",
    "print(control_desc.equals(curr_desc))\n",
    "# False\n",
    "# почему-то фреймы не совпали, читаем хелп по equals: Corresponding columns must be of the same dtype.\n",
    "# получается нам такое сравнение не подходит, так как мы хотим сравнить значения, но не типы (они гарантированно рзаличаются)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "b64e236f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n"
     ]
    }
   ],
   "source": [
    "# сравним фреймы иначе:\n",
    "print (curr_desc.loc[curr_desc.index != 'std'].to_dict() == control_desc.loc[control_desc.index != 'std'].to_dict())\n",
    "# True\n",
    "# почему не совпадает  std - непонятно, но различия минимальны, можно убидетьися вот так:\n",
    "# for c in numcols:     print (c, curr_desc[c] - control_desc[c])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "c980c355",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "115 2147482888 393656\n"
     ]
    }
   ],
   "source": [
    "# отдельно разберемся с user_id                         \n",
    "print(train_df.user_id.min(), train_df.user_id.max(), train_df.user_id.nunique())\n",
    "# 115      2 147 482 888     393656    \n",
    "# максимальное значение int32 = 2 147 483 647\n",
    "# вывод - user_id помещается в int32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "78f91a27",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df.user_id = train_df.user_id.astype('int32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "ec6f1025",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "bool"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# отдельно разберемся с prior_question_had_explanation\n",
    "# поймем, что за тип там хранится:\n",
    "type(train_df.prior_question_had_explanation.iloc[1024342]) # случайная строка\n",
    "#bool\n",
    "# не меняем это поле\n",
    "# так же не будем менять типы у timestamp и prior_question_elapsed_time     "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "f7ad8b81",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 101230332 entries, 0 to 101230331\n",
      "Data columns (total 10 columns):\n",
      " #   Column                          Dtype  \n",
      "---  ------                          -----  \n",
      " 0   row_id                          int32  \n",
      " 1   timestamp                       int64  \n",
      " 2   user_id                         int32  \n",
      " 3   content_id                      int16  \n",
      " 4   content_type_id                 int8   \n",
      " 5   task_container_id               int16  \n",
      " 6   user_answer                     int8   \n",
      " 7   answered_correctly              int8   \n",
      " 8   prior_question_elapsed_time     float64\n",
      " 9   prior_question_had_explanation  object \n",
      "dtypes: float64(1), int16(2), int32(2), int64(1), int8(3), object(1)\n",
      "memory usage: 3.7+ GB\n"
     ]
    }
   ],
   "source": [
    "# финальный размер фрейма:\n",
    "train_df.info()\n",
    "# memory usage: 3.7+ GB - в два раза меньше, чем раньше.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fdf0aefe",
   "metadata": {},
   "source": [
    "## 1.3) Сокращение объема lectures"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "35f83d94",
   "metadata": {},
   "outputs": [],
   "source": [
    "lect_df = pd.read_csv('data\\lectures.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "4f52e489",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 418 entries, 0 to 417\n",
      "Data columns (total 4 columns):\n",
      " #   Column      Non-Null Count  Dtype \n",
      "---  ------      --------------  ----- \n",
      " 0   lecture_id  418 non-null    int64 \n",
      " 1   tag         418 non-null    int64 \n",
      " 2   part        418 non-null    int64 \n",
      " 3   type_of     418 non-null    object\n",
      "dtypes: int64(3), object(1)\n",
      "memory usage: 13.2+ KB\n"
     ]
    }
   ],
   "source": [
    "lect_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "5c2ab2dd",
   "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>lecture_id</th>\n",
       "      <th>tag</th>\n",
       "      <th>part</th>\n",
       "      <th>type_of</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>89</td>\n",
       "      <td>159</td>\n",
       "      <td>5</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>185</td>\n",
       "      <td>45</td>\n",
       "      <td>6</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   lecture_id  tag  part  type_of\n",
       "0          89  159     5  concept\n",
       "1         100   70     1  concept\n",
       "2         185   45     6  concept"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "5a4083c5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 7\n",
      "89 32736\n",
      "0 187\n",
      "['concept' 'solving question' 'intention' 'starter']\n"
     ]
    }
   ],
   "source": [
    "print(lect_df.part.min(), lect_df.part.max())\n",
    "print(lect_df.lecture_id.min(), lect_df.lecture_id.max())\n",
    "print(lect_df.tag.min(), lect_df.tag.max())\n",
    "print((lect_df.type_of.unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "29e644fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "lect_df.part = lect_df.part.astype('int8')\n",
    "lect_df.tag = lect_df.tag.astype('uint8')\n",
    "lect_df.lecture_id = lect_df.lecture_id.astype('int16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "b5531fa7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 418 entries, 0 to 417\n",
      "Data columns (total 4 columns):\n",
      " #   Column      Non-Null Count  Dtype \n",
      "---  ------      --------------  ----- \n",
      " 0   lecture_id  418 non-null    int16 \n",
      " 1   tag         418 non-null    uint8 \n",
      " 2   part        418 non-null    int8  \n",
      " 3   type_of     418 non-null    object\n",
      "dtypes: int16(1), int8(1), object(1), uint8(1)\n",
      "memory usage: 5.0+ KB\n"
     ]
    }
   ],
   "source": [
    "lect_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "a512d8be",
   "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>lecture_id</th>\n",
       "      <th>tag</th>\n",
       "      <th>part</th>\n",
       "      <th>type_of</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>89</td>\n",
       "      <td>159</td>\n",
       "      <td>5</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>100</td>\n",
       "      <td>70</td>\n",
       "      <td>1</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>185</td>\n",
       "      <td>45</td>\n",
       "      <td>6</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   lecture_id  tag  part  type_of\n",
       "0          89  159     5  concept\n",
       "1         100   70     1  concept\n",
       "2         185   45     6  concept"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.head(3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "6f886a73",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 418 entries, 0 to 417\n",
      "Data columns (total 4 columns):\n",
      " #   Column      Non-Null Count  Dtype   \n",
      "---  ------      --------------  -----   \n",
      " 0   lecture_id  418 non-null    int16   \n",
      " 1   tag         418 non-null    uint8   \n",
      " 2   part        418 non-null    int8    \n",
      " 3   type_of     418 non-null    category\n",
      "dtypes: category(1), int16(1), int8(1), uint8(1)\n",
      "memory usage: 2.4 KB\n"
     ]
    }
   ],
   "source": [
    "lect_df.type_of = lect_df.type_of.astype('category')\n",
    "lect_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "b5fb162a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0      0\n",
       "1      0\n",
       "2      0\n",
       "3      2\n",
       "4      2\n",
       "      ..\n",
       "413    2\n",
       "414    2\n",
       "415    0\n",
       "416    0\n",
       "417    0\n",
       "Length: 418, dtype: int8"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.type_of.cat.codes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "476f99e2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['concept', 'intention', 'solving question', 'starter'], dtype='object')"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.type_of.cat.categories"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c0db844",
   "metadata": {},
   "source": [
    "## 1.3) Сокращение объема questions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "50f03011",
   "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>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "      <th>tags</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>51 131 162 38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>131 36 81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>131 101 162 92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>131 149 162 29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>131 5 162 38</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   question_id  bundle_id  correct_answer  part            tags\n",
       "0            0          0               0     1   51 131 162 38\n",
       "1            1          1               1     1       131 36 81\n",
       "2            2          2               0     1  131 101 162 92\n",
       "3            3          3               0     1  131 149 162 29\n",
       "4            4          4               3     1    131 5 162 38"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(quest_df.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "8a10f5f5",
   "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>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>13523.00000</td>\n",
       "      <td>13523.000000</td>\n",
       "      <td>13523.000000</td>\n",
       "      <td>13523.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>6761.00000</td>\n",
       "      <td>6760.510907</td>\n",
       "      <td>1.455298</td>\n",
       "      <td>4.264956</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>3903.89818</td>\n",
       "      <td>3903.857783</td>\n",
       "      <td>1.149707</td>\n",
       "      <td>1.652553</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>3380.50000</td>\n",
       "      <td>3379.500000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>6761.00000</td>\n",
       "      <td>6761.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>10141.50000</td>\n",
       "      <td>10140.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>13522.00000</td>\n",
       "      <td>13522.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>7.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       question_id     bundle_id  correct_answer          part\n",
       "count  13523.00000  13523.000000    13523.000000  13523.000000\n",
       "mean    6761.00000   6760.510907        1.455298      4.264956\n",
       "std     3903.89818   3903.857783        1.149707      1.652553\n",
       "min        0.00000      0.000000        0.000000      1.000000\n",
       "25%     3380.50000   3379.500000        0.000000      3.000000\n",
       "50%     6761.00000   6761.000000        1.000000      5.000000\n",
