eda_new.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\denis\\Anaconda3\\envs\\tf\\lib\\site-packages\\IPython\\core\\interactiveshell.py:2785: DtypeWarning: Columns (12) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  interactivity=interactivity, compiler=compiler, result=result)\n"
     ]
    }
   ],
   "source": [
    "d1 = pd.read_csv('D:/tele2_data_hack/dialogs_1.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ID</th>\n",
       "      <th>Дата.изменения.статуса</th>\n",
       "      <th>Статус</th>\n",
       "      <th>Автор</th>\n",
       "      <th>Пользователь</th>\n",
       "      <th>Категория.автора</th>\n",
       "      <th>Контент</th>\n",
       "      <th>Тип.обращения</th>\n",
       "      <th>Хост</th>\n",
       "      <th>Подписчики</th>\n",
       "      <th>Тема</th>\n",
       "      <th>Локация</th>\n",
       "      <th>Тональность</th>\n",
       "      <th>Тип</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>282893.0</td>\n",
       "      <td>2088-04-02 03:01:12</td>\n",
       "      <td>Новый</td>\n",
       "      <td>token_102441</td>\n",
       "      <td>token_111457</td>\n",
       "      <td>Обычный</td>\n",
       "      <td>'Если я перейду с тарифа мой разговор на мой о...</td>\n",
       "      <td>Комментарий</td>\n",
       "      <td>'ВКонтакте</td>\n",
       "      <td>260.0</td>\n",
       "      <td>Тарифы</td>\n",
       "      <td>Москва</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0-пост</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>282893.0</td>\n",
       "      <td>2088-04-02 03:46:55</td>\n",
       "      <td>Инцидент</td>\n",
       "      <td>token_NA</td>\n",
       "      <td>token_111457</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Тарифы</td>\n",
       "      <td>Москва</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2-инцидент</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>282893.0</td>\n",
       "      <td>2088-04-02 05:52:52</td>\n",
       "      <td>Ожидание ответа автора</td>\n",
       "      <td>token_111418</td>\n",
       "      <td>token_111438</td>\n",
       "      <td>NaN</td>\n",
       "      <td>'[xxxxx], здравствуйте. При смене тарифного пл...</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",
       "      <td>1-ответ</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>282893.0</td>\n",
       "      <td>2088-04-02 05:52:53</td>\n",
       "      <td>Ожидание ответа автора</td>\n",
       "      <td>token_NA</td>\n",
       "      <td>token_111438</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Тарифы</td>\n",
       "      <td>Санкт-Петербург</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2-инцидент</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>282893.0</td>\n",
       "      <td>2088-04-03 05:58:11</td>\n",
       "      <td>Закрытые</td>\n",
       "      <td>token_NA</td>\n",
       "      <td>token_111424</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Тарифы</td>\n",
       "      <td>Санкт-Петербург</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2-инцидент</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         ID Дата.изменения.статуса                  Статус         Автор  \\\n",
       "0  282893.0    2088-04-02 03:01:12                   Новый  token_102441   \n",
       "1  282893.0    2088-04-02 03:46:55                Инцидент      token_NA   \n",
       "2  282893.0    2088-04-02 05:52:52  Ожидание ответа автора  token_111418   \n",
       "3  282893.0    2088-04-02 05:52:53  Ожидание ответа автора      token_NA   \n",
       "4  282893.0    2088-04-03 05:58:11                Закрытые      token_NA   \n",
       "\n",
       "   Пользователь Категория.автора  \\\n",
       "0  token_111457          Обычный   \n",
       "1  token_111457              NaN   \n",
       "2  token_111438              NaN   \n",
       "3  token_111438              NaN   \n",
       "4  token_111424              NaN   \n",
       "\n",
       "                                             Контент Тип.обращения  \\\n",
       "0  'Если я перейду с тарифа мой разговор на мой о...   Комментарий   \n",
       "1                                                NaN           NaN   \n",
       "2  '[xxxxx], здравствуйте. При смене тарифного пл...           NaN   \n",
       "3                                                NaN           NaN   \n",
       "4                                                NaN           NaN   \n",
       "\n",
       "         Хост  Подписчики    Тема          Локация Тональность         Тип  \n",
       "0  'ВКонтакте       260.0  Тарифы           Москва         NaN      0-пост  \n",
       "1         NaN         NaN  Тарифы           Москва         NaN  2-инцидент  \n",
       "2         NaN         NaN     NaN              NaN         NaN     1-ответ  \n",
       "3         NaN         NaN  Тарифы  Санкт-Петербург         NaN  2-инцидент  \n",
