customers_churn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "toc": true
   },
   "source": [
    "<h1>Содержание<span class=\"tocSkip\"></span></h1>\n",
    "<div class=\"toc\"><ul class=\"toc-item\"><li><span><a href=\"#Подготовка-данных\" data-toc-modified-id=\"Подготовка-данных-1\"><span class=\"toc-item-num\">1&nbsp;&nbsp;</span>Подготовка данных</a></span></li><li><span><a href=\"#Исследование-задачи\" data-toc-modified-id=\"Исследование-задачи-2\"><span class=\"toc-item-num\">2&nbsp;&nbsp;</span>Исследование задачи</a></span></li><li><span><a href=\"#Борьба-с-дисбалансом\" data-toc-modified-id=\"Борьба-с-дисбалансом-3\"><span class=\"toc-item-num\">3&nbsp;&nbsp;</span>Борьба с дисбалансом</a></span></li><li><span><a href=\"#Тестирование-модели\" data-toc-modified-id=\"Тестирование-модели-4\"><span class=\"toc-item-num\">4&nbsp;&nbsp;</span>Тестирование модели</a></span></li><li><span><a href=\"#Чек-лист-готовности-проекта\" data-toc-modified-id=\"Чек-лист-готовности-проекта-5\"><span class=\"toc-item-num\">5&nbsp;&nbsp;</span>Чек-лист готовности проекта</a></span></li></ul></div>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Отток клиентов"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Цель:** сохранять текущих клиентов.\n",
    "\n",
    "**Задача:** \n",
    "- спрогнозировать, уйдёт клиент из банка в ближайшее время или нет. \n",
    "- построить модель с предельно большим значением *F1*-меры. Нужно довести метрику до 0.59. \n",
    "- Дополнительно измерить *AUC-ROC*, сравнить её значение с *F1*-мерой.\n",
    "\n",
    "**Данные:* предоставлены исторические данные о поведении клиентов и расторжении договоров с банком.\n",
    "\n",
    "Источник данных: [https://www.kaggle.com/barelydedicated/bank-customer-churn-modeling](https://www.kaggle.com/barelydedicated/bank-customer-churn-modeling)\n",
    "\n",
    "*Признаки*\n",
    "- RowNumber — индекс строки в данных\n",
    "- CustomerId — уникальный идентификатор клиента\n",
    "- Surname — фамилия\n",
    "- CreditScore — кредитный рейтинг\n",
    "- Geography — страна проживания\n",
    "- Gender — пол\n",
    "- Age — возраст\n",
    "- Tenure — сколько лет человек является клиентом банка\n",
    "- Balance — баланс на счёте\n",
    "- NumOfProducts — количество продуктов банка, используемых клиентом\n",
    "- HasCrCard — наличие кредитной карты\n",
    "- IsActiveMember — активность клиента\n",
    "- EstimatedSalary — предполагаемая зарплата\n",
    "\n",
    "*Целевой признак*\n",
    "- Exited — факт ухода клиента"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Подготовка данных"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Импортируем библиотеки и ознакомимся с данными."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.metrics import roc_curve, auc, mean_squared_error, f1_score, roc_auc_score\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.utils import shuffle\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "\n",
    "sns.set(style=\"darkgrid\")\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame = pd.read_csv(\"/datasets/Churn.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10000 entries, 0 to 9999\n",
      "Data columns (total 14 columns):\n",
      "RowNumber          10000 non-null int64\n",
      "CustomerId         10000 non-null int64\n",
      "Surname            10000 non-null object\n",
      "CreditScore        10000 non-null int64\n",
      "Geography          10000 non-null object\n",
      "Gender             10000 non-null object\n",
      "Age                10000 non-null int64\n",
      "Tenure             9091 non-null float64\n",
      "Balance            10000 non-null float64\n",
      "NumOfProducts      10000 non-null int64\n",
      "HasCrCard          10000 non-null int64\n",
      "IsActiveMember     10000 non-null int64\n",
      "EstimatedSalary    10000 non-null float64\n",
      "Exited             10000 non-null int64\n",
      "dtypes: float64(3), int64(8), object(3)\n",
      "memory usage: 1.1+ MB\n"
     ]
    }
   ],
   "source": [
    "frame.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Столбец `Tenure` содержит около 10% пропущенных значений, их нужно будет заполнить.\n",
    "<hr>\n",
    "Изучим значения статистик и проверим на наличие дубликатов."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>RowNumber</th>\n",
       "      <th>CustomerId</th>\n",
       "      <th>Surname</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>Geography</th>\n",
       "      <th>Gender</th>\n",
       "      <th>Age</th>\n",
       "      <th>Tenure</th>\n",
       "      <th>Balance</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Exited</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>15634602</td>\n",
       "      <td>Hargrave</td>\n",
       "      <td>619</td>\n",
       "      <td>France</td>\n",
       "      <td>Female</td>\n",
       "      <td>42</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>101348.88</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>15647311</td>\n",
       "      <td>Hill</td>\n",
       "      <td>608</td>\n",
       "      <td>Spain</td>\n",
       "      <td>Female</td>\n",
       "      <td>41</td>\n",
       "      <td>1.0</td>\n",
       "      <td>83807.86</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>112542.58</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>15619304</td>\n",
       "      <td>Onio</td>\n",
       "      <td>502</td>\n",
       "      <td>France</td>\n",
       "      <td>Female</td>\n",
       "      <td>42</td>\n",
       "      <td>8.0</td>\n",
       "      <td>159660.80</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>113931.57</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>15701354</td>\n",