       "75%    10141.50000  10140.000000        3.000000      5.000000\n",
       "max    13522.00000  13522.000000        3.000000      7.000000"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(quest_df.describe())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "d510be37",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 13523 entries, 0 to 13522\n",
      "Data columns (total 5 columns):\n",
      " #   Column          Non-Null Count  Dtype \n",
      "---  ------          --------------  ----- \n",
      " 0   question_id     13523 non-null  int64 \n",
      " 1   bundle_id       13523 non-null  int64 \n",
      " 2   correct_answer  13523 non-null  int64 \n",
      " 3   part            13523 non-null  int64 \n",
      " 4   tags            13522 non-null  object\n",
      "dtypes: int64(4), object(1)\n",
      "memory usage: 528.4+ KB\n"
     ]
    }
   ],
   "source": [
    "quest_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "8125dc9c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 13523 entries, 0 to 13522\n",
      "Data columns (total 5 columns):\n",
      " #   Column          Non-Null Count  Dtype \n",
      "---  ------          --------------  ----- \n",
      " 0   question_id     13523 non-null  int16 \n",
      " 1   bundle_id       13523 non-null  int16 \n",
      " 2   correct_answer  13523 non-null  int8  \n",
      " 3   part            13523 non-null  int8  \n",
      " 4   tags            13522 non-null  object\n",
      "dtypes: int16(2), int8(2), object(1)\n",
      "memory usage: 185.0+ KB\n"
     ]
    }
   ],
   "source": [
    "quest_df.question_id = quest_df.question_id.astype('int16')\n",
    "quest_df.bundle_id = quest_df.bundle_id.astype('int16')\n",
    "quest_df.correct_answer = quest_df.correct_answer.astype('int8')\n",
    "quest_df.part = quest_df.part.astype('int8')\n",
    "quest_df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "a100b4c5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1519"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(quest_df.tags.value_counts())\n",
    "#1519\n",
    "# не будем переводить в категорию"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "d813aff1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "189\n"
     ]
    }
   ],
   "source": [
    "s = set()\n",
    "x = []\n",
    "for row in quest_df.itertuples():\n",
    "    x = x + str(row.tags).split(sep = ' ')\n",
    "print(len(set(x)))\n",
    "# всего 189 уникальных тэгов. Не будем разворачивать их в колонки, слишком много уникальных значений."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1f66eb73",
   "metadata": {},
   "source": [
    "# 2) Поиск и очистка данных от аномальных значений"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "1461e099",
   "metadata": {},
   "outputs": [],
   "source": [
    "# теперь когда датафрейм стал меньше, посмотрим на данные более внимательно"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "d8dd7d79",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "row_id                            101230332\n",
       "timestamp                         101230332\n",
       "user_id                           101230332\n",
       "content_id                        101230332\n",
       "content_type_id                   101230332\n",
       "task_container_id                 101230332\n",
       "user_answer                       101230332\n",
       "answered_correctly                101230332\n",
       "prior_question_elapsed_time        98878794\n",
       "prior_question_had_explanation    100837826\n",
       "dtype: int64"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.count()\n",
    "# видим, что в prior_question_elapsed_time есть nan, что соответствует постановке"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1ee1acc7",
   "metadata": {},
   "source": [
    "## 2.1) Исследование prior_question_elapsed_time (pqe_time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "5072d4f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "etime_stat_ser = train_df.prior_question_elapsed_time.value_counts().sort_values(ascending = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "59f17cd9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "17000.0     5154588\n",
       "16000.0     4752104\n",
       "18000.0     4670915\n",
       "19000.0     4017031\n",
       "15000.0     3640279\n",
       "             ...   \n",
       "259666.0          1\n",
       "298600.0          1\n",
       "299666.0          1\n",
       "297200.0          1\n",
       "256400.0          1\n",
       "Name: prior_question_elapsed_time, Length: 3258, dtype: int64"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "etime_stat_ser"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "cd507812",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "17000.0    5.213\n",
       "16000.0    4.806\n",
       "18000.0    4.724\n",
       "19000.0    4.063\n",
       "15000.0    3.682\n",
       "20000.0    3.638\n",
       "21000.0    3.364\n",
       "22000.0    3.076\n",
       "23000.0    2.708\n",
       "14000.0    2.664\n",
       "24000.0    2.351\n",
       "13000.0    2.085\n",
       "25000.0    2.045\n",
       "12000.0    1.845\n",
       "26000.0    1.793\n",
       "11000.0    1.680\n",
       "10000.0    1.595\n",
       "27000.0    1.587\n",
       "9000.0     1.525\n",
       "28000.0    1.410\n",
       "Name: prior_question_elapsed_time, dtype: float64"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# посмотрим, как распределены проценты по самы частым временам\n",
    "etime_stat_ser1 = etime_stat_ser.apply(lambda x: round(100*x/98878794, 3))\n",
    "etime_stat_ser1.head(20)\n",
    "# обратим внимание, что самые часты времена приходятся на целое число секунды (кратны 1000), это не соотвествует ожиданиям "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "a33cb792",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "d9508ef2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 792x792 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "etime_stat_ser1.sort_index().plot(figsize = (11,11))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "d25f79ac",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 288x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# видим небольшой всплеск на больших значениях абсциссы, исследуем их\n",
    "etime_stat_ser1.sort_index().tail(5).plot(figsize = (4,4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "22de61fa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "299600.0        3\n",
       "299666.0        1\n",
       "299750.0       13\n",
       "299800.0       13\n",
       "300000.0    71896\n",
       "Name: prior_question_elapsed_time, dtype: int64"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "etime_stat_ser.sort_index().tail(5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "272f7a4a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# видим, что 72 тыс. строк имеет значение 300 секунд. Видимо, это максимальное значение в системе.\n",
    "# удалять или как-то менять эти данные было бы неправильно, но будем помнить об этой особенности, \n",
    "# на случай если мы решим ввести категорию по prior_question_elapsed_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "c8f6d769",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 792x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# более детально исследуется тот факт, что на целое число секун приходятся пики:\n",
    "etime_stat_ser.sort_index().head(200).plot(figsize = (11,4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "574f509f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 792x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# обратим внимание, что распределения для целых и нецелых - разные. На графике ниже видно, что ни максимумы ни форма двух (?) мод не совпадают.\n",
    "#Это еще одна странность.\n",
    "etime_stat_ser.sort_index().head(1200).plot(figsize = (11,4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "8532bf36",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 792x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# покажем это более наглядно\n",
    "whole_sec_ind_list = list(range(0,301000,1000))\n",
    "#print(whole_sec_ind_list)\n",
    "z = etime_stat_ser[[i in whole_sec_ind_list for i in etime_stat_ser.index]]\n",
    "z.sort_index().head(200).plot(figsize = (11,4))\n",
    "w = etime_stat_ser[[i not in whole_sec_ind_list for i in etime_stat_ser.index]]\n",
    "w.sort_index().head(800).plot(figsize = (11,4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "e902578e",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 792x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "w = etime_stat_ser[[i not in whole_sec_ind_list for i in etime_stat_ser.index]]\n",
    "w.sort_index().head(400).plot(figsize = (11,4))\n",
    "# этот шум выглядит как несколько смешанных рядов\n",
    "# таких строк очень много, миллионы строк."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "id": "978452e3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# пока непонятно, что с ними делать. Нет повода утверждать, что это точно ошибки,\n",
    "# оставим эти строки в датафрейме, не будем их удалять."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "09a2ae8c",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>prior_question_had_explanation</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>37000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>55000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>19000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>11000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly  pqe_time prior_question_had_explanation  \n",
       "0            3                   1       NaN                            NaN  \n",
       "1            2                   1   37000.0                          False  \n",
       "2            0                   1   55000.0                          False  \n",
       "3            0                   1   19000.0                          False  \n",
       "4            1                   1   11000.0                          False  "
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# переименуем пару колонок\n",
    "train_df.rename(inplace=True, columns={'prior_question_elapsed_time': 'pqe_time'})\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "id": "ea39f4c6",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>pq_had_expl</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>37000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>55000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>19000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>11000.0</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly  pqe_time pq_had_expl  \n",
       "0            3                   1       NaN         NaN  \n",
       "1            2                   1   37000.0       False  \n",
       "2            0                   1   55000.0       False  \n",
       "3            0                   1   19000.0       False  \n",
       "4            1                   1   11000.0       False  "