       "4         NaN         NaN  Тарифы  Санкт-Петербург         NaN  2-инцидент  "
      ]
     },
     "execution_count": 242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [],
   "source": [
    "d2 = pd.read_csv('D:/tele2_data_hack/dialogs_2.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 244,
   "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>appeal_id</th>\n",
       "      <th>datetime</th>\n",
       "      <th>client_id</th>\n",
       "      <th>user_id</th>\n",
       "      <th>msg_type</th>\n",
       "      <th>channel_id</th>\n",
       "      <th>channel_name</th>\n",
       "      <th>channel_type</th>\n",
       "      <th>answer_time</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>323874</td>\n",
       "      <td>2088-08-08 16:44:02+03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19670.0</td>\n",
       "      <td>0</td>\n",
       "      <td>73</td>\n",
       "      <td>Webdealer</td>\n",
       "      <td>Веб-чат</td>\n",
       "      <td>12.0</td>\n",
       "      <td>Привет.))</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>323870</td>\n",
       "      <td>2088-08-08 16:44:01+03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19671.0</td>\n",
       "      <td>0</td>\n",
       "      <td>73</td>\n",
       "      <td>Webdealer</td>\n",
       "      <td>Веб-чат</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Все подписки отключены. Как это делать самосто...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>323861</td>\n",
       "      <td>2088-08-08 16:44:01+03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19418.0</td>\n",
       "      <td>0</td>\n",
       "      <td>133</td>\n",
       "      <td>Нижний Новгород B2C</td>\n",
       "      <td>Веб-чат</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Вы забыли пароль и хотите войти в кабинет?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>323850</td>\n",
       "      <td>2088-08-08 16:44:01+03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19578.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>Viber</td>\n",
       "      <td>Viber</td>\n",
       "      <td>54.0</td>\n",
       "      <td>Можем отправить заявку на подключение новых но...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>322276</td>\n",
       "      <td>2088-08-08 16:44:01+03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Обращение закрыто в связи с неактивностью поль...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   appeal_id                datetime  client_id  user_id  msg_type  \\\n",
       "0     323874  2088-08-08 16:44:02+03        NaN  19670.0         0   \n",
       "1     323870  2088-08-08 16:44:01+03        NaN  19671.0         0   \n",
       "2     323861  2088-08-08 16:44:01+03        NaN  19418.0         0   \n",
       "3     323850  2088-08-08 16:44:01+03        NaN  19578.0         0   \n",
       "4     322276  2088-08-08 16:44:01+03        NaN      NaN         2   \n",
       "\n",
       "   channel_id         channel_name channel_type  answer_time  \\\n",
       "0          73            Webdealer      Веб-чат         12.0   \n",
       "1          73            Webdealer      Веб-чат          NaN   \n",
       "2         133  Нижний Новгород B2C      Веб-чат          NaN   \n",
       "3           2                Viber        Viber         54.0   \n",
       "4           0                  NaN          NaN          NaN   \n",
       "\n",
       "                                                text  \n",
       "0                                          Привет.))  \n",
       "1  Все подписки отключены. Как это делать самосто...  \n",
       "2         Вы забыли пароль и хотите войти в кабинет?  \n",
       "3  Можем отправить заявку на подключение новых но...  \n",
       "4  Обращение закрыто в связи с неактивностью поль...  "
      ]
     },
     "execution_count": 244,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(125938, 10)\n"
     ]
    }
   ],
   "source": [
    "d1 = pd.read_csv('D:/tele2_data_hack/email_reports.csv')\n",
    "print(d1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SENDER_ID</th>\n",
       "      <th>SENDER_EMAIL</th>\n",
       "      <th>OPERATOR_ID</th>\n",
       "      <th>ACCEPTOR</th>\n",
       "      <th>SUBJECT</th>\n",
       "      <th>To.queue.time</th>\n",
       "      <th>agent.receive.time</th>\n",
       "      <th>treatment.end.time</th>\n",
       "      <th>CONTENT</th>\n",
       "      <th>ANSWERTEXT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>token_7747</td>\n",
       "      <td>xxxx@yandex.ru</td>\n",
       "      <td>token_48</td>\n",
       "      <td>spb.vopros@tele2.ru</td>\n",
       "      <td>Обратная связь</td>\n",
       "      <td>2088-06-17 02:31:32</td>\n",
       "      <td>2088-06-18 19:48:26</td>\n",
       "      <td>2088-06-18 20:24:59</td>\n",
       "      <td>Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...</td>\n",