       "      <td>Boni</td>\n",
       "      <td>699</td>\n",
       "      <td>France</td>\n",
       "      <td>Female</td>\n",
       "      <td>39</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>93826.63</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>15737888</td>\n",
       "      <td>Mitchell</td>\n",
       "      <td>850</td>\n",
       "      <td>Spain</td>\n",
       "      <td>Female</td>\n",
       "      <td>43</td>\n",
       "      <td>2.0</td>\n",
       "      <td>125510.82</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>79084.10</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   RowNumber  CustomerId   Surname  CreditScore Geography  Gender  Age  \\\n",
       "0          1    15634602  Hargrave          619    France  Female   42   \n",
       "1          2    15647311      Hill          608     Spain  Female   41   \n",
       "2          3    15619304      Onio          502    France  Female   42   \n",
       "3          4    15701354      Boni          699    France  Female   39   \n",
       "4          5    15737888  Mitchell          850     Spain  Female   43   \n",
       "\n",
       "   Tenure    Balance  NumOfProducts  HasCrCard  IsActiveMember  \\\n",
       "0     2.0       0.00              1          1               1   \n",
       "1     1.0   83807.86              1          0               1   \n",
       "2     8.0  159660.80              3          1               0   \n",
       "3     1.0       0.00              2          0               0   \n",
       "4     2.0  125510.82              1          1               1   \n",
       "\n",
       "   EstimatedSalary  Exited  \n",
       "0        101348.88       1  \n",
       "1        112542.58       0  \n",
       "2        113931.57       1  \n",
       "3         93826.63       0  \n",
       "4         79084.10       0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>RowNumber</th>\n",
       "      <th>CustomerId</th>\n",
       "      <th>CreditScore</th>\n",
       "      <th>Age</th>\n",
       "      <th>Tenure</th>\n",
       "      <th>Balance</th>\n",
       "      <th>NumOfProducts</th>\n",
       "      <th>HasCrCard</th>\n",
       "      <th>IsActiveMember</th>\n",
       "      <th>EstimatedSalary</th>\n",
       "      <th>Exited</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>count</td>\n",
       "      <td>10000.00000</td>\n",
       "      <td>1.000000e+04</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>9091.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>10000.00000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "      <td>10000.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>mean</td>\n",
       "      <td>5000.50000</td>\n",
       "      <td>1.569094e+07</td>\n",
       "      <td>650.528800</td>\n",
       "      <td>38.921800</td>\n",
       "      <td>4.997690</td>\n",
       "      <td>76485.889288</td>\n",
       "      <td>1.530200</td>\n",
       "      <td>0.70550</td>\n",
       "      <td>0.515100</td>\n",
       "      <td>100090.239881</td>\n",
       "      <td>0.203700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>std</td>\n",
       "      <td>2886.89568</td>\n",
       "      <td>7.193619e+04</td>\n",
       "      <td>96.653299</td>\n",
       "      <td>10.487806</td>\n",
       "      <td>2.894723</td>\n",
       "      <td>62397.405202</td>\n",
       "      <td>0.581654</td>\n",
       "      <td>0.45584</td>\n",
       "      <td>0.499797</td>\n",
       "      <td>57510.492818</td>\n",
       "      <td>0.402769</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>min</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.556570e+07</td>\n",
       "      <td>350.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>11.580000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25%</td>\n",
       "      <td>2500.75000</td>\n",
       "      <td>1.562853e+07</td>\n",
       "      <td>584.000000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>51002.110000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50%</td>\n",
       "      <td>5000.50000</td>\n",
       "      <td>1.569074e+07</td>\n",
       "      <td>652.000000</td>\n",
       "      <td>37.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>97198.540000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>100193.915000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75%</td>\n",
       "      <td>7500.25000</td>\n",
       "      <td>1.575323e+07</td>\n",
       "      <td>718.000000</td>\n",
       "      <td>44.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>127644.240000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>149388.247500</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>max</td>\n",
       "      <td>10000.00000</td>\n",
       "      <td>1.581569e+07</td>\n",
       "      <td>850.000000</td>\n",
       "      <td>92.000000</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>250898.090000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>199992.480000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         RowNumber    CustomerId   CreditScore           Age       Tenure  \\\n",
       "count  10000.00000  1.000000e+04  10000.000000  10000.000000  9091.000000   \n",
       "mean    5000.50000  1.569094e+07    650.528800     38.921800     4.997690   \n",
       "std     2886.89568  7.193619e+04     96.653299     10.487806     2.894723   \n",
       "min        1.00000  1.556570e+07    350.000000     18.000000     0.000000   \n",
       "25%     2500.75000  1.562853e+07    584.000000     32.000000     2.000000   \n",
       "50%     5000.50000  1.569074e+07    652.000000     37.000000     5.000000   \n",