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.rename(inplace=True, columns={'prior_question_had_explanation': 'pq_had_expl'})\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f04538e1",
   "metadata": {},
   "source": [
    "## 2.2) Избавимся от nan  в PQE_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "6b7144e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import math as m\n",
    "train_df['pqe_nan'] = train_df.pqe_time.apply(lambda x: m.isnan(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "7196edbf",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_nan</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>37000.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>55000.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>19000.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>11000.0</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly  pqe_time pq_had_expl  pqe_nan  \n",
       "0            3                   1       NaN         NaN     True  \n",
       "1            2                   1   37000.0       False    False  \n",
       "2            0                   1   55000.0       False    False  \n",
       "3            0                   1   19000.0       False    False  \n",
       "4            1                   1   11000.0       False    False  "
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "de588fa5",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df['pqe_time_2'] = train_df['pqe_time'].apply(lambda x: -1 if m.isnan(x) == True else int(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "id": "205fc742",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df.drop(labels = 'pqe_time', axis=1,inplace = True )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "1fd18638",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_nan</th>\n",
       "      <th>pqe_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>True</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>37000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>55000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>19000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>11000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly pq_had_expl  pqe_nan  pqe_time  \n",
       "0            3                   1         NaN     True        -1  \n",
       "1            2                   1       False    False     37000  \n",
       "2            0                   1       False    False     55000  \n",
       "3            0                   1       False    False     19000  \n",
       "4            1                   1       False    False     11000  "
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.rename(columns={'pqe_time_2':'pqe_time'}, inplace=True)\n",
    "train_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "id": "a5a7bd4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df.drop(labels = 'pqe_nan', axis=1,inplace = True )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "17fe61e2",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>37000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>55000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>19000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>11000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly pq_had_expl  pqe_time  \n",
       "0            3                   1         NaN        -1  \n",
       "1            2                   1       False     37000  \n",
       "2            0                   1       False     55000  \n",
       "3            0                   1       False     19000  \n",
       "4            1                   1       False     11000  "
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "662ae010",
   "metadata": {},
   "source": [
    "## 2.3) Попарная консистентность полей: train_df.content_id и quest_df.question_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "id": "d04c3d8f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "13523 13523 13523 13523 0 0 0\n"
     ]
    }
   ],
   "source": [
    "# content_id - это ключ для question_id в файле questions (для строк content_type_id = 0)\n",
    "# проверим, что в train нет значений content_id, которых нет в questions\n",
    "\n",
    "list1 = train_df[train_df.content_type_id == 0]['content_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = quest_df['question_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 13523 13523 13523 13523 0 0 0\n",
    "# все сходится: в train нет id вопросов, которых не было бы в соотвествующем справочнике вопросов"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "d31ad47c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Для того, чтобы заново не прогонять все исследования, запишем текущий вид train_df в файл\n",
    "train_df.to_csv('data/train_2.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "5698df7d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\n 0   row_id                          int32  \\n 1   timestamp                       int64  \\n 2   user_id                         int32  \\n 3   content_id                      int16  \\n 4   content_type_id                 int8   \\n 5   task_container_id               int16  \\n 6   user_answer                     int8   \\n 7   answered_correctly              int8   \\n 8   prior_question_elapsed_time     float64\\n 9   prior_question_had_explanation  object \\n '"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "'''\n",
    " 0   row_id                          int32  \n",
    " 1   timestamp                       int64  \n",
    " 2   user_id                         int32  \n",
    " 3   content_id                      int16  \n",
    " 4   content_type_id                 int8   \n",
    " 5   task_container_id               int16  \n",
    " 6   user_answer                     int8   \n",
    " 7   answered_correctly              int8   \n",
    " 8   prior_question_elapsed_time     float64\n",
    " 9   prior_question_had_explanation  object \n",
    " '''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "95256a44",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>37000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>55000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>19000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>11000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>157063</td>\n",
       "      <td>115</td>\n",
       "      <td>156</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>5000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>6</td>\n",
       "      <td>176092</td>\n",
       "      <td>115</td>\n",
       "      <td>51</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>17000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "      <td>194190</td>\n",
       "      <td>115</td>\n",
       "      <td>50</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>17000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>212463</td>\n",
       "      <td>115</td>\n",
       "      <td>7896</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>16000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>9</td>\n",
       "      <td>230983</td>\n",
       "      <td>115</td>\n",
       "      <td>7863</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>16000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0  row_id  timestamp  user_id  content_id  content_type_id  \\\n",
       "0           0       0          0      115        5692                0   \n",
       "1           1       1      56943      115        5716                0   \n",
       "2           2       2     118363      115         128                0   \n",
       "3           3       3     131167      115        7860                0   \n",
       "4           4       4     137965      115        7922                0   \n",
       "5           5       5     157063      115         156                0   \n",
       "6           6       6     176092      115          51                0   \n",
       "7           7       7     194190      115          50                0   \n",
       "8           8       8     212463      115        7896                0   \n",
       "9           9       9     230983      115        7863                0   \n",
       "\n",
       "   task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \n",
       "0                  1            3                   1         NaN        -1  \n",
       "1                  2            2                   1       False     37000  \n",
       "2                  0            0                   1       False     55000  \n",
       "3                  3            0                   1       False     19000  \n",
       "4                  4            1                   1       False     11000  \n",
       "5                  5            2                   1       False      5000  \n",
       "6                  6            0                   1       False     17000  \n",
       "7                  7            3                   1       False     17000  \n",
       "8                  8            2                   1       False     16000  \n",
       "9                  9            0                   1       False     16000  "
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tst_train_df = pd.read_csv('data/train_2.csv', nrows = 100)\n",
    "tst_train_df.head(10)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "6288c7a6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "21"
      ]
     },
     "execution_count": 76,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "del tst_train_df\n",
    "import gc\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5e2871f7",
   "metadata": {},
   "source": [
    "## 2.4) Попарная консистентность полей: train_df.content_id и lect_df.lecture_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "id": "b681697b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "415 418 415 415 3 0 3\n"
     ]
    }
   ],
   "source": [
    "# выполним похожую проверку для лекций\n",
    "# проверим, что в train нет значений content_id, которых нет в questions\n",
    "\n",
    "#фильтруем оп content_type_id == 1, нам нужны только лекции\n",
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = train_df[train_df.content_type_id == 1]['content_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = lect_df['lecture_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 415 418 415 415 3 0 3\n",
    "# все в порядке: в train нет лекци, отсутствющих в справочнике"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "73fc142f",
   "metadata": {},
   "source": [
    "## 2.5) Попарная консистентность полей: train_df.task_container_id и lect_df.bundle_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "id": "fb71a300",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000 9765 7476 7476 4813 2524 2289\n"
     ]
    }
   ],
   "source": [
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = train_df[train_df.content_type_id == 0]['task_container_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = quest_df['bundle_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 10000 9765 7476 7476 4813 2524 2289\n",
    "# Данне не консистентны\n",
    "# Этот результат говорит о том, что в train.task_container_id  встречаются значения, которых нет в QUESTIONS.bundle_id\n",