       "      <td>\\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx!\\r\\n\\r...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>token_14226</td>\n",
       "      <td>xxxx@mail.ru</td>\n",
       "      <td>token_50</td>\n",
       "      <td>Saransk.Vopros@tele2.ru</td>\n",
       "      <td>Re[2]: Обратная связь</td>\n",
       "      <td>2088-06-17 02:31:52</td>\n",
       "      <td>2088-06-18 18:43:45</td>\n",
       "      <td>2088-06-18 18:58:43</td>\n",
       "      <td>\\r\\nвся необходимая информация отображена в ...</td>\n",
       "      <td>Здравствуйте, xxxxx xxxxx xxxxx!\\r\\nБлагодарим...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>token_23942</td>\n",
       "      <td>xxxx@mail.ru</td>\n",
       "      <td>token_3</td>\n",
       "      <td>rostov.vopros@tele2.ru</td>\n",
       "      <td>Обратная связь</td>\n",
       "      <td>2088-06-17 02:33:14</td>\n",
       "      <td>2088-06-18 18:43:45</td>\n",
       "      <td>2088-06-18 18:47:06</td>\n",
       "      <td>Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...</td>\n",
       "      <td>\\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx!\\r\\nБл...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>token_18368</td>\n",
       "      <td>xxxx@mail.ru</td>\n",
       "      <td>token_53</td>\n",
       "      <td>krasnodar.vopros@tele2.ru</td>\n",
       "      <td>Обратная связь</td>\n",
       "      <td>2088-06-17 02:33:36</td>\n",
       "      <td>2088-06-18 19:49:10</td>\n",
       "      <td>2088-06-18 19:53:57</td>\n",
       "      <td>Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...</td>\n",
       "      <td>\\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx! ¦\\r\\n...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>token_21039</td>\n",
       "      <td>xxxx@mail.ru</td>\n",
       "      <td>token_34</td>\n",
       "      <td>kostroma.vopros@tele2.ru</td>\n",
       "      <td>Обратная связь</td>\n",
       "      <td>2088-06-17 02:33:56</td>\n",
       "      <td>2088-06-18 19:49:17</td>\n",
       "      <td>2088-06-18 19:59:48</td>\n",
       "      <td>Тело письма\\r\\n\\r\\nОтметьте, в каком регионе В...</td>\n",
       "      <td>\\r\\n\\r\\nАлександр, доброго времени суток! ;)\\r...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     SENDER_ID    SENDER_EMAIL OPERATOR_ID                   ACCEPTOR  \\\n",
       "0   token_7747  xxxx@yandex.ru    token_48        spb.vopros@tele2.ru   \n",
       "1  token_14226    xxxx@mail.ru    token_50    Saransk.Vopros@tele2.ru   \n",
       "2  token_23942    xxxx@mail.ru     token_3     rostov.vopros@tele2.ru   \n",
       "3  token_18368    xxxx@mail.ru    token_53  krasnodar.vopros@tele2.ru   \n",
       "4  token_21039    xxxx@mail.ru    token_34   kostroma.vopros@tele2.ru   \n",
       "\n",
       "                 SUBJECT        To.queue.time   agent.receive.time  \\\n",
       "0         Обратная связь  2088-06-17 02:31:32  2088-06-18 19:48:26   \n",
       "1  Re[2]: Обратная связь  2088-06-17 02:31:52  2088-06-18 18:43:45   \n",
       "2         Обратная связь  2088-06-17 02:33:14  2088-06-18 18:43:45   \n",
       "3         Обратная связь  2088-06-17 02:33:36  2088-06-18 19:49:10   \n",
       "4         Обратная связь  2088-06-17 02:33:56  2088-06-18 19:49:17   \n",
       "\n",
       "    treatment.end.time                                            CONTENT  \\\n",
       "0  2088-06-18 20:24:59  Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...   \n",
       "1  2088-06-18 18:58:43    \\r\\nвся необходимая информация отображена в ...   \n",
       "2  2088-06-18 18:47:06  Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...   \n",
       "3  2088-06-18 19:53:57  Тело письма\\r\\n\\r\\nMSISDN: +xxx xxx xx xx\\r\\nО...   \n",
       "4  2088-06-18 19:59:48  Тело письма\\r\\n\\r\\nОтметьте, в каком регионе В...   \n",
       "\n",
       "                                          ANSWERTEXT  \n",
       "0  \\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx!\\r\\n\\r...  \n",
       "1  Здравствуйте, xxxxx xxxxx xxxxx!\\r\\nБлагодарим...  \n",
       "2  \\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx!\\r\\nБл...  \n",
       "3  \\r\\n\\r\\nЗдравствуйте, xxxxx xxxxx xxxxx! ¦\\r\\n...  \n",
       "4  \\r\\n\\r\\nАлександр, доброго времени суток! ;)\\r...  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d1.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "44289"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(set(d1['SENDER_ID']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### number of emails "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp = d1[['SENDER_ID','SENDER_EMAIL']].groupby('SENDER_ID').agg('count').reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([   1,    2,    3,    4,    5,    6,    7,    8,    9,   10,   11,\n",
       "          12,   13,   14,   15,   16,   17,   18,   19,   20,   21,   22,\n",