       "75%     7500.25000  1.575323e+07    718.000000     44.000000     7.000000   \n",
       "max    10000.00000  1.581569e+07    850.000000     92.000000    10.000000   \n",
       "\n",
       "             Balance  NumOfProducts    HasCrCard  IsActiveMember  \\\n",
       "count   10000.000000   10000.000000  10000.00000    10000.000000   \n",
       "mean    76485.889288       1.530200      0.70550        0.515100   \n",
       "std     62397.405202       0.581654      0.45584        0.499797   \n",
       "min         0.000000       1.000000      0.00000        0.000000   \n",
       "25%         0.000000       1.000000      0.00000        0.000000   \n",
       "50%     97198.540000       1.000000      1.00000        1.000000   \n",
       "75%    127644.240000       2.000000      1.00000        1.000000   \n",
       "max    250898.090000       4.000000      1.00000        1.000000   \n",
       "\n",
       "       EstimatedSalary        Exited  \n",
       "count     10000.000000  10000.000000  \n",
       "mean     100090.239881      0.203700  \n",
       "std       57510.492818      0.402769  \n",
       "min          11.580000      0.000000  \n",
       "25%       51002.110000      0.000000  \n",
       "50%      100193.915000      0.000000  \n",
       "75%      149388.247500      0.000000  \n",
       "max      199992.480000      1.000000  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Четверть клиентов имеет нулевой баланс, тоже самое касается признака активности. Возможно здесь есть связь."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Количесвто дубликатов:'"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "display(\"Количесвто дубликатов:\",frame.duplicated().sum())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<hr>\n",
    "Фамилия, айди и номер строки - лишние столбцы, удалим их."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame = frame.drop(['Surname', 'RowNumber', 'CustomerId'], axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Переименуем столбцы и приведем к одному регистру."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['creditscore', 'geography', 'gender', 'age', 'tenure', 'balance',\n",
       "       'numofproducts', 'hascrcard', 'isactivemember', 'estimatedsalary',\n",
       "       'exited'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.columns = frame.columns.str.lower()\n",
    "frame.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame.rename({'creditscore': 'credit_score',\n",
    "             'geography': 'country',\n",
    "             'isactivemember': 'is_active',\n",
    "             'numofproducts': 'num_of_products',\n",
    "             'hascrcard': 'has_cr_card',\n",
    "             'estimatedsalary': 'salary',\n",
    "             'exited': 'churn'}, axis=1, inplace=True)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Обработка пропусков**\n",
    "\n",
    "В столбце `tenure` есть пропуски. Поищем аномалии среди этих клиентов."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>credit_score</th>\n",
       "      <th>age</th>\n",
       "      <th>tenure</th>\n",
       "      <th>balance</th>\n",
       "      <th>num_of_products</th>\n",
       "      <th>has_cr_card</th>\n",
       "      <th>is_active</th>\n",
       "      <th>salary</th>\n",
       "      <th>churn</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>count</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "      <td>909.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>mean</td>\n",
       "      <td>648.451045</td>\n",
       "      <td>38.647965</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76117.341474</td>\n",
       "      <td>1.530253</td>\n",
       "      <td>0.710671</td>\n",
       "      <td>0.510451</td>\n",
       "      <td>99180.389373</td>\n",
       "      <td>0.201320</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>std</td>\n",
       "      <td>99.079381</td>\n",
       "      <td>9.785438</td>\n",
       "      <td>NaN</td>\n",
       "      <td>63105.690715</td>\n",
       "      <td>0.588452</td>\n",
       "      <td>0.453701</td>\n",
       "      <td>0.500166</td>\n",
       "      <td>56378.063765</td>\n",
       "      <td>0.401207</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>min</td>\n",
       "      <td>359.000000</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>106.670000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25%</td>\n",
       "      <td>580.000000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>49872.330000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>50%</td>\n",
       "      <td>647.000000</td>\n",
       "      <td>37.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96674.550000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>99444.020000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>75%</td>\n",
       "      <td>718.000000</td>\n",
       "      <td>43.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>128554.980000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>145759.700000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>max</td>\n",
       "      <td>850.000000</td>\n",
       "      <td>92.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>206663.750000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>199390.450000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       credit_score         age  tenure        balance  num_of_products  \\\n",
       "count    909.000000  909.000000     0.0     909.000000       909.000000   \n",
       "mean     648.451045   38.647965     NaN   76117.341474         1.530253   \n",
       "std       99.079381    9.785438     NaN   63105.690715         0.588452   \n",