    "# таких значений 4813, посмотрим на строки с такими train.task_container_id:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "5dcab047",
   "metadata": {},
   "outputs": [],
   "source": [
    "#q_train_df = train_df[train_df.content_type_id == 0]\n",
    "\n",
    "#cont_id = pd.merge(left = train_df[train_df.content_type_id == 0], \n",
    "#                   right = quest_df, \n",
    "#                   how = 'left', \n",
    "#                   left_on = ['content_id','task_container_id'], \n",
    "#                   right_on = ['question_id','bundle_id'],\n",
    "#                   suffixes=('_left', '_right'),\n",
    "#                   indicator=True)\n",
    "#display(cont_id.head(20))     \n",
    "\n",
    "\n",
    "# MemoryError  - печаль, но не повод сдаваться"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "a377fea5",
   "metadata": {},
   "outputs": [],
   "source": [
    "#quest_df.filter?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "03026eea",
   "metadata": {},
   "outputs": [],
   "source": [
    "#list2 = train_df[train_df.content_type_id == 0][['content_id','task_container_id']]\n",
    "#list2['cnt_content_id'] = list2['content_id'].apply(lambda x: 1)\n",
    "#list3 = pd.Series(list2)\n",
    "#list3 = list2[['task_container_id','content_id', 'cnt_content_id']].groupby(by = ['task_container_id','content_id']).sum()\n",
    "#list3.sort_values(by = ['cnt_content_id'])\n",
    "#display(list2)\n",
    "\n",
    "\n",
    "# все равно MemoryError"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "6f7274b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "#x = tuple()\n",
    "#y = set()\n",
    "#for i, r in train_df.iterrows():\n",
    "#    if row['content_type_id'] == 0:\n",
    "#        x = (row['content_id'], row['task_container_id'])\n",
    "#        y.add(x)\n",
    "#    i += 1\n",
    "#    if i % 10 == 0: \n",
    "#         print(i)\n",
    "    \n",
    "#for i in range(0, 101230333):\n",
    "#    if train_df.loc[i].content_type_id == 0:\n",
    "#        x = (train_df.loc[i].content_id, train_df.loc[i].task_container_id)\n",
    "#        y.add(x)\n",
    "#    i += 1\n",
    "#    if i % 10000 == 0:\n",
    "#        print(i)\n",
    "#print(len(y))\n",
    "\n",
    "# все равно MemoryError  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "47919607",
   "metadata": {},
   "outputs": [],
   "source": [
    "#pd.read_csv?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "id": "30ddc185",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%reset Out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "c14b1024",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "101230333\n"
     ]
    }
   ],
   "source": [
    "#import gc\n",
    "#del train_df\n",
    "#gc.collect()\n",
    "big_file = open('data/train_2.csv', 'rb')\n",
    "line_count = 0\n",
    "for line in big_file:\n",
    "    line_count += 1\n",
    "print(line_count)\n",
    "# 101 230 333\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "ee7d8dc9",
   "metadata": {},
   "outputs": [],
   "source": [
    "#%reset Out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "256c9a45",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = pd.read_csv('data/train_2.csv',\n",
    "                       dtype= {'row_id':'int32','timestamp':'int64','user_id':'int32','content_id':'int16','content_type_id':'int8','task_container_id':'int16','user_answer':'int8','answered_correctly':'int8','prior_question_elapsed_time':'float64','pq_had_expl':'object','pqe_time':'float64' }, \n",
    "                       usecols = ['row_id','timestamp','user_id','content_id','content_type_id','task_container_id','user_answer','answered_correctly','pq_had_expl','pqe_time'],\n",
    "                       )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "id": "0f8faccc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "99271300"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "row_cnt  = train_df[train_df.content_type_id == 0].shape[0]\n",
    "row_cnt\n",
    "#99 271 300"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "c08cba53",
   "metadata": {
    "scrolled": false
   },
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>37000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>55000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>19000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>11000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>157063</td>\n",
       "      <td>115</td>\n",
       "      <td>156</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>5000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>176092</td>\n",
       "      <td>115</td>\n",
       "      <td>51</td>\n",
       "      <td>0</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>17000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>194190</td>\n",
       "      <td>115</td>\n",
       "      <td>50</td>\n",
       "      <td>0</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>17000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>212463</td>\n",
       "      <td>115</td>\n",
       "      <td>7896</td>\n",
       "      <td>0</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>16000.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>230983</td>\n",
       "      <td>115</td>\n",
       "      <td>7863</td>\n",
       "      <td>0</td>\n",
       "      <td>9</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>16000.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "5       5     157063      115         156                0                  5   \n",
       "6       6     176092      115          51                0                  6   \n",
       "7       7     194190      115          50                0                  7   \n",
       "8       8     212463      115        7896                0                  8   \n",
       "9       9     230983      115        7863                0                  9   \n",
       "\n",
       "   user_answer  answered_correctly pq_had_expl  pqe_time  \n",
       "0            3                   1         NaN      -1.0  \n",
       "1            2                   1       False   37000.0  \n",
       "2            0                   1       False   55000.0  \n",
       "3            0                   1       False   19000.0  \n",
       "4            1                   1       False   11000.0  \n",
       "5            2                   1       False    5000.0  \n",
       "6            0                   1       False   17000.0  \n",
       "7            3                   1       False   17000.0  \n",
       "8            2                   1       False   16000.0  \n",
       "9            0                   1       False   16000.0  "
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_df.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7d3c49d3",
   "metadata": {},
   "source": [
    "## 2.6) Попарная консистентность полей: train_df.task_container_id ДЛЯ ЛЕКЦИЙ и quest_df.bundle_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "d810b125",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9369 9765 6943 6943 5248 2426 2822\n"
     ]
    }
   ],
   "source": [
    "\n",
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = train_df[train_df.content_type_id == 1]['task_container_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = quest_df['bundle_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 9369 9765 6943 6943 5248 2426 2822\n",
    "# данные не консистентны. Все еще не ясно, что такое task_container_id для лекций."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "926a87e3",
   "metadata": {},
   "source": [
    "## 2.7) Попарная консистентность полей: train_df.task_container_id ДЛЯ ЛЕКЦИЙ и train_df.task_container_id ДЛЯ ВОПРОСОВ"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "id": "1fc583f7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "9369 10000 9369 9369 631 0 631\n"
     ]
    }
   ],
   "source": [
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = train_df[train_df.content_type_id == 1]['task_container_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = train_df[train_df.content_type_id == 0]['task_container_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 9369 10000 9369 9369 631 0 631\n",
    "# таким образом все бандлы лекций есть и у вопросов. Обратное не верно.\n",
    "# Делаем вывод, что лекции поставляются в одном бандле с вопросами, то есть \n",
    "# у них нет своей группировки в бандлы и они наследую бандл от той группы вопросов, в рамках которой они были показаны студенту"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d9ca4c23",
   "metadata": {},
   "source": [
    "## 2.8) Вывод из 2.5 - 2.8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "id": "e747b554",
   "metadata": {},
   "outputs": [],
   "source": [
    "# мы нашли такие task_container_id, каких нет в questions, но пока не ясно, что это значит. Удалять эти строки пока не будем.\n",
    "# Дальнейшие попытки исследовать этот момент заканчиваются MemoryError, \n",
    "# поэтому просто запишем в фрейм еще один столбец-индикатор того, есть ли соотвествующий task_container_id среди bundle_id:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "id": "8d1708a0",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = train_df[train_df.content_type_id == 0]['task_container_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = quest_df['bundle_id'].unique()\n",
    "set2 = set(list2)\n",
    "dset = set1 - set2\n",
    "\n",
    "train_df['cont_is_bundle_ind'] = train_df['task_container_id'].apply(lambda x: 0 if x in dset else 1)    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "0f540206",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>115</td>\n",
       "      <td>5692</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>56943</td>\n",
       "      <td>115</td>\n",
       "      <td>5716</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>37000.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>118363</td>\n",
       "      <td>115</td>\n",
       "      <td>128</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>55000.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>131167</td>\n",
       "      <td>115</td>\n",
       "      <td>7860</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>19000.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>137965</td>\n",
       "      <td>115</td>\n",
       "      <td>7922</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>False</td>\n",
       "      <td>11000.0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   row_id  timestamp  user_id  content_id  content_type_id  task_container_id  \\\n",
       "0       0          0      115        5692                0                  1   \n",
       "1       1      56943      115        5716                0                  2   \n",
       "2       2     118363      115         128                0                  0   \n",
       "3       3     131167      115        7860                0                  3   \n",
       "4       4     137965      115        7922                0                  4   \n",
       "\n",