       "          23,   25,   26,   27,   29,   30,   31,   32,   34,   37,   40,\n",
       "          41,   43,   45,   48,   56,   61,   63,   69,  137,  143,  173,\n",
       "         174,  205,  293,  447,  450,  555,  583,  712,  794,  825,  878,\n",
       "        1799, 2038, 2797, 5795], dtype=int64),\n",
       " array([14977,  7122, 16235,  3880,  1199,   416,   166,    96,    34,\n",
       "           17,    18,    17,    13,    13,     4,     6,     8,     9,\n",
       "            5,     3,     3,     4,     1,     2,     4,     2,     1,\n",
       "            2,     1,     1,     1,     2,     1,     1,     1,     1,\n",
       "            1,     1,     1,     1,     1,     1,     1,     1,     1,\n",
       "            1,     1,     1,     1,     1,     1,     1,     1,     1,\n",
       "            1,     1,     1,     1,     1], dtype=int64))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.unique(temp['SENDER_EMAIL'], return_counts=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# determine fraction of each problem "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(125938, 10)\n",
      "(102898, 10)\n"
     ]
    }
   ],
   "source": [
    "print(d1.shape)\n",
    "d1 = d1[d1['CONTENT'].apply(lambda x: 'Тело письма' in str(x))].reset_index(drop=True)\n",
    "print(d1.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_topic(x):\n",
    "    ret = ''\n",
    "    try:\n",
    "        ret = x.split('Выберите тему обращения: ')[1].split('\\r')[0]\n",
    "    except:\n",
    "        pass\n",
    "    return ret\n",
    "\n",
    "def get_question(x):\n",
    "    ret = ''\n",
    "    try:\n",
    "        ret = x.split('Выберите тему обращения: ')[1].split('\\r')[1].split('\\nЗадайте ваш вопрос или опишите проблему: ')[1]\n",
    "    except:\n",
    "        pass\n",
    "    return ret\n",
    "\n",
    "\n",
    "d1['topic'] = d1['CONTENT'].apply(lambda x: get_topic(x))\n",
    "d1['question'] = d1['CONTENT'].apply(lambda x: get_question(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp = d1[['topic','CONTENT']].groupby('topic').agg('count').reset_index()\n",
    "num_q = np.sum(temp['CONTENT'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp = temp[temp['CONTENT']>1200].reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "97.50335283484615\n"
     ]
    }
   ],
   "source": [
    "temp['CONTENT_prc'] = temp['CONTENT'] / num_q * 100\n",
    "print(np.sum(temp['CONTENT_prc']))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "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>topic</th>\n",
       "      <th>CONTENT</th>\n",
       "      <th>CONTENT_prc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>прочее</td>\n",
       "      <td>24434</td>\n",
       "      <td>23.745845</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>тарифыИУслуги</td>\n",
       "      <td>18770</td>\n",
       "      <td>18.241365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>мобильныйИнтернет</td>\n",
       "      <td>15918</td>\n",
       "      <td>15.469688</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>подключенныеУслуги</td>\n",
       "      <td>9112</td>\n",
       "      <td>8.855371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>личныйКабинет</td>\n",
       "      <td>7379</td>\n",
       "      <td>7.171179</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>роумингИЗонаОбслуживания</td>\n",
       "      <td>6732</td>\n",
       "      <td>6.542401</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>поступлениеПлатежейОшибочныеПлатежи</td>\n",
       "      <td>6045</td>\n",
       "      <td>5.874750</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>детализацияСчётаИлиЗвонков</td>\n",
       "      <td>5474</td>\n",
       "      <td>5.319831</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>блокировкаИРазблокировкаSimКарты</td>\n",
       "      <td>2412</td>\n",
       "      <td>2.344069</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>подключениеКTele2</td>\n",
       "      <td>2398</td>\n",
       "      <td>2.330463</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>переносНомераВTele2</td>\n",
       "      <td>1655</td>\n",
       "      <td>1.608389</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                  topic  CONTENT  CONTENT_prc\n",
       "8                                прочее    24434    23.745845\n",
       "10                        тарифыИУслуги    18770    18.241365\n",
       "3                     мобильныйИнтернет    15918    15.469688\n",
       "6                    подключенныеУслуги     9112     8.855371\n",
       "2                         личныйКабинет     7379     7.171179\n",
       "9              роумингИЗонаОбслуживания     6732     6.542401\n",
       "7   поступлениеПлатежейОшибочныеПлатежи     6045     5.874750\n",
       "1            детализацияСчётаИлиЗвонков     5474     5.319831\n",