       "min      359.000000   18.000000     NaN       0.000000         1.000000   \n",
       "25%      580.000000   32.000000     NaN       0.000000         1.000000   \n",
       "50%      647.000000   37.000000     NaN   96674.550000         1.000000   \n",
       "75%      718.000000   43.000000     NaN  128554.980000         2.000000   \n",
       "max      850.000000   92.000000     NaN  206663.750000         4.000000   \n",
       "\n",
       "       has_cr_card   is_active         salary       churn  \n",
       "count   909.000000  909.000000     909.000000  909.000000  \n",
       "mean      0.710671    0.510451   99180.389373    0.201320  \n",
       "std       0.453701    0.500166   56378.063765    0.401207  \n",
       "min       0.000000    0.000000     106.670000    0.000000  \n",
       "25%       0.000000    0.000000   49872.330000    0.000000  \n",
       "50%       1.000000    1.000000   99444.020000    0.000000  \n",
       "75%       1.000000    1.000000  145759.700000    0.000000  \n",
       "max       1.000000    1.000000  199390.450000    1.000000  "
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.query('tenure.isna()').describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Очевидных аномалий я не нашёл, поделим выборку на пять возрастных групп и заполним медианным значением внутри каждой из групп все пропуски."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(17.999, 31.0]    2372\n",
       "(31.0, 35.0]      1781\n",
       "(35.0, 40.0]      2266\n",
       "(40.0, 46.0]      1696\n",
       "(46.0, 92.0]      1885\n",
       "Name: group_age, dtype: int64"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "frame['group_age'] = pd.qcut(frame['age'], 5)\n",
    "display(frame['group_age'].value_counts().sort_index())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 9091 entries, 0 to 9998\n",
      "Data columns (total 12 columns):\n",
      "credit_score       9091 non-null int64\n",
      "country            9091 non-null object\n",
      "gender             9091 non-null object\n",
      "age                9091 non-null int64\n",
      "tenure             9091 non-null float64\n",
      "balance            9091 non-null float64\n",
      "num_of_products    9091 non-null int64\n",
      "has_cr_card        9091 non-null int64\n",
      "is_active          9091 non-null int64\n",
      "salary             9091 non-null float64\n",
      "churn              9091 non-null int64\n",
      "group_age          9091 non-null category\n",
      "dtypes: category(1), float64(3), int64(6), object(2)\n",
      "memory usage: 861.4+ KB\n"
     ]
    }
   ],
   "source": [
    "frame_del_drop = frame.copy()\n",
    "frame_del_drop = frame_del_drop.dropna()\n",
    "frame_del_drop.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame['tenure'] = frame.groupby(\n",
    "        ['country',\n",
    "         'gender',\n",
    "         'group_age']\n",
    ")['tenure'].transform(lambda x: x.fillna(x.median()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 10000 entries, 0 to 9999\n",
      "Data columns (total 11 columns):\n",
      "credit_score       10000 non-null int64\n",
      "country            10000 non-null object\n",
      "gender             10000 non-null object\n",
      "age                10000 non-null int64\n",
      "tenure             10000 non-null int64\n",
      "balance            10000 non-null float64\n",
      "num_of_products    10000 non-null int64\n",
      "has_cr_card        10000 non-null int64\n",
      "is_active          10000 non-null int64\n",
      "salary             10000 non-null float64\n",
      "churn              10000 non-null int64\n",
      "dtypes: float64(2), int64(7), object(2)\n",
      "memory usage: 859.5+ KB\n"
     ]
    }
   ],
   "source": [
    "frame['tenure'] = frame['tenure'].astype('int')\n",
    "frame = frame.drop('group_age', axis=1)\n",
    "frame.info()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Поправили типы данных, удалили ненужный стобец с возрастной группой, проверили названия, пропуски - всё ок, двигаемся дальше.\n",
    "<hr>\n",
    "Построим графики распределения по кредитному скорингу, возрасту, остатку на счетах, заработной плате и изучим зависимости."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(2, 2, figsize=(10, 10))\n",
    "axes[0, 0].hist(frame['credit_score'], bins=32)\n",
    "axes[0, 0].set_title(\"credit_score\")\n",
    "axes[0, 1].hist(frame['age'], bins=24)\n",
    "axes[0, 1].set_title(\"age\")\n",
    "axes[1, 0].hist(frame['balance'], bins=30)\n",
    "axes[1, 0].set_title(\"balance\")\n",
    "axes[1, 1].hist(frame['salary'], bins=30)\n",
    "axes[1, 1].set_title(\"salary\");"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Распределение таких характеристик близки к нормальному. При этом мы можем видеть, что распределение зарплат довольно однородное. Это говорит о том, что, вероятно, выборка стратифицирована.\n",
    "\n",
    "Есть подозрительный выброс в районе нуля на графике балансов. Удалять мы его не будем, нет причин полагать, что это ошибка, скорее всего этот критерий будет очень важен для модели. Интересно, что нет клиентов с нулевым балансом из Германии. Во франции таких вдвое больше, но это обсуловлено тем, что и самих клиентов из Франции практически вдвое больше. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "country  is_active\n",
       "France   0            1158\n",
       "         1            1260\n",
       "Spain    0             586\n",
       "         1             613\n",