       "   user_answer  answered_correctly pq_had_expl  pqe_time  cont_is_bundle_ind  \n",
       "0            3                   1         NaN      -1.0                   1  \n",
       "1            2                   1       False   37000.0                   1  \n",
       "2            0                   1       False   55000.0                   1  \n",
       "3            0                   1       False   19000.0                   1  \n",
       "4            1                   1       False   11000.0                   1  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3552</th>\n",
       "      <td>3552</td>\n",
       "      <td>5253595800</td>\n",
       "      <td>24418</td>\n",
       "      <td>218</td>\n",
       "      <td>0</td>\n",
       "      <td>1401</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>18000.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3553</th>\n",
       "      <td>3553</td>\n",
       "      <td>5253620336</td>\n",
       "      <td>24418</td>\n",
       "      <td>505</td>\n",
       "      <td>0</td>\n",
       "      <td>1402</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>16000.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3556</th>\n",
       "      <td>3556</td>\n",
       "      <td>5253693471</td>\n",
       "      <td>24418</td>\n",
       "      <td>779</td>\n",
       "      <td>0</td>\n",
       "      <td>1404</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>18000.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3557</th>\n",
       "      <td>3557</td>\n",
       "      <td>5253717572</td>\n",
       "      <td>24418</td>\n",
       "      <td>828</td>\n",
       "      <td>0</td>\n",
       "      <td>1405</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>17000.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3559</th>\n",
       "      <td>3559</td>\n",
       "      <td>5253801393</td>\n",
       "      <td>24418</td>\n",
       "      <td>986</td>\n",
       "      <td>0</td>\n",
       "      <td>1407</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>17000.0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      row_id   timestamp  user_id  content_id  content_type_id  \\\n",
       "3552    3552  5253595800    24418         218                0   \n",
       "3553    3553  5253620336    24418         505                0   \n",
       "3556    3556  5253693471    24418         779                0   \n",
       "3557    3557  5253717572    24418         828                0   \n",
       "3559    3559  5253801393    24418         986                0   \n",
       "\n",
       "      task_container_id  user_answer  answered_correctly pq_had_expl  \\\n",
       "3552               1401            3                   1        True   \n",
       "3553               1402            0                   1        True   \n",
       "3556               1404            0                   1        True   \n",
       "3557               1405            3                   1        True   \n",
       "3559               1407            3                   0        True   \n",
       "\n",
       "      pqe_time  cont_is_bundle_ind  \n",
       "3552   18000.0                   0  \n",
       "3553   16000.0                   0  \n",
       "3556   18000.0                   0  \n",
       "3557   17000.0                   0  \n",
       "3559   17000.0                   0  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(train_df[train_df['cont_is_bundle_ind'] == 1].head(5))\n",
    "display(train_df[train_df['cont_is_bundle_ind'] == 0].head(5))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "80cf1966",
   "metadata": {},
   "source": [
    "## 3.1) Корректность полей: корректность поля answered_correctly"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "4626a369",
   "metadata": {},
   "outputs": [],
   "source": [
    "# в quet есть индикатор корректного ответа и сам ответ. Проверим по quest_df.correct_answer, верно ли заполнен индикатор."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "b42d4688",
   "metadata": {},
   "outputs": [],
   "source": [
    "# сведем задачу к консистентности полей, но будем сравнивать не поле с полем, а пару полей с парой полей"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4c9410c9",
   "metadata": {},
   "source": [
    "### 3.1.1) Корректность строк с answered_correctly =  1`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "cd599196",
   "metadata": {},
   "outputs": [],
   "source": [
    "# напомним, что все id вопросов, которые есть в train, есть и в quest, мы проверили это на этапе проверок консистентности"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "9800ce2d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13523, 2)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corr_quest_df =quest_df[['question_id','correct_answer']]\n",
    "corr_quest_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "81dbea6d",
   "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>content_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>6266</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6589</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>24890</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>17906</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>19461</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>8383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>2</td>\n",
       "      <td>26910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>0</td>\n",
       "      <td>14363</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "      <td>7738</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>3</td>\n",
       "      <td>14389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>21219</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>3</td>\n",
       "      <td>5937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>1</td>\n",
       "      <td>15421</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>1</td>\n",
       "      <td>7977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>2</td>\n",
       "      <td>12095</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>15</td>\n",
       "      <td>3</td>\n",
       "      <td>14321</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>3</td>\n",
       "      <td>2315</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>0</td>\n",
       "      <td>7396</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "      <td>17362</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>9203</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    content_id  user_answer      0\n",
       "0            0            0   6266\n",
       "1            1            1   6589\n",
       "2            2            0  24890\n",
       "3            3            0  17906\n",
       "4            4            3  19461\n",
       "5            5            2   8383\n",
       "6            6            2  26910\n",
       "7            7            0  14363\n",
       "8            8            3   7738\n",
       "9            9            3  14389\n",
       "10          10            0  21219\n",
       "11          11            3   5937\n",
       "12          12            1  15421\n",
       "13          13            1   7977\n",
       "14          14            2  12095\n",
       "15          15            3  14321\n",
       "16          16            3   2315\n",
       "17          17            0   7396\n",
       "18          18            1  17362\n",
       "19          19            0   9203"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "corr_train_df = train_df[(train_df.answered_correctly == 1) \n",
    "                         & (train_df.content_type_id == 0)].groupby(['content_id','user_answer']).size().reset_index(level=[0,1])\n",
    "display(corr_train_df.head(20))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "id": "2cb3a3ea",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13519, 3)"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "corr_train_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "id": "84e1fe24",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13523, 5)"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "comp_df = pd.merge(left=corr_quest_df, \n",
    "                   right=corr_train_df, \n",
    "                   how='outer',\n",
    "                   left_on = 'question_id', \n",
    "                   right_on='content_id', \n",
    "                   validate = '1:m',\n",
    "                   suffixes = ('_quest','_train'))\n",
    "comp_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "id": "874907c8",
   "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>question_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>content_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1484</th>\n",
       "      <td>1484</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1485</th>\n",
       "      <td>1485</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1486</th>\n",
       "      <td>1486</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10007</th>\n",
       "      <td>10007</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       question_id  correct_answer  content_id  user_answer   0\n",
       "1484          1484               2         NaN          NaN NaN\n",
       "1485          1485               0         NaN          NaN NaN\n",
       "1486          1486               3         NaN          NaN NaN\n",
       "10007        10007               0         NaN          NaN NaN"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "comp_df[(comp_df.user_answer != comp_df.correct_answer)]\n",
    "# все строки должны быть с NaN в content_id\n",
    "# так и есть, значит все правильно"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "81ab2001",
   "metadata": {},
   "source": [
    "### 3.1.1) Корректность строк с answered_correctly =  0`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "id": "e7424720",
   "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>content_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>342</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>86</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>561</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>74</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   content_id  user_answer    0\n",
       "0           0            1  342\n",
       "1           0            2  209\n",
       "2           0            3   86\n",
       "3           1            0  561\n",
       "4           1            2   74"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(38570, 5)\n"
     ]
    },
    {
     "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>question_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>content_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>0</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [question_id, correct_answer, content_id, user_answer, 0]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "corr_train_df = train_df[(train_df.answered_correctly == 0) \n",