       "0      блокировкаИРазблокировкаSimКарты     2412     2.344069\n",
       "5                     подключениеКTele2     2398     2.330463\n",
       "4                   переносНомераВTele2     1655     1.608389"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "temp.sort_values(by='CONTENT_prc', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "temp.set_index('topic')['CONTENT'].plot(kind='pie', subplots=True, figsize=(12, 12))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Тело письма\r\n",
      "\r\n",
      "MSISDN: +xxx xxx xx xx\r\n",
      "Отметьте, в каком регионе Вы постоянно проживаете:: Москва и Московская область\r\n",
      "Как к вам обращаться: xxxxx xxxxx xxxxx\r\n",
      "Контактный номер телефона: +xxx xxx xx xx\r\n",
      "Контактный email: xxx@yahoo.com\r\n",
      "Выберите тему обращения: тарифыИУслуги\r\n",
      "Задайте ваш вопрос или опишите проблему: Здравствуйте. Как можно оплатить приложение Zvooq (премиум на месяц) с помощью банковской карты, а не с помрщью средств на балансе телефонного номера?\r\n",
      "Выражаю согласие* на обработку персональных данных в соответствии со ст.9 ФЗ №152 \"О персональных данных\".: true\r\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(d1[d1['topic']=='тарифыИУслуги']['CONTENT'].reset_index(drop=True)[18])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "# remove all xxxxxx\n",
    "import re\n",
    "\n",
    "xs_seq = []\n",
    "for n_x in np.arange(2,10):\n",
    "    xs_seq.append('x'*n_x)\n",
    "    \n",
    "def clean_xs(x):\n",
    "    for xs in xs_seq:\n",
    "        x = re.sub(xs, ' ', x)\n",
    "    return x\n",
    "\n",
    "d1['question'] = d1['question'].apply(lambda x: clean_xs(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of stop words: 158\n"
     ]
    }
   ],
   "source": [
    "from nltk.corpus import stopwords\n",
    "stopWords = set(stopwords.words('russian'))\n",
    "stopWords = list(stopWords)\n",
    "stopWords = stopWords + ['вечер', 'добрый', 'день', 'скажите', 'пожалуйста', 'спасибо', 'ли', 'почему', 'вопрос']\n",
    "stopWords.remove('не')\n",
    "stopWords = set(stopWords)\n",
    "print('Number of stop words:', len(stopWords))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer, CountVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [],
   "source": [
    "#tfidf = TfidfVectorizer(ngram_range=(1, 2), min_df=100, max_df=0.99, stop_words=stopWords, preprocessor=clean_xs)\n",
    "tfidf = CountVectorizer(ngram_range=(1, 2), min_df=100, max_df=0.99, stop_words=stopWords, preprocessor=clean_xs, binary=True)\n",
    "X_tfidf = tfidf.fit_transform(d1['question'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 231,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(102898, 2813)"
      ]
     },
     "execution_count": 231,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_tfidf.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 232,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[x for x in tfidf.vocabulary_ if 'почему' in x]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# LDA -> topics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 228,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\denis\\Anaconda3\\envs\\tf\\lib\\site-packages\\sklearn\\decomposition\\online_lda.py:536: DeprecationWarning: The default value for 'learning_method' will be changed from 'online' to 'batch' in the release 0.20. This warning was introduced in 0.18.\n",
      "  DeprecationWarning)\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-228-428e8f9df92a>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[1;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdecomposition\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mLatentDirichletAllocation\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      2\u001b[0m \u001b[0mLDA\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mLatentDirichletAllocation\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mn_components\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m50\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmax_iter\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m10\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mn_jobs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m-\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mrandom_state\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m42\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 3\u001b[1;33m \u001b[0mLDA\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX_tfidf\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m~\\Anaconda3\\envs\\tf\\lib\\site-packages\\sklearn\\decomposition\\online_lda.py\u001b[0m in \u001b[0;36mfit\u001b[1;34m(self, X, y)\u001b[0m\n\u001b[0;32m    550\u001b[0m                     \u001b[1;32mfor\u001b[0m \u001b[0midx_slice\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mgen_batches\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mn_samples\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    551\u001b[0m                         self._em_step(X[idx_slice, :], total_samples=n_samples,\n\u001b[1;32m--> 552\u001b[1;33m                                       batch_update=False, parallel=parallel)\n\u001b[0m\u001b[0;32m    553\u001b[0m                 \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    554\u001b[0m                     \u001b[1;31m# batch update\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tf\\lib\\site-packages\\sklearn\\decomposition\\online_lda.py\u001b[0m in \u001b[0;36m_em_step\u001b[1;34m(self, X, total_samples, batch_update, parallel)\u001b[0m\n\u001b[0;32m    448\u001b[0m         \u001b[1;31m# update `component_` related variables\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    449\u001b[0m         self.exp_dirichlet_component_ = np.exp(\n\u001b[1;32m--> 450\u001b[1;33m             _dirichlet_expectation_2d(self.components_))\n\u001b[0m\u001b[0;32m    451\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mn_batch_iter_\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    452\u001b[0m         \u001b[1;32mreturn\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "from sklearn.decomposition import LatentDirichletAllocation\n",
    "LDA = LatentDirichletAllocation(n_components=50, max_iter=10, n_jobs=-1, verbose=1, random_state=42)\n",
    "LDA.fit(X_tfidf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def display_topics(model, feature_names, no_top_words):\n",
    "    for topic_idx, topic in enumerate(model.components_):\n",
    "        print(\"\\nTopic %d:\" % (topic_idx))\n",
    "        print(\" \".join([feature_names[i] for i in topic.argsort()[:-no_top_words - 1:-1]]))\n",
    "\n",
    "no_top_words = 10\n",
    "display_topics(LDA, list(tfidf.vocabulary_.keys()), no_top_words)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Clustering questions "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MiniBatchKMeans(batch_size=4048, compute_labels=True, init='k-means++',\n",
       "        init_size=None, max_iter=100, max_no_improvement=10, n_clusters=64,\n",
       "        n_init=3, random_state=2018, reassignment_ratio=0.01, tol=0.0,\n",
       "        verbose=0)"
      ]
     },
     "execution_count": 233,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.cluster import MiniBatchKMeans\n",
    "\n",
    "mbcl = MiniBatchKMeans(n_clusters=64, max_iter=100, batch_size=4048, random_state=2018)\n",
    "mbcl.fit(X_tfidf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 234,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1096650.4255176256"
      ]
     },
     "execution_count": 234,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mbcl.inertia_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [],
   "source": [
    "cluster_labels = mbcl.labels_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [],
   "source": [
    "(cluster_label, number_items) = np.unique(cluster_labels, return_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "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>cluster_label</th>\n",
       "      <th>number_items</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>19367</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>52</td>\n",
       "      <td>18849</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>13488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>43</td>\n",
       "      <td>11547</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>7276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>5477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>42</td>\n",
       "      <td>3979</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>24</td>\n",
       "      <td>2725</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>20</td>\n",
       "      <td>2271</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>61</td>\n",
       "      <td>2233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>49</td>\n",
       "      <td>2070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>46</td>\n",
       "      <td>1898</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>1477</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>38</td>\n",
       "      <td>1408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>1344</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>33</td>\n",
       "      <td>1276</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>1165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>62</td>\n",
       "      <td>905</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>31</td>\n",
       "      <td>778</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>47</td>\n",
       "      <td>730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>59</th>\n",
       "      <td>59</td>\n",
       "      <td>683</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>350</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>45</td>\n",