       "Name: churn, dtype: int64"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame[frame['balance'] == 0].groupby(['country', 'is_active'])['churn'].agg('count')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fa3c82c0150>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "(frame[frame['churn'] == 1].groupby(['country'])['churn'].count()).plot(kind='bar', rot=45)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Больше всего отток клиентов из Франции и Германии, Испания имеет вдвое меньший показатель."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Requirement already up-to-date: seaborn in /home/jovyan/.local/lib/python3.7/site-packages (0.11.2)\n",
      "Requirement already satisfied, skipping upgrade: numpy>=1.15 in /opt/conda/lib/python3.7/site-packages (from seaborn) (1.19.5)\n",
      "Requirement already satisfied, skipping upgrade: matplotlib>=2.2 in /opt/conda/lib/python3.7/site-packages (from seaborn) (3.1.0)\n",
      "Requirement already satisfied, skipping upgrade: scipy>=1.0 in /opt/conda/lib/python3.7/site-packages (from seaborn) (1.4.1)\n",
      "Requirement already satisfied, skipping upgrade: pandas>=0.23 in /opt/conda/lib/python3.7/site-packages (from seaborn) (0.25.1)\n",
      "Requirement already satisfied, skipping upgrade: python-dateutil>=2.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn) (2.8.1)\n",
      "Requirement already satisfied, skipping upgrade: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn) (2.4.7)\n",
      "Requirement already satisfied, skipping upgrade: kiwisolver>=1.0.1 in /opt/conda/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn) (1.3.1)\n",
      "Requirement already satisfied, skipping upgrade: cycler>=0.10 in /opt/conda/lib/python3.7/site-packages (from matplotlib>=2.2->seaborn) (0.10.0)\n",
      "Requirement already satisfied, skipping upgrade: pytz>=2017.2 in /opt/conda/lib/python3.7/site-packages (from pandas>=0.23->seaborn) (2021.1)\n",
      "Requirement already satisfied, skipping upgrade: six>=1.5 in /opt/conda/lib/python3.7/site-packages (from python-dateutil>=2.1->matplotlib>=2.2->seaborn) (1.15.0)\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "pip install -U seaborn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Баланс на момент оттока')"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAEcCAYAAAAGD4lRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nOzdeXhU1fnA8e+dNdtkDwGysHMAQRBEVHBHQdyLWtuq1dZWbWtta1u7ufRnbcW22qp1q1hcwLovdQNxQQEFZVfgEJaQkJAQyL5OZub+/phJDJA9k5kk836eJ08y99x77ntmJvPOvefccw3TNBFCCCG6wxLuAIQQQvRfkkSEEEJ0myQRIYQQ3SZJRAghRLdJEhFCCNFtkkSEEEJ0myQRIYQQ3WYLdwCi/1FK5QLpgLfF4tVa63PCEpAQImwkiYjuukBrvTzcQQghwkuSiAg6pdR5wJ+AUUAFsFBrfecR65hALWACTmCB1voPSqlRwL+ByYGypcCPtdblge1ygeuaEphS6jrgSq316YHHxwD/AKYBjcA/tdZ/biXGRcA+rfUfAo9HAzlaayPw+Frg10AmUBKI77E22nsN8B/gF1rr+wPL5gFvAXe32McPgFuBZGAlcIPWurDF87FZaz058NgK5AGm1jozsGwo8CBwKlAN3K+1fiBQdicwWmt9ZeCxLdD+EcAMYGEg3OjAcg+A1jqulfYkBPZzbuA1+jfwZ2AwsCOwmgMwgIbA43OB/+H/TLEBdqAuUHa91nqxUupC4C9ABrARuFFrvS2wz1wCr6tS6gTgdeAirfXawON/AuMDdb4ceK7drb0eIrSkT0T0hhrgaiAROA+4USl1cVOhUqrpfXds4ENscYttDfwfNEPxf2hkAXd2ZqdKKRewHHg3sP1o4P1utuEAcD4QD1wL3K+UmtrO+juB77Z4fB2wrUVsZ+Jv1+XAEGAv8N8j6nAopaYH/j4PfwJu2t6C/0N6E/4P4bOAnyml5nTUEK3181rruMBz/QnwkxaPW/MgkACMBE7D/1peq7UubLHdn4HmerXWn2itEwNlNwCftihbrJQaCzwH/AxIA94G/qeUcrTcsVJqPPAq/i8GawOLvcDPgVTgpEDbf9RRu0VoyJGI6K7XlFIe/N9odwF/0Vq/BqC1/qjFepuVUs/h/zB6LbCs6YPjqG+SWuud+D+QAUqUUvcBd3QypvOBIq313wOP64E1ndz2yDjeavFwhVJqGXAKsL6NTYqBg0qpk4DdwDBgbYvy7wBPaq3XAyilfguUKaWGa61zA+ssxJ98Pg/8Xoj/wxNgOpCmtf6/wOPdSql/A1fgP1oLisAR0BXAFK11FVCllPo7cBVfH810xzeBt7TW7wX28zfgZuBk4KPAOsPwH9H9QWvdnPy11uta1JOrlHoM//vpHz2IRwSJJBHRXRcHTj3YgEuAF5RSx2itc5RSM4B7gIn4E4YTeLHFtsmB32VHVqqUSsd/6uIUwIX/aPnI9ZoSGIH6mz6ss/AntM76pVLqJ4G/DzsqV0qdiz95jQ2UxQBbOqjvCfwf/hp4GjiuRdlQWiQgrXW1UuoQ/qOK3MDiN4HXA6fWhgAtPzyHAUOVUuUtllnxH1k0uVwpdX4HMXYkFf+pqL0tlu0NxNkTQ1vWqbX2KaXyj6j3QfxHWmfjTyYABI5i7gOOx/862Dj8uRFhJKezRI9orT1a6xfxn3o5NrB4CfAGkKW1TgAexX+aqslYYL/WurqVKv+Mvy9kktY6HrjyiG3Bn8AStdaJwE9bLM/Hfwqms/7Wop7mU1VKKSf+8+5/A9ID5W+3EseR3gFm4j+t9cwRZYX4E0HTPmKBFKCgxToe/KdyXgIWHbF9PrCnKd7Aj0trPa/FOi+0aE9qB7G25SD+PpNhLZZlHxFndxzZfgN/0m9Z71/xn6o6IdB/0uQRYDswJvCe+B0dvxYiRCSJiB5RShlKqYuAJGBrYLELKNVa1wc6Rb/dYv1U4Dd8fWrrSC78ncYVSqkM4FddCOdNYIhS6mdKKadSyhU4KuqqpqOnEsATOCrpcPiy1toLLACe1VqXHlH8HHCtUmpKIEn9GVjT4lRWk8fx96UsPmL5Wvynlm5VSkUrpaxKqYkt+lCCItCGF4C7A8/fMOAXwLM9rPoF4Dyl1FlKKTtwC/5O+dUt1vlEa10LfB94WCmVGFjuAiqBaqXUOODGHsYigkiSiOiu/ymlqvH/c98DfL9ppA3+Ts//U0pVAbfj/wBp8l/8/Qe/aaPeP+I/KqjAP7rplc4GFDiHfzZwAVAE5ABndHb7I+r5aSDuMvxJ8I1ObvsfrfVfWlm+HLgN/xHOfvwj165oZb3dWutvNY1Ga7Hci7/PZwqwB/8RwxP4O8CD7Sb8gyN24x9FtgR4sicVaq01/qPKB/HHfgH+YeKt9YutwP8l4/7Aol/ifw2q8I8Ue74nsYjgMuSmVEIIIbpLjkSEEEJ0myQRIYQQ3SZJRAghRLdJEhFCCNFtIbvYUCn1Gv55fHz4h3DepLXeGJgzpz7wA3Cr1nppYJsTgcfwz/eTi38qhAOd3KUT/1W++zl8tlkhhBBts+K/2PVzvp4brU2hvGL9u1rrCoDAdQVP8vUFXpdqrb9suXJgrqBngWu01iuVUn/AP5T0e53c33QOv5pXCCFE552Cf4h3u0KWRJoSSEAC/iOS9kwD6rXWTY14FP/RSGeTyH6AsrIafL6uD2NOSYnj0KHWLqgeuCKtzZHWXpA2R4KettdiMUhKioXAZ2hHQjp3llLqCfxX/hrA3BZFiwPTIKwEfhe40Cqbw+faOaiUsiilklu5Grg1XgCfz+xWEmnaNtJEWpsjrb0gbY4EQWpvp7oBQppEtNbXASilrsI/T8484BStdX5gKoh/AA/hv7I1KFJS2prtumNpaa5ghdFvRFqbI629IG2OBKFsb1hm8dVaP6OUelwplaK1zg8sa1BKPczX00vkcfiEbamAr5NHIc0OHaruVlZOS3NRUlLV5e36s0hrc6S1F6TNkaCn7bVYjC59+Q5JElFKxQFJTQlDKXUBUArUK6UStNYVgdNZV+C/4xn4p3qOVkrNCvSL3MDh04kLIQaouroaqqvL8Xo9Ha/cgQMHLPh8HXXBDhyda6+BwxFFUlIahtGzCZFDdSQSC7wYmP7aiz+BXACkAy8HboRjxT8L7I+g+X4DVwGPKaWiCAzxDVG8QogwqauroaqqjMTENOx2R88/5GwWPJ7ISSKdaa9p+igvP0h1dQUuV2K763a4vx5t3Ula62LgxDaKj2tjOVrr1cCkXglKCNEnVVeXk5iYhsPhDHcoA5ZhWHC5kigtLe5xEpEr1oUQfYrX68Fud3S8ougRq9WGz9fz67AliQgh+pyensISHQvWcyz3WG+D1+vr8TA5d6OXivLaIEUkhBB9jySRNlitFv79yqYe1fGDb0wOUjRCiP7q0ksv4NZb/8D06d25U3PfJ6ezhBAijDyeng9jDidJIkKIiFJcXMTvfvcrzj9/NvPmncV99y3A5/OxaNETzJ9/PueffzZ33XU71dX++afWr/+CSy6Zd1gdl156AZ9/vgaAhQsf47bbfsNdd93O2WefypVXXs727VsBuOuu2yguLuLWW3/B2WefwuLFT7F/fyGzZh3Pm2++xje+cR4333wjv