    "                         & (train_df.content_type_id == 0)].groupby(['content_id','user_answer']).size().reset_index(level=[0,1])\n",
    "display(corr_train_df.head())\n",
    "comp_df = pd.merge(left=corr_quest_df, \n",
    "                   right=corr_train_df, \n",
    "                   how='outer',\n",
    "                   left_on = 'question_id', \n",
    "                   right_on='content_id', \n",
    "                   validate = '1:m',\n",
    "                   suffixes = ('_quest','_train'))\n",
    "print(comp_df.shape)\n",
    "display(comp_df[(comp_df.user_answer == comp_df.correct_answer)])\n",
    "# убедились, что нет строк, в которых ответ совпадат с правильным ответом и при это они помечены answered_correctly = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b604707c",
   "metadata": {},
   "source": [
    "## 3.2 Проверка корректности заполнения полей user_answer, answered_correctly _для лекций_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "c548e017",
   "metadata": {},
   "outputs": [],
   "source": [
    "# теперь проверим следующие утверждения из постановки:\n",
    "# user_answer: ... Read -1 as null, for lectures.\n",
    "# answered_correctly: ... if the user responded correctly. Read -1 as null, for lectures.\n",
    "# сделаем это сразу несоклькими эквивалентными методами, чтобы потренироваться."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "7fc1961a",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(train_df[(train_df.content_type_id == 1) & ((train_df.user_answer != -1) |  (train_df.answered_correctly != -1))])\n",
    "display(train_df.query('content_type_id == 1 and (user_answer != -1 or  answered_correctly != -1)'))\n",
    "display(train_df.loc[(train_df.content_type_id == 1) & ((train_df.user_answer != -1) |  (train_df.answered_correctly != -1))])\n",
    "\n",
    "#мы убедились, что лекции не содержат в этих полях ничего кроме -1\n",
    "# но можеть быть есть не-лекции (вопросы), которые содержат -1 в этих полях? Этого тоже не дожно быть, проверим:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "9e749db4",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "Empty DataFrame\n",
       "Columns: [row_id, timestamp, user_id, content_id, content_type_id, task_container_id, user_answer, answered_correctly, pq_had_expl, pqe_time, cont_is_bundle_ind]\n",
       "Index: []"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(train_df[(train_df.content_type_id == 0) & ((train_df.user_answer == -1) |  (train_df.answered_correctly == -1))])\n",
    "display(train_df.query('content_type_id == 0 and (user_answer == -1 or  answered_correctly == -1)'))\n",
    "display(train_df.loc[(train_df.content_type_id == 0) & ((train_df.user_answer == -1) |  (train_df.answered_correctly == -1))])\n",
    "# все хорошо, результат - пустое множество"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd605c4e",
   "metadata": {},
   "source": [
    "## 3.3 Проверка корректности LECTURES.tag"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "63804d43",
   "metadata": {},
   "outputs": [],
   "source": [
    "# в условии сказано, что одни tag для каждой лекции, проверим визуально:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "4a5dac6f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([159,  70,  45,  79, 156, 114, 179, 134,  93,  80,  98, 101,  33,\n",
       "        74,  60, 181,  43,   4, 109,  89, 142, 140, 104,   1,  65, 155,\n",
       "        26, 178, 173,   7, 147, 171,  78,  62,  64, 141, 124, 161,  24,\n",
       "        23, 152,   8,  94, 119,  37, 177,  54, 157, 132,   6,  67, 123,\n",
       "        50,  25, 175, 116, 136,  20, 129, 127,  53, 130,  13,  42,  72,\n",
       "        36, 172,  57, 151,   9, 187,  58, 166,  49, 164, 163,  47, 139,\n",
       "        85,  91, 158,  48,  69,  76,  28, 167,  18,   0,  27,  19, 149,\n",
       "        99,  39,  34,  61,  14, 113, 182,  55, 145,  96, 133,  12, 106,\n",
       "       126, 105, 125, 168,   5,  71,  82,  44, 108, 143,  90,  16, 112,\n",
       "        73,  40, 111,  51,  97,  31,  52, 180, 138, 150, 120, 128,  68,\n",
       "        95, 170, 184,  30, 174, 153,  66,  17,  75,  83, 183, 148, 144,\n",
       "       103, 118,  15, 115, 176, 137, 169,   2], dtype=uint8)"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lect_df.tag.unique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0a311d45",
   "metadata": {},
   "source": [
    "# 4) Обогащение данных train данными из lectures и questions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "1a127fb5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# так как файл train очень большой, а при исследовании могуть потребоваться отсутствующие в train данные из lect и quest\n",
    "# постоянно делать merge не удобно, а удобно один раз сделать это и работать уже с обогащенным фреймом\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "id": "07fcb56e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# так как мой комп не тянет объединения train c чем либо еще, \n",
    "# а делать это в цикле и потом объединять сабс-фреймы не кажется интеллектуальной задачей,\n",
    "# я продемонстрирую свои навыки на укороченном фрейме train # в которм 1 000 000 записей\n",
    "\n",
    "# так же я не буду писать единый merge трех фреймов, а сделаю это поочереди"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "746f8ffd",
   "metadata": {},
   "source": [
    "## 4.0) Укороченный фрейм"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "id": "ddf469e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "sample_train_df = train_df.head(2000000).tail(1000000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "b9781a4f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000000, 11)"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_train_df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20682604",
   "metadata": {},
   "source": [
    "## 4.1) Обогатим train полями из lectures"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "id": "d9eeb3b3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "418 13523 158 158 13625 260 13365\n"
     ]
    }
   ],
   "source": [
    "# проверим, что множество lecture_id НЕ ПЕРЕСЕКАЕТСЯ с множеством question_id:\n",
    "set1 = set()\n",
    "set2 = set()\n",
    "list1 = lect_df['lecture_id'].unique()\n",
    "set1 = set(list1)\n",
    "list2 = quest_df['question_id'].unique()\n",
    "set2 = set(list2)\n",
    "print(len(set1), len(set2), len(set1 & set2), len(set2 & set1), len(set1 ^ set2), len(set1-set2), len(set2-set1))\n",
    "# 418 9765 122 122 9939 296 9643\n",
    "#!!! Оказалось, что пересекаются! Надем примеры:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "b536fa7f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{89,\n",
       " 100,\n",
       " 185,\n",
       " 192,\n",
       " 317,\n",
       " 335,\n",
       " 484,\n",
       " 641,\n",
       " 761,\n",
       " 814,\n",
       " 1003,\n",
       " 1077,\n",
       " 1105,\n",
       " 1144,\n",
       " 1253,\n",
       " 1308,\n",
       " 1450,\n",
       " 1501,\n",
       " 1604,\n",
       " 1731,\n",
       " 1887,\n",
       " 1907,\n",
       " 1967,\n",
       " 2033,\n",
       " 2049,\n",
       " 2084,\n",
       " 2278,\n",
       " 2282,\n",
       " 2299,\n",
       " 2327,\n",
       " 2476,\n",
       " 2771,\n",
       " 2857,\n",
       " 2904,\n",
       " 2918,\n",
       " 3078,\n",
       " 3113,\n",
       " 3123,\n",
       " 3153,\n",
       " 3255,\n",
       " 3625,\n",
       " 3668,\n",
       " 3697,\n",
       " 3780,\n",
       " 3803,\n",
       " 3852,\n",
       " 3879,\n",
       " 3937,\n",
       " 4057,\n",
       " 4072,\n",
       " 4100,\n",
       " 4118,\n",
       " 4173,\n",
       " 4187,\n",
       " 4385,\n",
       " 4425,\n",
       " 4441,\n",
       " 4652,\n",
       " 4705,\n",
       " 4883,\n",
       " 5120,\n",
       " 5240,\n",
       " 5266,\n",
       " 5588,\n",
       " 5694,\n",
       " 5752,\n",
       " 5990,\n",
       " 6081,\n",
       " 6101,\n",
       " 6137,\n",
       " 6220,\n",
       " 6316,\n",
       " 6481,\n",
       " 6619,\n",
       " 6761,\n",
       " 6797,\n",
       " 6808,\n",
       " 6948,\n",
       " 7008,\n",
       " 7010,\n",
       " 7097,\n",
       " 7385,\n",
       " 7477,\n",
       " 7704,\n",
       " 7785,\n",
       " 7877,\n",
       " 7920,\n",
       " 7926,\n",
       " 7949,\n",
       " 8138,\n",
       " 8207,\n",
       " 8233,\n",
       " 8240,\n",
       " 8288,\n",
       " 8317,\n",
       " 8320,\n",
       " 8330,\n",
       " 8411,\n",
       " 8461,\n",
       " 8513,\n",
       " 8693,\n",
       " 8708,\n",
       " 8821,\n",
       " 8976,\n",
       " 8977,\n",
       " 9174,\n",
       " 9212,\n",
       " 9245,\n",
       " 9477,\n",
       " 9554,\n",
       " 9758,\n",
       " 9831,\n",
       " 9835,\n",
       " 9856,\n",
       " 9896,\n",
       " 10010,\n",
       " 10032,\n",
       " 10040,\n",
       " 10052,\n",
       " 10075,\n",
       " 10180,\n",
       " 10217,\n",
       " 10354,\n",
       " 10382,\n",
       " 10383,\n",
       " 10540,\n",
       " 10688,\n",
       " 10742,\n",
       " 10852,\n",
       " 10928,\n",
       " 11151,\n",
       " 11230,\n",
       " 11236,\n",
       " 11485,\n",
       " 11559,\n",
       " 11772,\n",
       " 11873,\n",
       " 11877,\n",
       " 12007,\n",
       " 12187,\n",
       " 12294,\n",
       " 12354,\n",
       " 12469,\n",
       " 12472,\n",
       " 12509,\n",
       " 12638,\n",
       " 12655,\n",
       " 12811,\n",
       " 12831,\n",
       " 12846,\n",
       " 12942,\n",
       " 13138,\n",
       " 13256,\n",
       " 13336,\n",
       " 13379,\n",
       " 13384,\n",
       " 13471,\n",
       " 13487}"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set1 & set2\n",
    "\n",
    "# выберем значение 192"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "id": "1fafde07",
   "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>lecture_id</th>\n",
       "      <th>tag</th>\n",
       "      <th>part</th>\n",
       "      <th>type_of</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>192</td>\n",
       "      <td>79</td>\n",
       "      <td>5</td>\n",
       "      <td>solving question</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   lecture_id  tag  part           type_of\n",
       "3         192   79     5  solving question"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "      <th>tags</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>192</th>\n",
       "      <td>192</td>\n",
       "      <td>192</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>131 104 38</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     question_id  bundle_id  correct_answer  part        tags\n",
       "192          192        192               1     1  131 104 38"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(lect_df[lect_df['lecture_id'] == 192])\n",
    "display(quest_df[quest_df['question_id'] == 192])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "id": "7612e55f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# мы убедились, что, увы нельзя делать объединение train с другими двумя фреймами просто по id.\n",
    "\n",