       "      <td>198</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>41</td>\n",
       "      <td>148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>144</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>48</td>\n",
       "      <td>123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>54</th>\n",
       "      <td>54</td>\n",
       "      <td>56</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>50</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>15</td>\n",
       "      <td>34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>58</th>\n",
       "      <td>58</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>27</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>53</th>\n",
       "      <td>53</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>28</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>37</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>63</th>\n",
       "      <td>63</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>21</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>26</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>50</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>60</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>57</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>56</th>\n",
       "      <td>56</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>55</th>\n",
       "      <td>55</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>39</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>36</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>51</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>23</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>25</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>29</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>30</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>44</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>35</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>32</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>64 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    cluster_label  number_items\n",
       "0               0         19367\n",
       "52             52         18849\n",
       "13             13         13488\n",
       "43             43         11547\n",
       "2               2          7276\n",
       "12             12          5477\n",
       "42             42          3979\n",
       "24             24          2725\n",
       "20             20          2271\n",
       "61             61          2233\n",
       "49             49          2070\n",
       "46             46          1898\n",
       "7               7          1477\n",
       "38             38          1408\n",
       "22             22          1344\n",
       "33             33          1276\n",
       "11             11          1165\n",
       "62             62           905\n",
       "31             31           778\n",
       "47             47           730\n",
       "8               8           693\n",
       "59             59           683\n",
       "16             16           350\n",
       "45             45           198\n",
       "41             41           148\n",
       "9               9           144\n",
       "48             48           123\n",
       "54             54            56\n",
       "3               3            50\n",
       "15             15            34\n",
       "..            ...           ...\n",
       "58             58            11\n",
       "27             27             9\n",
       "53             53             6\n",
       "28             28             5\n",
       "37             37             4\n",
       "63             63             3\n",
       "21             21             3\n",
       "14             14             2\n",
       "26             26             2\n",
       "50             50             2\n",
       "18             18             1\n",
       "5               5             1\n",
       "6               6             1\n",
       "60             60             1\n",
       "10             10             1\n",
       "57             57             1\n",
       "56             56             1\n",
       "55             55             1\n",
       "17             17             1\n",
       "39             39             1\n",
       "36             36             1\n",
       "51             51             1\n",
       "23             23             1\n",
       "25             25             1\n",
       "29             29             1\n",
       "30             30             1\n",
       "44             44             1\n",
       "1               1             1\n",
       "35             35             1\n",
       "32             32             1\n",
       "\n",
       "[64 rows x 2 columns]"
      ]
     },