    "\n",
    "# придумаем структуру полей обогащенного фрейма:\n",
    "#    сначала все поля train\n",
    "#    l_tag: tag лекции, -1 для вопроса\n",
    "#    l_part: part для лекции, -1 для вопроса\n",
    "#    type_of_lect: type_of для лекции, nan для вопроса\n",
    "#    q_part: part для вопроса, -1 для лекции. Не объединяем в одни столбец c l_part, судя по всему разные сущности\n",
    "#    q_tags: тэг для вопроса, nan для лекции \n",
    "\n",
    "lect_df['ind'] = 1\n",
    "\n",
    "# обхединяемся с lect\n",
    "ext_sample_train_df = sample_train_df.merge(right=lect_df,\n",
    "                                            how='left',\n",
    "                                            validate='m:1',\n",
    "                                            left_on = ['content_id','content_type_id'],\n",
    "                                            right_on = ['lecture_id','ind']\n",
    "                                           )\n",
    "#extr_train_df    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "id": "34772f1b",
   "metadata": {
    "scrolled": true
   },
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>lecture_id</th>\n",
       "      <th>tag</th>\n",
       "      <th>part</th>\n",
       "      <th>type_of</th>\n",
       "      <th>ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1000012</td>\n",
       "      <td>139567501</td>\n",
       "      <td>20949024</td>\n",
       "      <td>29544</td>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>29544.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1000017</td>\n",
       "      <td>172554699</td>\n",
       "      <td>20949024</td>\n",
       "      <td>32312</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>32312.0</td>\n",
       "      <td>176.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "12  1000012  139567501  20949024       29544                1   \n",
       "17  1000017  172554699  20949024       32312                1   \n",
       "\n",
       "    task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "12                 47           -1                  -1       False      -1.0   \n",
       "17                 51           -1                  -1       False      -1.0   \n",
       "\n",
       "    cont_is_bundle_ind  lecture_id    tag  part  type_of  ind  \n",
       "12                   1     29544.0   30.0   2.0  concept  1.0  \n",
       "17                   1     32312.0  176.0   2.0  concept  1.0  "
      ]
     },
     "execution_count": 117,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ext_sample_train_df[ext_sample_train_df.content_type_id == 1].head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "id": "2a59b617",
   "metadata": {},
   "outputs": [],
   "source": [
    "ext_sample_train_df.drop(inplace=True, axis=1, labels = ['lecture_id', 'ind'])\n",
    "ext_sample_train_df.rename(inplace=True, columns = {'tag':'l_tag','part':'l_part','type_of':'type_of_lect'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "id": "cd89c483",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1000012</td>\n",
       "      <td>139567501</td>\n",
       "      <td>20949024</td>\n",
       "      <td>29544</td>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>30.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1000017</td>\n",
       "      <td>172554699</td>\n",
       "      <td>20949024</td>\n",
       "      <td>32312</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>176.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "12  1000012  139567501  20949024       29544                1   \n",
       "17  1000017  172554699  20949024       32312                1   \n",
       "\n",
       "    task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "12                 47           -1                  -1       False      -1.0   \n",
       "17                 51           -1                  -1       False      -1.0   \n",
       "\n",
       "    cont_is_bundle_ind  l_tag  l_part type_of_lect  \n",
       "12                   1   30.0     2.0      concept  \n",
       "17                   1  176.0     2.0      concept  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000000</td>\n",
       "      <td>26624921</td>\n",
       "      <td>20949024</td>\n",
       "      <td>8207</td>\n",
       "      <td>0</td>\n",
       "      <td>34</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>9000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1000001</td>\n",
       "      <td>26641933</td>\n",
       "      <td>20949024</td>\n",
       "      <td>5583</td>\n",
       "      <td>0</td>\n",
       "      <td>35</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>14000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "0  1000000   26624921  20949024        8207                0   \n",
       "1  1000001   26641933  20949024        5583                0   \n",
       "\n",
       "   task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "0                 34            1                   1        True    9000.0   \n",
       "1                 35            0                   0        True   14000.0   \n",
       "\n",
       "   cont_is_bundle_ind  l_tag  l_part type_of_lect  \n",
       "0                   1    NaN     NaN          NaN  \n",
       "1                   1    NaN     NaN          NaN  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 1].head(2))\n",
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 0].head(2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "id": "2ad7c97a",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "      <th>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "      <th>tags</th>\n",
       "      <th>ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1000012</td>\n",
       "      <td>139567501</td>\n",
       "      <td>20949024</td>\n",
       "      <td>29544</td>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>30.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1000017</td>\n",
       "      <td>172554699</td>\n",
       "      <td>20949024</td>\n",
       "      <td>32312</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>176.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "12  1000012  139567501  20949024       29544                1   \n",
       "17  1000017  172554699  20949024       32312                1   \n",
       "\n",
       "    task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "12                 47           -1                  -1       False      -1.0   \n",
       "17                 51           -1                  -1       False      -1.0   \n",
       "\n",
       "    cont_is_bundle_ind  l_tag  l_part type_of_lect  question_id  bundle_id  \\\n",
       "12                   1   30.0     2.0      concept          NaN        NaN   \n",
       "17                   1  176.0     2.0      concept          NaN        NaN   \n",
       "\n",
       "    correct_answer  part tags  ind  \n",
       "12             NaN   NaN  NaN  NaN  \n",
       "17             NaN   NaN  NaN  NaN  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "      <th>question_id</th>\n",
       "      <th>bundle_id</th>\n",
       "      <th>correct_answer</th>\n",
       "      <th>part</th>\n",
       "      <th>tags</th>\n",
       "      <th>ind</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000000</td>\n",
       "      <td>26624921</td>\n",
       "      <td>20949024</td>\n",
       "      <td>8207</td>\n",
       "      <td>0</td>\n",
       "      <td>34</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>9000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>8207.0</td>\n",
       "      <td>8207.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>8</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1000001</td>\n",
       "      <td>26641933</td>\n",
       "      <td>20949024</td>\n",
       "      <td>5583</td>\n",
       "      <td>0</td>\n",
       "      <td>35</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>14000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5583.0</td>\n",
       "      <td>5583.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>54</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "0  1000000   26624921  20949024        8207                0   \n",
       "1  1000001   26641933  20949024        5583                0   \n",
       "\n",
       "   task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "0                 34            1                   1        True    9000.0   \n",
       "1                 35            0                   0        True   14000.0   \n",
       "\n",
       "   cont_is_bundle_ind  l_tag  l_part type_of_lect  question_id  bundle_id  \\\n",
       "0                   1    NaN     NaN          NaN       8207.0     8207.0   \n",
       "1                   1    NaN     NaN          NaN       5583.0     5583.0   \n",
       "\n",
       "   correct_answer  part tags  ind  \n",
       "0             1.0   5.0    8  0.0  \n",
       "1             2.0   5.0   54  0.0  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "quest_df['ind'] = 0\n",
    "\n",
    "# обхединяемся с quest\n",
    "ext_sample_train_df = ext_sample_train_df.merge(right=quest_df,\n",
    "                                            how='left',\n",
    "                                            validate='m:1',\n",
    "                                            left_on = ['content_id','content_type_id'],\n",
    "                                            right_on = ['question_id','ind']\n",
    "                                           )\n",
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 1].head(2))\n",
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 0].head(2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "id": "ecf47ab2",
   "metadata": {},
   "outputs": [],
   "source": [
    "ext_sample_train_df.drop(inplace=True, axis=1, labels = ['question_id','bundle_id','ind','correct_answer'])\n",
    "ext_sample_train_df.rename(inplace=True, columns = {'tags':'q_tags','part':'q_part'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "efac217e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# удалим ранее созданные колонки: quest_df['ind'] = 0 lect_df['ind'] = 1\n",
    "quest_df.drop(inplace=True, axis=1, labels = ['ind'])\n",
    "lect_df.drop(inplace=True, axis=1, labels = ['ind'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "fe8ba695",
   "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "      <th>q_part</th>\n",
       "      <th>q_tags</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1000012</td>\n",
       "      <td>139567501</td>\n",
       "      <td>20949024</td>\n",
       "      <td>29544</td>\n",
       "      <td>1</td>\n",
       "      <td>47</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>30.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>1000017</td>\n",
       "      <td>172554699</td>\n",
       "      <td>20949024</td>\n",
       "      <td>32312</td>\n",
       "      <td>1</td>\n",
       "      <td>51</td>\n",
       "      <td>-1</td>\n",
       "      <td>-1</td>\n",
       "      <td>False</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>176.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>concept</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "12  1000012  139567501  20949024       29544                1   \n",
       "17  1000017  172554699  20949024       32312                1   \n",
       "\n",
       "    task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "12                 47           -1                  -1       False      -1.0   \n",
       "17                 51           -1                  -1       False      -1.0   \n",
       "\n",
       "    cont_is_bundle_ind  l_tag  l_part type_of_lect  q_part q_tags  \n",
       "12                   1   30.0     2.0      concept     NaN    NaN  \n",
       "17                   1  176.0     2.0      concept     NaN    NaN  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "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>row_id</th>\n",
       "      <th>timestamp</th>\n",
       "      <th>user_id</th>\n",
       "      <th>content_id</th>\n",
       "      <th>content_type_id</th>\n",
       "      <th>task_container_id</th>\n",
       "      <th>user_answer</th>\n",
       "      <th>answered_correctly</th>\n",
       "      <th>pq_had_expl</th>\n",
       "      <th>pqe_time</th>\n",
       "      <th>cont_is_bundle_ind</th>\n",
       "      <th>l_tag</th>\n",
       "      <th>l_part</th>\n",
       "      <th>type_of_lect</th>\n",
       "      <th>q_part</th>\n",
       "      <th>q_tags</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1000000</td>\n",
       "      <td>26624921</td>\n",
       "      <td>20949024</td>\n",
       "      <td>8207</td>\n",
       "      <td>0</td>\n",
       "      <td>34</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>True</td>\n",
       "      <td>9000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5.0</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1000001</td>\n",
       "      <td>26641933</td>\n",
       "      <td>20949024</td>\n",
       "      <td>5583</td>\n",
       "      <td>0</td>\n",
       "      <td>35</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>True</td>\n",
       "      <td>14000.0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>5.0</td>\n",
       "      <td>54</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    row_id  timestamp   user_id  content_id  content_type_id  \\\n",
       "0  1000000   26624921  20949024        8207                0   \n",
       "1  1000001   26641933  20949024        5583                0   \n",
       "\n",
       "   task_container_id  user_answer  answered_correctly pq_had_expl  pqe_time  \\\n",
       "0                 34            1                   1        True    9000.0   \n",
       "1                 35            0                   0        True   14000.0   \n",
       "\n",
       "   cont_is_bundle_ind  l_tag  l_part type_of_lect  q_part q_tags  \n",
       "0                   1    NaN     NaN          NaN     5.0      8  \n",
       "1                   1    NaN     NaN          NaN     5.0     54  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 1].head(2))\n",
    "display(ext_sample_train_df[ext_sample_train_df.content_type_id == 0].head(2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "db9d5de7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000000 entries, 0 to 999999\n",
      "Data columns (total 16 columns):\n",
      " #   Column              Non-Null Count    Dtype   \n",
      "---  ------              --------------    -----   \n",
      " 0   row_id              1000000 non-null  int32   \n",
      " 1   timestamp           1000000 non-null  int64   \n",
      " 2   user_id             1000000 non-null  int32   \n",
      " 3   content_id          1000000 non-null  int16   \n",
      " 4   content_type_id     1000000 non-null  int8    \n",
      " 5   task_container_id   1000000 non-null  int16   \n",
      " 6   user_answer         1000000 non-null  int8    \n",
      " 7   answered_correctly  1000000 non-null  int8    \n",
      " 8   pq_had_expl         996123 non-null   object  \n",
      " 9   pqe_time            1000000 non-null  float64 \n",
      " 10  cont_is_bundle_ind  1000000 non-null  int64   \n",
      " 11  l_tag               19076 non-null    float64 \n",
      " 12  l_part              19076 non-null    float64 \n",
      " 13  type_of_lect        19076 non-null    category\n",
      " 14  q_part              980924 non-null   float64 \n",
      " 15  q_tags              980924 non-null   object  \n",
      "dtypes: category(1), float64(4), int16(2), int32(2), int64(2), int8(3), object(2)\n",
      "memory usage: 83.9+ MB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 418 entries, 0 to 417\n",
      "Data columns (total 4 columns):\n",
      " #   Column      Non-Null Count  Dtype   \n",
      "---  ------      --------------  -----   \n",
      " 0   lecture_id  418 non-null    int16   \n",
      " 1   tag         418 non-null    uint8   \n",
      " 2   part        418 non-null    int8    \n",
      " 3   type_of     418 non-null    category\n",
      "dtypes: category(1), int16(1), int8(1), uint8(1)\n",
      "memory usage: 2.4 KB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 13523 entries, 0 to 13522\n",
      "Data columns (total 5 columns):\n",
      " #   Column          Non-Null Count  Dtype \n",
      "---  ------          --------------  ----- \n",
      " 0   question_id     13523 non-null  int16 \n",
      " 1   bundle_id       13523 non-null  int16 \n",
      " 2   correct_answer  13523 non-null  int8  \n",
      " 3   part            13523 non-null  int8  \n",
      " 4   tags            13522 non-null  object\n",
      "dtypes: int16(2), int8(2), object(1)\n",
      "memory usage: 185.0+ KB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(ext_sample_train_df.info())\n",
    "display(lect_df.info())\n",
    "display(quest_df.info())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "0f698816",
   "metadata": {},
   "outputs": [],
   "source": [
    "# заменим nan прежде чем преобразовывать столбцы к другим типам:\n",
    "for col in ['l_part','q_part','l_tag']:\n",
    "    ext_sample_train_df[col] = ext_sample_train_df[col].apply(lambda x: -1 if m.isnan(x) == True else int(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "0ba0a001",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# изменим типы данных столбцов:\n",
    "ext_sample_train_df[['l_part','q_part']] =  ext_sample_train_df[['l_part','q_part']].astype('int8')\n",
    "ext_sample_train_df[['l_tag']] = ext_sample_train_df[['l_tag']].astype('int16') # тут мы отказались от uint8 в пользу более широкого типа, так как тепрь в столбце есть -1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "id": "7a8ad418",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000000 entries, 0 to 999999\n",
      "Data columns (total 16 columns):\n",
      " #   Column              Non-Null Count    Dtype   \n",
      "---  ------              --------------    -----   \n",
      " 0   row_id              1000000 non-null  int32   \n",
      " 1   timestamp           1000000 non-null  int64   \n",
      " 2   user_id             1000000 non-null  int32   \n",
      " 3   content_id          1000000 non-null  int16   \n",
      " 4   content_type_id     1000000 non-null  int8    \n",
      " 5   task_container_id   1000000 non-null  int16   \n",
      " 6   user_answer         1000000 non-null  int8    \n",
      " 7   answered_correctly  1000000 non-null  int8    \n",
      " 8   pq_had_expl         996123 non-null   object  \n",
      " 9   pqe_time            1000000 non-null  float64 \n",
      " 10  cont_is_bundle_ind  1000000 non-null  int64   \n",
      " 11  l_tag               1000000 non-null  int16   \n",
      " 12  l_part              1000000 non-null  int8    \n",
      " 13  type_of_lect        19076 non-null    category\n",
      " 14  q_part              1000000 non-null  int8    \n",
      " 15  q_tags              980924 non-null   object  \n",
      "dtypes: category(1), float64(1), int16(3), int32(2), int64(2), int8(5), object(2)\n",
      "memory usage: 64.9+ MB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "display(ext_sample_train_df.info())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "04bfbc65",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.series.Series'>\n",
      "Int64Index: 1000000 entries, 0 to 999999\n",
      "Series name: q_tags\n",
      "Non-Null Count   Dtype \n",
      "--------------   ----- \n",
      "980924 non-null  object\n",
      "dtypes: object(1)\n",
      "memory usage: 15.3+ MB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.series.Series'>\n",
      "Int64Index: 1000000 entries, 0 to 999999\n",
      "Series name: q_tags\n",
      "Non-Null Count   Dtype   \n",
      "--------------   -----   \n",
      "980924 non-null  category\n",
      "dtypes: category(1)\n",
      "memory usage: 9.6 MB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# при сокращении объема quest_df мы отказались от категоризации tags, так как уникальных значений много\n",
    "# но в ext_sample_train_df это имеет смысл, так как строк тоже много, докажем это:\n",
    "ser1 = ext_sample_train_df['q_tags']\n",
    "ser2 = ext_sample_train_df['q_tags'].astype('category')\n",
    "display(ser1.info())\n",
    "display(ser2.info())\n",
    "# Выигрыш 5.7 мегабайт на миллион строк, то есть 570 Мб на 100 млн строк. Пол гига не помешают, это 8% от фрейма"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "4bdd1fd5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000000 entries, 0 to 999999\n",
      "Data columns (total 16 columns):\n",
      " #   Column              Non-Null Count    Dtype   \n",
      "---  ------              --------------    -----   \n",
      " 0   row_id              1000000 non-null  int32   \n",
      " 1   timestamp           1000000 non-null  int64   \n",
      " 2   user_id             1000000 non-null  int32   \n",
      " 3   content_id          1000000 non-null  int16   \n",
      " 4   content_type_id     1000000 non-null  int8    \n",
      " 5   task_container_id   1000000 non-null  int16   \n",
      " 6   user_answer         1000000 non-null  int8    \n",
      " 7   answered_correctly  1000000 non-null  int8    \n",
      " 8   pq_had_expl         996123 non-null   object  \n",
      " 9   pqe_time            1000000 non-null  float64 \n",
      " 10  cont_is_bundle_ind  1000000 non-null  int64   \n",
      " 11  l_tag               1000000 non-null  int16   \n",
      " 12  l_part              1000000 non-null  int8    \n",
      " 13  type_of_lect        19076 non-null    category\n",
      " 14  q_part              1000000 non-null  int8    \n",
      " 15  q_tags              980924 non-null   category\n",
      "dtypes: category(2), float64(1), int16(3), int32(2), int64(2), int8(5), object(1)\n",
      "memory usage: 59.2+ MB\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "None"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ext_sample_train_df['q_tags'] = ext_sample_train_df['q_tags'].astype('category')\n",
    "display(ext_sample_train_df.info())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}