     "execution_count": 237,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame({'cluster_label': cluster_label, 'number_items': number_items}).sort_values(by='number_items', ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster size: 7276 \n",
      "\n",
      "Здравствуйте.у меня тариф \"беспредельно черный\". Сегодня списалась абонентская плата 340р, а остатки СМС с прошлого месяца не перенеслись почему-то. Деньги на телефон я положил заранее ещё днём  x  x    D_\n",
      "---\n",
      "Хочу поменять тариф\n",
      "---\n",
      "Почему мой тариф изменился , но мне об этом не сообщили ? И почему он изменился?\n",
      "---\n",
      "Здравствуйте! Сегодня положил 240  рублей на счёт и после настройки тарифа мой онлайн (15гб интернета и 300 мин на звонки) активировал тариф . В итоге средства списаны и тариф мне не доступен . \n",
      "---\n",
      "Как сменить тариф?\n",
      "---\n",
      "Извините пожалуйста .Я только что разбирался в программе \"мой теле 2\" и хотел не много уменьшит тариф,но как понял что хотели снять с меня деньги и их просто не хватило и теперь я получается тупо потратил 350 рублей за 2 дня пользования .Не могли бы вы отправить пожалуйста и обедаю что такого больше не повторится \n",
      "---\n",
      "Проблема , полчаса лазил так и не нашёл!  С какого времени и по какое  действует мой тариф.   Например с 1 июня по 31 июня   x\n",
      "---\n",
      "Добрый день! Выезжаю в выходные дни на дачу, которая находится в д.   x   x   x области. На самом деле 30 минут от   x   x   x. Ребята, что там со связью?! Сигнал настолько мизерный, что в одной комнате через два шага связь обрывается и нет сигнала??!!! Да что это такое?! Просто безобразие. В прошлом году такой гадости не было. Сколько ещё можно мучится?! А ведь за эти дни нахождения на даче тариф не приостанавливается, а берётся за полноценную связь, которой практически нет. Сделайте что-нибудь!!!\n",
      "---\n",
      "Можно узнать,есть ли у вас  тариф,\"Вместе дешевле\"?Если есть,то как на него перейти?\n",
      "---\n",
      "Здравствуйте.При переходе на новый тариф \"Мой онлайн\" оплата за него списалась дважды. Деньги за повторное списание прошу вернуть на счёт телефона.\n",
      "---\n",
      "14 июня   x подключён новый тариф, 6 Гб, 500 мин разговора на других операторов, 50 смс, с 14 июня   x г. согласно детализации по 16 июня   x г. звонков на телефоны других операторов не осуществлялось, объясните куда списалось 108 мин разговора, это простите больше одного часа, жду подробных объяснений, просьба наказать мошенников!!! \n",
      "---\n",
      " Подключил тариф с 15 Гб интернета, при списание средств на следующие месяц на тариф зачислили 8 гб\n",
      "---\n",
      "Подключил тариф \"Мой разговор\", где соц. сеть \"В контакте\"- без лимита трафика. Посмотрел в \"VK\"фильм, за что сняли 1Гб . Чё за беспредел, разберитесь.\n",
      "---\n",
      "Доброго времени суток! Я пополнила баланс и хотела перейти на другой тариф, и случайно нажала на смену прямо сейчас, все заблокировалось. Получается у меня 2 раза списали абонентскую плату. Отмените пожалуйста переход или списание первой абонентской платы\n",
      "---\n",
      "Где в новой версии кнопка тариф\\менить тариф?\n",
      "---\n",
      "Предстоит поездка в Крым. У меня тариф \"Оранжевый\", Алтайский край, использую исключительно для звонков на телефоны любых операторов. Какую опцию подключить для ИИ средств на счёту?\n",
      "---\n",
      "Можете сделать так что бы я подключил тариф сегодня?\n",
      "---\n",
      "Почему мне поменяли тариф, без моего согласия!!!!???\n",
      "---\n",
      "Здравствуйте. У меня тариф \"Мой Теле2\" за 399р. Звонки/смс/интернет по тарифу действуют только по России? В   x   x   x какие условия использования тарифа? \n",
      "---\n",
      "Добрый день, вопрос в том, что ваш сотрудник ежедневно, по несколько раз надоедает моей маме звонками с навязчивыми предложениями сменить тариф! Тариф мою маму полностью устраивает, а вот назойливость вашего сотрудника уже выше сил! все это длится уже неделю примерно! Телефон мамы:+ x  x    , вашего назойливого сотрудника+ x  x    .\n",
      "---\n"
     ]
    }
   ],
   "source": [
    "check = np.where(cluster_labels==2)[0]\n",
    "print('Cluster size:', len(check), '\\n')\n",
    "for j in check[0:20]:  \n",
    "    print(d1['question'][j])\n",
    "    print('---')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp = d1[['appeal_id','client_id','text']].groupby(['appeal_id','client_id'])['text'].apply(list)\n",
    "temp = pd.DataFrame(temp).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp['text'] = temp['text'].apply(lambda x: [str(x) for x in x])\n",
    "temp['text'] = temp['text'].apply(lambda x: x[0:2])\n",
    "temp['text'] = temp['text'].apply(lambda x: ' '.join(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "temp.iloc[92]['text']"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}