/rVzbz00n8P28d3v3sFK1Z8GIJnpGckiQghIobX6+XXv/45gwcP5sUX/8err77NWWfN4e23/8c777zJAw88ygsvvE5dXR3337+g0/WuWvUxs2efw7vvfsisWady3333AnDbbXeRnj6YBQvu4733PuE73/lu8zYbNqxn8eKXuO++B5k793yWLn2nuSwnZwclJSWcfPKs4DW+l0gSEUJEjG3bvuLQoRJ+9KObiY6Oxul0MnnyFN57712++c3vkJGRSUxMDDfc8GOWL1/W6VNNkyZN4aSTZmG1WpkzZx47d+Z0uM33vvfDQAxRzJp1Kvn5eeTn5wGwdOnbnHnm2djt9h61NxQkiQghIkZxcTHp6UOw2Q4fU3TwYAmDBw9pfpyePgSv10tZWeem6ktJSWn+OyoqCre7ocMElJ4+uPlvp9PJWWedzbJl7+Dz+Vi+fClz585rZ+u+Q5KIECJipKenU1xcdNQHfGpqGkVFX98+o7i4CKvVSlJSMtHR0dTX1zeXeb1eysvLOr3Pzl6Pce6557Ns2TusW7eWqKgoJk48ttP7CCdJIkKIiDF+/DGkpKTy6KMPUVdXR0NDA5s3b2T27Dm88MISCgsLqK2t5fHH/8VZZ52DzWYjK2sYbncDq1evxOPx8NRTC2lsbOz0PpOSkiksLOhwvYkTj8UwLDz00D+YM6d/HIWAJBEhRASxWq0sWHA/BQX5zJ9/Pt/4xjw++OA9zjvvQubMmcdPfvJDLr/8QhwOJz/72a8AiIuL45ZbfsOCBXdx8cXnEhUVTVraoE7v86qrruGppxYyd+7pLFnyTLvrzp07j127dnLOOef2qJ2hZJjmgL3j13BgT0/uJxKMiw37030M5L4LA19/aHNR0V4GDx7W8Yqd1J9m8X3nnTd5441XeeSRhd2uoyvtbe25bnE/kRH4Z09vlxyJCCFEH1BfX8+rr77EhRdeEu5QukSSiBBChNmaNZ9y/vmzSU5O5uyz54Y7nC6RubOEECLMZsw4ieXLV4Y7jG6RIxEhhBDdJklECCFEt0kSEUII0W2SRIQQQnSbJBEhhBDdJqOzhBB9XkJiDA67Nej1dvYW1pdeegEOhwOHwwnA1KnT+OlPbwl6PP1RyJKIUuo1/FdA+oBq4Cat9Ual1FjgKSAFOARcrbXOCWzTZpkQInI47NZuzyBhsRhtzlrRlVtY/+lPCxg5cnSrZR6P56iZgSNFKFv9Xa11BYBS6iLgSWAq8CjwL631s0qpK4HHgDMD27RXJoQQYXP33XditVrJy9tLbW0tixYt4Y9//AN5eXtpbHSTkZHFb397O/Hx8axf/wUPPHAfEyYcw1dfbQEM/vjHPzN8+AgA3nzzdV580X9nQ7vdzr333k9ycgqffrqSp59+koYGN3a7nZtu+gUTJ04KY6uPFrIk0pRAAhIAn1JqEP5EcnZg+XPAQ0qpNMBoq0xrXRKisIUQAoA//OHW5tNZN954E+C/A+FDDz1OdHQ0ADff/EsSExMBePzxh1m8+Knmdffs2cXvfnc7v/7173nqqYU89dRC7rjjT6xf/wXPPPMfHn74CVJSUqmtrcVqtVJQsI9FixZy330PEhsbx+7du/jlL3/KK6+8FYbWty2kx19KqSeAc/AniLlAFlCgtfYCaK29SqnCwHKjnbJOJ5HARGLdEhvr7Pa2TdLSXD2uI5T6W7w9FWnthb7f5gMHLNhsR4/5sVg6d1+O1rS3bWv7as1f/vJXRo36+nTW++8v46yzZuNyxTYvW7bsbZYufRuPx0NdXR3Z2dnYbBasVgvDhg1nwoQJABx77LGsXv0JNpuFNWtWMW/e+aSn+2cGjo/3f2a9885nFBbu4yc/+WFz/T6fl4qKssNugtWTNlkslh6/H0KaRLTW1wEopa4C/grc1tv77MksvjU1DT3ef1+fMbWl/jDDazBFWnuhf7TZ5/O1Ogttd/6Pof0+EaDTM956vYfHZZomTmdU87JNmzbwyisv8sgjT5KUlMSyZe/yxhuv4PH48Hp92O2O5nVN08Dj8eLx+PD5/G07Mg6v18cJJ5zEbbf9X5di7sosvj6f76j3Q4tZfDslLEN8tdbPAGcA+4AMpZQVIPB7KJAf+GmrTAgh+pSqqipiY+NISEjA7Xbz1ltvdGq7k06aybvvvkVp6SEAamtraWho4IQTTmTNmk/ZvXtX87rbtn3VK7H3REiORJRScUCS1jo/8PgCoBQ4AGwEvgU8G/i9oanPQynVZpkQInK4G71dGknVlXqD5cQTT2bZsnf41re+QUJCIlOmHMfWrR1/6E+dejxXXXUNP/vZjzAMCw6HnQUL7icrK5vbb7+Le+65i4aGBjyeRiZNmsz48ccELeZgCMlNqZRS6cDrQCzgxZ9Afqm1Xq+UGod/GG8SUIZ/GK8ObNdmWScMR25K1SX94VRHMEVae6F/tDmSb0oVDKG+KVVIjkS01sXAiW2UbQdmdLVMCCFE+Mm0J0IIIbpNkogQQohukyQihBCi2ySJCCGE6DZJIkIIIbotMqedFEL0K8mJUVjt9qDX621spLS8vsP1PvhgOc888ySmCW53A2PHjuPOO+/u1j63b9/K888v4Y47/tSt7fsaSSJCiD7Parez752nu7Vte9OeZJ57NdB+Ejl48CD33XcPCxc+S3r6YEzTJCens5erHW3cuAkDJoGAJBEhhGhXaelBrFYbCQn+2XkNw2Ds2HEAzJp1PNde+wM++WQFDQ31XH/9jzn99LMA2p0W/l//+icLFz7D/v2FXHfdVVx44Tf47LNV1NfX85vf3M7kyVPC1t6ukiQihBDtGD16LBMmHMP8+edx3HHTOPbYKcyZM685qVgsFhYtWkJeXi433PB9Jk8+jqSk5HanhW+poqKCiROP5frrf8yyZe/w6KMP8MgjT4a0jT0hSUQIIdphsVj4y1/+zu7dO9mwYT2ffPIRS5Y8w9NP+28idf75FwGQnT2csWMVX321hVmzTuPdd99k2bJ38XgaqaurJysru9X6o6NjmDnzFACOOWYSDz30j9A0LEgkiQghRCeMHDmakSNHM3/+5Vx55WVs2LCuzXU3bdrAa6+9fNS08K1xOL4eMGCxWPB6PUGPvTfJEF8hhGhHSckBvvxyc/PjAweKKS8vY8iQoQDNU77n5+eRk6M55phJ3Z4Wvj+SIxEhhGiH1+tl4cLHKCraj9MZhWn6uO66G5s7171eL9de+23q6+v51a9+R1JScrenhe+PQjIVfJgMR6aC75L+ME14MEVae6F/tLm16cnDfZ1IW2bNOp5lyz4mJiYmiFH1zICcCl4IIXrC/0HfvQ/7SLufSKhJEhFCiG5aufKLcIcQdtKxLoQQotskiQghhOg2SSJCCCG6TZKIEEKIbpMkIoQQottCMjpLKZUCPAOMAtxADnC91rpEKWUCW4CmMXhXaa23BLa7APhrIM51wLVa69pQxCyE6DsSk6Kx24L/cdXo8VBeVtfheh6Ph6eeWsjy5UuxWm1YrVaysrL4/vdvYMSIkUGPqz8J1RBfE7hXa/0RgFLqr8A9wPcD5SdrratbbqCUigP+DZyitc5RSj0B/BL4vxDFLIToI+w2G4vWvNytbdu7n8g1M+Z3qo4///mP1NfX8/jjT+FyuTBNk08/XUVe3t5OJxGPx4OtFxJhuIWkRVrrUuCjFos+A27sYLNzgS+01jmBx48CTyFJRAgRQvn5eXz88Ye88srbuFwuwH9PkZNPngVAY2Mjjz/+MBs3rsPtbmT06NHccstviYmJ4e6778RqtZKXt5fa2loWLVrCrFnH84Mf3Mgnn6ygoqKCW2/9PV98sZY1a1bj8Xi4664FDB8+gkOHDnLnnb+npqYGt9vNySfP5Ec/uhmAhQsfIy9vLzU11RQWFpCRkclddy3AMAwuuugiFi58ltTUVAD+8Y+/kpycwtVXf69Xnp+Q94kopSz4E0jLGck+UkptVEr9RSnlDCzLBva2WCcPyApRmEIIAcCOHZrMzGzi4+NbLV+8+CliY2P597+f5qmnniMlJY1nnvlPc3lOzg7+/vcHWbRoSfOyuDgXTzzxNDfeeBO//e0tTJo0mf/8Zwlz557H008/2bzOggX38+STz7Jo0RK2b9/GZ5+tbq5D623cccfdLF78Eh6Ph2XL3sHpdDJv3vnNMwbX1tayfPkyLrjg4t54aoDwXLH+IFANPBR4nK21zldKxePvN7kN+EOwdhaYA6ZbYmOdHa/UgbQ0V4/rCKX+Fm9PRVp7oe+3+cABCzbb0d9vLRaj23W2t21r+2rJajUwjK/X27NnN7ff/jvq6+s56aSZfPnlZmpqalix4gMA3G43Y8aMxWazYBgGZ501G5cr9rA658yZg81mYcKE8RiGwWmnnQbAhAkT+PjjD7HZLDQ2wkMPPcCWLZswTZPS0kPs3p3DrFmzsFgMTjzxJJKSEgCYOHES+/cXYLNZuPTSy7n++u/zve9dx/Ll7zJjxomkpaW28bxYevx+CGkSUUr9DRgDXKC19gForfMDvysD/R6/CKyeB5zRYvNsIL+r++zJBIw1NQ1d3u5IfX2yu5b6w+R8wRRp7YX+0Wafz9fqXFfd+T+G9vtEgA7n1Ro1aiz5+XmUlVXgcrnIyhrOf/6zhJdffp7t27fh85n84he3Mm3a9KPqNU0TpzPqqH1YLPZAuYHdbm8uN00Dj8eLx+Nj8eJnqKys4LHHFuF0Olmw4G7q6urxeHz4fCY2m6NFvQaNjR48Hh/p6YMZN248H330IS+99Dy//vXv22yjz+c76v3QYgLGTgnZ6Syl1J+BacDFWuuGwLIkpVR04G8bcCmwMbDJu8B0pdSYwOMbgBdCFa8QQgBkZWUza9ZpLFjwJ6qrvx7/U1fnH9U1a9apPP/8Yhoa/BNE1tbWkJu7p8f7raqqIiUlFafTSUnJAVauXNHpbefP/yYPPHAfNpuNiROP7XEs7QnVEN9jgN8CO4DVSimAPcC9wGOBYb52YDX+01lorauUUj8E3lRKWYENwM2hiFcI0bc0ejydHknV1Xo74/e/v5NFi57guuuuxmaz4XK5SE1N48orr2HUqNEsXPgY1113NRaLBTD43vd+wPDhI3oU22WXXcFtt93KVVddTlpa+lFHOu057rhpOBwOLrnksh7F0BlyP5E2yP1EBr5Iay/0jza3do+Lnoi0qeBtNgt5efnceOP3ef7514iKimpzXbmfiBBCiMM8/vgj/O9/r/OTn/ys3QQSLJJEhBBiAPnhD2/ke9+7PmT7k7mzhBB9zgA+zd5nBOs5liQihOhTrFYbjY3ucIcx4Hm9HiwWa4/rkSQihOhT4uISKS8vwe1ukCOSXmKaPqqqyoiO7v7F2E2kT0SIbjBNk62lmo/yV1FYU0SMLZrp6cdxWtZMnFZHuMPr16Kj/Vd3V1QcxOvt3BDc9lgsFny+yBmd1bn2GjgcUcTFJfR4f5JEhOgin+njtZ1v837+xyQ44hmXPIbS+jJe3/0OnxR+xk1TrmNQTFq4w+zXoqNjm5NJT/WHYc3BFOr2ShIRooteynmDFftWc1rmycwffQHWwHnlnLLdPPHlM9y//lFuPu56BscOCnOkQvQ+6RMRogs2lXzFin2rOSNrFpeNuag5gQCMSRrJz6begM/08e8tT9Pglc5hMfBJEhGik6rc1SzZ/hJZcUO5eNQ8DOPomWGHxKZz7THfpri2hOf1q2GIUojQkiQiRCct2/shNY21XD3hCmyWts8Ej0sew9zhZ7KmaB1fHtwWwgiFCD3pExGiEyoaKvmk4FNOGDyVCVkjOrzf95XJF7H50Fe8vOsNTh4zBaft8BFbnb23txB9nSQRITrhvb0f4TV9nDt8dqfv950Zm8GaonX8cfk/GJs06rCy3piRVohwkNNZQnSg3tPA6v1rOT59CmkxKZ3eLiUqicExg9hTmSed7GLAkiQiRAfWH9hMg9fNrKEndnnbsYmj8JpedlX0/CZFQvRFkkSE6MDqwjUMjhnEyISu3+MizhFLZtwQ8ir3Ueep74XohAgvSSJCtKOwuog9lXmcPPSEVof0dsaYRH9/SE757mCGJkSfIElEiHasP7AZA4MTBk/tdh3RtiiyXZnsqy6k2l0TxOiECD9JIkK0Y1PJl4xOHIHL0bPZTkcljsBqWNkpfSNigJEkIkQbimtLKKwpYnLaxB7X5bQ6yHZlUFhTRG1jbRCiE6JvkCQiRBs2HfgSgClBSCIAI+KHYcFgd+XeoNQnRF8gSUSINmw8+CXZrkySohKDUl+UzUlG3FD2Ve2nvK4iKHUKEW4huWJdKZUCPAOMAtxADnC91rpEKXUi8BgQDeQCV2qtDwS2a7NMiN5U3VhDXuU+5o2YHdR6RyYMI7+6gDd3fMCcocGtW4hwCNWRiAncq7VWWutJwC7gHqWUBXgW+LHWeizwMXAPQHtlQvQ2XboTE5NxyWODWm+sPYYhMem8t/Njahtl7izR/3U6iSilLmtj+aUdbau1LtVaf9Ri0WfAMGAaUK+1XhlY/ihweeDv9sqE6FXbS3OItkUxzJUZ9LpHJQynzlPPxwWfBr1uIUKtK0ciC9tY/nhXdhg4wrgReAPIBpp7GbXWBwGLUiq5gzIheo1pmmwr3YFKGn3YTaeCJd7p4tj08XxS8Clenzfo9QsRSh32iSilRgb+tCilRgAtL9sdCXR1LocHgWrgIeCSLm7bZSkp3R/fHxvr7PH+09JcPa4jlPpbvD3VWnsLK4soayhnfva5bT4fMT18b8zNPJ17Vz5Crns3J2Z1/0LG7oi01xgir82hbG9nOtZ34u/TMPD3ZbRUBNzZ2Z0ppf4GjAEu0Fr7lFJ5+E9rNZWnAj6tdWl7ZZ3dH8ChQ9X4fGZXNgH8L0JNTUOXtztSSUlVj+sIlbQ0V7+Kt6faau+n+zYBkGHParU8Lc1FbQ/fG1MnTCQ5Kon/bX2fUVFjelRXV0TaawyR1+aettdiMbr05bvDJKK1tgAopVZorU/rbmBKqT/j7+c4T2vd9B+4DohWSs0K9H3cALzYiTIhes3O8j0kOhNIi+78tO9dZbFYODXjJF7b9TaF1UUMjRvca/sSojd1uk+khwnkGOC3wFBgtVJqo1LqVa21D7gKeEQplQOcBvwmsL82y4ToLaZpsrN8N6MTR3R7wsXOOmnodOwWGysKVvfqfoToTZ2+TiTQH3I3MAU47FhHa53d3rZa6684vC+lZdlqYFJXy4ToDQfrSqlwVzE6cUSv7yvOHsu09Cms3b+Oi0aeS4w9utf3KUSwdeViwyX4+0RuAWTyHzEg7QxM1z46cWQHawbHaZkn89n+L1hTtI4zsmaFZJ9CBFNXksgxwMzAaSYh+qXEpGjstq/f9keOYtm3ex8uZxyTho3q9dNZANmuTIbHZ7OqcA2nZ84MyT6FCKauJJGPgePwd3gL0S/ZbTYWrXkZ8A/TPXKU1dp9G3E54ieIjF4AACAASURBVHhq7Stt1nHNjPlBjenEIdP4r36VfdWFZLkyglq3EL2tK0kkF3hXKfUq/qG9zbTWtwczKCHCocHbQK2njuxeuEq9PVMHTealHW+wpmidJBHR73TlivVY4E3ADmQd8SNEv1dW759ZN8mZENL9xtpjmJg6gS+KNsoV7KLf6fSRiNb62t4MRIhwK2+owMAg3hH6q5tnDJ7KxpItbCvdwcTU8SHfvxDd1ZUhvm0OV9Fa7w5OOEKET1lDBQkOV6/Ml9WRCSmKWHsMa4rWSRIR/UpX+kRaTn/SpGk+kdD/1wkRRD7TR4W7kuww9UnYLDaOT5/CqsK11DbWyTUjot/oyhXrFq21NfDbgv/q88fxX1UuRL9W6a7GZ/pIcgbnLobdMWPwNDw+DxsObA5bDEJ0VbdvSqW1LgJ+BvwleOEIER7lDf5O9cQQd6q3lO3KJD1mEGuK1octBiG6qqd3NlRATDACESKcKhoqcFodRFl7Pv1/dxmGwfHpk9ldkduc1ITo67rSsf4JX/eBgD95HAP8X7CDEiLUyt2VJDjiw37F+NRBk3lrz3tsOLBFpkER/UJXOtafOOJxDbBJa50TxHiECLlGn4eaxloyYoeEbJ9en7fVGwelpbkYtj2DLaVfcvnUc9uto9HjobxM7tMuwqsr14k81ZuBCBEuFQ2VACQ440O2T6vF2jz9ypGcRhT60C4eWbWYaFtUm3UEe/oVIbqjK6ez7MAf8I/GGgoUAs8Ad2ut3b0TnhC9r7lT3RG6JNKeIbHp7Cjfxf6aYkYmDOt4AyHCqCsd6/cCs/HfYXBy4PeZwIJeiEuIkKlwVxJji8FutYc7FMA/DUq8w0VRTXG4QxGiQ13pE7kMmKy1PhR4rJVS64FNwM+DHpkQIVLRUElyVFK4wzjMkJh0dPlOufBQ9HldSSJtDVsZsDdAiI0N33BPERr1ngbqvQ0khrA/pDOGxA5Cl++kqLaYkQnDwx2OEG3qShJ5EfifUuqPQB4wDH8fyYu9EVhfUJmzqWcVzBkXnEBEr6lw+/tDEvpIf0iTGHsMCY54CmskiYi+rSt9Ir8GlgP/wn9jqgeBD4Bf9UJcQoREeUNl2Gbu7ciQ2HQq3VXUNMrdqEXf1eGRiFJqJnCh1vpW4PbAT1PZAmAq8FmvRShELypvqMTliAvLzL0dGRI7iO1lORTVHGBU4vBwhyNEqzpzJPI7/LfGbc2HwO+DF44QoWOaJhWBK9X7omhbNInOBPbXFnW8shBh0pk+kSnAu22ULQee7MyOlFJ/A+YDw4FJWusvA8tzgfrAD8CtWuulgbITgceAaPy3571Sa32gM/sToiPV7ho8Pk+f61RvaUjMILaV5VDdWEOcPTbc4QhxlM4cicQDjjbK7EBnTya/BpwK7G2l7FKt9ZTAT1MCsQDPAj/WWo/FfzR0Tyf3JUSHSuvKgfDO3NuRwbHpAHLNiOizOpNEtgPntFF2TqC8Q1rrlVrr/M4GBkwD6rXWKwOPHwUu78L2QrSrrK4Cq2Ht09/wo21RJDkT2F8jB+Cib+rM6az7gceUUlbgNa21L3CUcDH+kVq/CEIci5VSBrAS+J3WuhzIpsVRi9b6oFLKopRK1lqXdrbilJS4bgdld3RlBHTrWptkry/rb/F2R0zg+p/S4nKSohOIjWt7fqqO6ghGHB3JTspgU9FWPDY38c7DX5/uvF6R8BofKdLaHMr2dvgpqbVeopQaDDwFOJVSB4FUoAG4Q2v9XA9jOEVrna+UcgL/AB4Cruxhnc0OHarG5zM7XvEIaWkuGt2eHu+/pKSqx3WESlqaq1/F2x1paS5qaxrwmT7K6ysZ7sqitqahy/V0Z5vu1pFiSwFgz8F9jEkceVhZV1+vSHiNjxRpbe5pey0Wo0tfvjt1nYjW+j4gA7gA+GXgd0ZgeY80neLSWjcADwMzA0VNFzQCoJRKBXxdOQoRoi1VgdvhhnLm3u6KsjlJdiayX/pFRB/UlangK4Glwdy5UioWsGmtKwKns64ANgaK1wHRSqlZgX6RGxjAV8eL0CoPTP/el0dmtTQkNp2vSjVV7mpcju6fohUi2Hp6e9xOU0o9oJTaB2QCy5VSXwHpwEdKqc3Al8BY4EcAWmsf/mnnH1FK5QCnAb8JVbxiYCt3N90Ot+v9IeEwOHYQgByNiD6n5z3HnaS1/inw01aKjmtnm9XApF4LSkSsioZKkqMTw3473M5yWp2kRCWxv6aYMYkj+03cYuAL2ZGIEH1Fo89DdWMNSdGJ4Q6lS4bEplPjqaXKXR3uUIRoJklERJzKQH9Ickz/SiKDYwZhYLCvZn+4QxGimSQREXHK3YEkEtV3r1RvjcPqID0mjYLq/XhNX7jDEQKQJCIiUHlDBTG2aBy2tmbz6buyXBk0+hoplivYRR8hSUREnIqGyj49X1Z7UqOSibZFkV9dEO5QhAAkiYgIU1pXTr23oV9cZNgawzDIisvgUH0ZRVVyNCLCT5KIiCi7Sv3TsSX20XuIdEZm3FAMDN7fvSrcoQghSURElp2Hcvvs7XA7K8rmJC06hY9yP8Pr84Y7HBHhJImIiLKzNLfP3g63K7JdmVTUV7L54NZwhyIinCQRETF8po9dpXv7zXxZ7UmLTiEtNoUP8z8JdygiwkkSERGjpPYgtY11JDj658islgzDYN6YM9hVkUtuZV64wxERTJKIiBi5lf4baw6EIxGAM0fOJMoaxQd5cjQiwkeSiIgYe6vycdqcffp2uF0RbY9iZsYJbCjZwqG6snCHIyKUJBERMXIr8xmVlD2gZsA9I3MWAB/tWxnmSESkkiQiIkKjz0NBVSGjU4aHO5SgSopKZOqgY1lduJY6T324wxERSJKIiAj7qgrwmF5GJw8PdyhBd1bWqdR7G1hVuCbcoYgIJElERIQ9Ff4r1VXqqDBHEnzZ8ZmMSRzJh/kr5eJDEXKSRERE2F2xl5SoZJKi+//w3tbMzj6N8oYKvijeGO5QRISRJCIGPNM02V2xl5EJw8IdSq85JmUcQ2LTWZ63AtM0wx2OiCCSRFqxNbeU/76nyau0UtM4cEbyRKrS+nIq3JWMGMBJxDAMZmefRmFNEdtKd4Q7HBFBJIm04vPtB1j87na2HLLz0T4nKwscFNdakC94/dOeilyAAX0kAnB8+hQSnQm8l7ci3KGICGILxU6UUn8D5gPDgUla6y8Dy8cCTwEpwCHgaq11Tkdlve27c8dx87em8cCDL3Gg1kJupZUvih0MjvEyKbURR/+euy/i7K7ci8PqYGjs4HCH0qtsFhunZ87ktV1vk1e5j+z4zHCHJCJAqI5EXgNOBfYesfxR4F9a67HAv4DHOlnW6+w2C7F2kxEJXk7LdDMuqZHiWgsrC51Uu+UUV3+yu2Ivw+Oz+/3MvZ0xK2MGUdYolsvRiAiRkCQRrfVKrXV+y2VKqUHAVOC5wKLngKlKqbT2ykIR75EsBoxK9HLyUDdeH6ze76CiQRJJf1DvaaCgev+AP5XVJNoWzayMGaw/sJmDdaXhDkdEgHD2iWQBBVprL0Dgd2FgeXtlYZPoNDl5qBubBdYWOeSIpB/Iq8rHZ/oiJokAnJE1C4th4QOZJl6EQEj6RMIpJSWu29vaHUc/PYkOmJnl5ZM8K2uLHZw+zIuznWcxLa1/3UGvv8XbkU9KigA4fuQE4hz+iRdjYp3N5S3/7orubhfMOrw+b6uvVxouThl2Aqvz1/Ld4y8hznn4hJMtt/H6vBFxmm+gva87Esr2hjOJ5AMZSimr1tqrlLICQwPLjXbKuuTQoWp8vq4Pq0pLc9Ho9rRa5gSmp3tZvd/Bmn0WZgxxY2njoKSkpKrL+w6XtDRXv4q3M7YU7mBwbDp1FT7qqCItzUVtTQPg/xBv+rururtdMOuwWqwsWvNyq2Uetw+3t5F7VzzKyIThzcuPbPM1M+YPuNf8SAPxfd2enrbXYjG69OU7bKeztNYHgI3AtwKLvgVs0FqXtFcW+khbl+A0OTa1kdIGC7pswB/Q9Us+08eeir2MjI+cU1lN4h0ukqOSyK3ch8/0hTscMYCFJIkopR5QSu0DMoHlSqmvAkU3ADcppXYANwUe04myPiEjzke2y8PuCiuH6uSSm76mqOYAtZ66AX2RYXuGu7Ko99ZzoPZguEMRA1hIvkJrrX8K/LSV5duBGW1s02ZZXzI+2cOhOgsbS+ycmtGAfeCfXu43csp3AzA2aWSYIwmP9Jg0oq1R5FbmMTh2ULjDEQOUfH3uIZsFpgxqpMELXx6yhzsc0UJO2S6SnImkRCWHO5SwMAyDYfFZlDaUU+mOnD4BEVqSRIIg0WkyOtFDYY2V/TXylPYFpmmSU76bsUmjBtSdDLsqK24oVsPSfH95IYJNPvGCZHSil3iHj68O2WmUfsyw219TTHVjDWMSI/NUVhO71c7Q2CEU1hTh9rrDHY4YgCSJBInFgEmp/tNaO2S0Vtg19YeMSRp4N6HqquHxWfhMH/uq94c7FDEASRIJokSnSbbLS26lVaZFCbMdzf0hSeEOJexcjjiSnAnkVRXIvUZE0EkSCbJxSR4cFthy0I63Gxc5ip7zmT52Sn/IYbJdmdR6aimpORTuUMQAI0kkyOxWmJDSSIXbwrLPcsMdTkQqqjng7w+RU1nNBscMwm6xs7ssL9yhiAFGkkgvGBrrIznKyzPvbKO6rjHc4UScHeW7ABgb4Z3qLVktVjLjhlBQWUSDt+dTtgjRRJJILzAMOCbFQ01dI69/sifc4UScnLLdJEclkRIdmdeHtCXLlYGJSX5VYbhDEQOIJJFeEu8wmXvScD7cUMC+kupwhxMxfKaPnPJdjE2UU1lHirPHkhabQn61dLCL4JEk0ou+M3c80U4rzy3PkX/aENlfU0xNYy2jI3Sqk46MTMqmzlPPwTrpYBfBIUmkF8XHOrj4lJFs21vGur4zAfGAtvWQBmB88pgwR9I3DXWl47DYya+WU1oiOOSquF52+nFDWbGxgOc/2Mmxo1JwyAyNvWrrIU1G3BASnQnhDqVdXq+Pa2bM73EdXWW1WMmIG0JuZb50sIugkCTSy6wWC9+aPZa/PreBd9fmceHMEeEOacCq99SzqyKXM7NOCXcoHbJaLfx64as9quPe71/Sre2y4jLYU5knV7CLoJDTWSEwflgSx6s03v50L6WV9eEOZ8DaUbYLr+llQsrYcIfSp8U5YklyJpJfVSh9daLH5EgkRC4/czSbdh3ihQ93csNFE8MdzoC0tXQHDqvjsNvB9oZgnIoKtyzXUDYf3Mq2khzSjCHhDkf0Y5JEQiQ1IZpzZ2TzxqpczpxaztisxHCHNKCYpsmWg1sZlzQGm6V339btnYoyTfB4wO2Gxkb/3x6PgdcLpg98JmDC+OzB7NtnYFjAZgWbzcRmA7sDnE6w9nLX2ZCYdLYaO1i+exXfGnVp7+5MDGiSRELo3BOHsXLLfpa8t4Pbr5mOxSLzOgVLXtU+yhsquGDknJDt0zT9yaKuDurrDOrroaEBfL7DX1eLxcRqBcPin+3ZMKCsqp4Gtz+xeDxgmoefWbZaTaKiIDoaoqJNYmKCm1isFitD4wazJn89F2bPI9YeE7zKRUSRJBJCTruVy88YzaOvf8XHmwo5/biMcIc0YGwu+QqLYWFi6vhe3U9NNSxbs5eCfQa1tf6jDPAniqgoSEgEh8OH0+E/qrDZwNJKz+O93z/9sKMZn8+ksdGflNxuaGgwqK+DgwfB33VpEh0NsbEQ5/Lvq6ey4zLIq9rH2qL1nJE1q+cViogkSSTEpo8bxAfrC3jl491MHz+I2Ci5pW4wbDr4FaMTRhBnjw1qvaYJ5WVQWGChsMCgstLgXTZitfo/0GNifcTEgMPhP8LoLovFfxrL6WzeMwA+H9TVmdTUGNTW+JPKwYMW7HaThW98SbkNkpO7t+94p4tRScNYXbiW0zNnyozHolskiYSYYRh8e/YY/rjoc17/ZA/fPltGEnVWYlI0dtvRb9n9VQfYX1PMnONOJS3N1W4dnekUN02TnPxyVqzfx+rNhRysqMdiwDEjU5lxxmCOG5vGg2++36Ok0VkWiz9ZxcaagfihqsqkstLgzZW78XhtpCVFM3t6NrOnZzMouWunpZbmrGDh+v+SW5nPiITs3miCGOAkiYRBdrqL06Zk8MH6Ak6bMpSMtLhwh9Qv2G02Fq15+ajlTXcx3Fuyn0VlR5e3dM2M+c2nkewOK41ub3OZ2w0VFVBZYeB2GxiGSWwcDBlqEhcHHtsBVuUc4KJTLwlJAmmN1QqJiZCYaHL7t8/lloffpKKilueWbee5ZduJjfWXueJbPzo5ss13Xj2XZza+zOrCNZJERLf0iSSilMoF6gM/ALdqrZcqpU4EHgOigVzgSq31gXDEGGyXnDKCtVuLee79HG755hQ5ldBNpmlSUF1ESlQS0baudxSYJlRVQVmpQW2tAfg7sVNSfLjie3+UVE/ERdtJSISERJNGN5SXG5RXQEGBBWuxSVKSSVKSv1+mLTH2aKalT+GLA5uYP+YCorrxHIrI1pcuNrxUaz0l8LNUKWUBngV+rLUeC3wM3BPeEIPHFePg4lNGsDW3jA05B8MdTr9V4a6k1lPL0NjBXdqusRGK9pvszDEo2GehsRHS0nyMHmMybLhJYlLfTiBHsjsgbZDJ6NEmWVk+opxwsMTCzhyDwkL/yLG2zBx6Am6vm3XFm0IXsBgw+lISOdI0oF5rvTLw+FHg8jDGE3RnTM0gIzWW/76fg7vR2/EG4igF1fuxYGFwbHqH65ombN5Zwr58g505BgeK/R3ZmVk+Ro02SU0Dez8f52AYEOeC7GEmI0f5SEiEygrYs9vC3lyDykqTIy9SHx6fzZDYdFYVrg1P0KJf60tJZLFSarNS6mGlVCKQDextKtRaHwQsSqkBc6chq8XCt2eP4WBFPa+vkptXdZXX56Wgpoj0mDTs7Vxg2OiGnTkG7y218vtHVlNbC8kpoMb7P2xdrp6NrOqrnE4YMsRkzFiTtEE+3G7I3Q17dhtUVNCcTAzDYObQGeytypcbVoku6xN9IsApWut8pZQT+AfwENCz2ekCUlK632ltd/T86elotFBamotNe8pYumYvs2cMZ2x2Uo/32RMdxRtuMbHNY2DZU5aHx+dh7KCRhy1vUlZqoreb7Nll4vFASirc/M3jeH/bhhYXenbvnJXd0fNzXaGqww4MiYbBQ0zKyqCk2KCwwKCkBN5atYfZJ2QzL/5U/rf7XT47uIapI6/ucVx9TV9/XwdbKNvbJ5KI1jo/8LtBKfUw8AbwT2BY0zpKqVTAp7Uu7Urdhw5V4/N1bZI5T8FWnJW5DK3ajhcbDZYYqqwJ1BsxXf7KWlJS1eE6F540jM+3FvH3xeu445rp2G3hOUBMS3N1Kt5wSUtzUVvjn77cNE12lOzBZY8jxoxpXu7zQkGBwa6dFg4dNLBYTbKyTEaO9pGcDLNPyGbZlnV4OXqkUld0d7tw15GcbCUu1kt1NRw8aPDoK5tZ8u42zp6exdTkaazMXcucjNnEOwbOh25ff18HW0/ba7EYXfryHfYkopSKBWxa6wqllAFcAWwE1gHRSqlZgX6RG4AXQxGTZ9da6rZ/hDpieSN2ym0plNiGUmLPoMESHZT9xUTZ+O7ccfzjxU28sWoP80+TW7t2pLS+jKrGaiamjMcw/Bfi7dltYc9ug4YGg9g4k0mTvQwbbra4gE+A/3uQywVxcSZXnzGLJe9u4+UVu3Hak/CmjmLpoE+57Jhzwh2m6CfCnkSAdOBlpZQV/7mFrcCPtNY+pdRVwGNKqSgCQ3xDEVDUqdeQecmPeOJfz2EzPUT5anB5y3H5yknxHCCtfj3Ur6fMmkq+YxTF9kxMo2enJo4dlcLMSYN557M8poxOZVRG376pUjiZpsmO8t04LA7s1UP4dLOFwkIDTP81HSNH+UgfbA7Ifo5gMgyYNCqVX1w+hb1FVbyzZi9rt3l5502TqryvOO/EEaQnyZxaon1hTyJa693AcW2UrQYmhTYiP8NixWvY8Rp2GizRVNhS/QWmSayvkkGNBQxtzOXYujU01G9kn2MUex1j8Fi6/7X3W2eNQeeV8+jrX3LHtScQF93PhwoFWdPV5mtyv+R/a3YSW3Eyqyt8xMc6uPSMYcw9aTjpXbxiW/gNG+zihosmcsLUKB5ZtpLVWyys2lzM9HGDOHfGMIYNHjint0RwhT2J9DuGQY01gT3WBPY4x5PiKSbLvZNRDVvJbsghzzmGvc6xeAxHl6uOibJzw0UT+cuz63jyrW3cNH+SXITYQu7+Su74z4dUVJhgjqch2svQoSau+Hq+KtJ89arusI7u3g0wUkzNGs3U4z9mZ8mnHG+Zzycbi1m77QAThicx54RsJo5IlvekOIwkkZ4wDA7ZB3PIPpg4bzmj6rf6k4l7Jzudx2D6ut5xOnJoPJefMZrn3s9h6dp85s6I7KkoGhq9fL7tACs2FrCrsBLDAtbUQlITnSTbU8Md3oB03ohzWHDoAWKG7eFvJ8/mww0FLF+3j/tf2ERGWiznTM/ixAmDwzYARPQtkkSCpNqayKbYk3F5yxhbt4nx9RvY98QvsZ1wBbaMCV2qa/bxmezIL+flFbsYMcSFCvOw33AoOFjDig0FrP6yiNoGD4OTY7hi7gheL36SaMNFUn2rZ0BFEGTHZzJj8DTez/uYE4ccz3knDWfOCdms2VrM0rV5/Oft7byyYjdnTsvkjOMy5LRrhJMkEmRV1iTWxZ7GIE8Bx7t3UvfWvdiGT8N50hVYXGmdqsMwDK6dN467n1nHAy9v4bdXTiUzAiZpbPR4+UKXsGJDATv2VWC1GBw/bhCnTxnK8KGxPLjlMQybl5Ta8RjIKZXedNGoeWwq+ZIXd7zOjyd/H5vVwsxJQzh54mC25paxdG0er368m7c+zWXWpCGcPT1LOuEjlCSR3mAYHLBnknnDTRS+/yLujW/ieWELjqkX4jh2Loa146c9JsrOzy+fzN3PrOP+Fzbx+6umkRw/MCfHKyqtZcXGAlZtKaK6rpFBSdFcdsYoZk4aQnyMA5/p46mt/yW3bB9p9VOwmcEZWi3aluB0ccHIubyY8zqfFHzKqZknA/4vOMeMSOaYEcnsK6lm2dp8Pt5UyIfrC5g8OpUzp2UwYXgyFuk3iRiSRHqRxebAOfVC7GNn0rB6Ce7PX8KTswrnzKs6dYorNSGan182mXsWr+f+Fzfxm+9MHTA3saqtb2Tt9gOs3lLEzgL/UcdxY1I57bgMxg9Lav4Q8pk+ntv+Cl8Ub+Tbx17M6tW1YY48eKZk9u5dGDvD6/W1eXXzpalz2Fm9k1d2vcXxI45hWGLmYeVpaS6OmzCE4tIaXvtwJx9vLGDjzoOkJ0VzxnEZzDx2yIB5v4q2SRIJAUtcCtHn3IQnbxP1q571n+IafRLOE7+JJSax3W2z01385BuTuP+FTfztuY3ccsWUfnkOOiExBqvFYMOOEt7/PI81XxXR6PGRle7imvMmcObxWSQdcaTl9rh5+PNnWL3/C74x4VwuGncOq1e/1ruBmiY2vDh9bhw+N07T/9uGF6vpxWZ6qVj7JlOqv8Rmfj1wwuDwWRF8hgUvVjyGFa/x9W+3YafB4qCxrIh9m9+jEVv3J+6aM64nLQXAarWwZOn2NsvTfSexzZfL7cv+yamubxJtOfq06jdnj+X6+ZP53kUTWbV5P2+v2sN/P9jJK5/s4fSpmcw7eTijMtt/n7sbvVSUD5wvCJFEkkgI2bInEzt0PO6Nb+He+BaevRtxTp+PfcKZGK3diDtgwvBkbpo/iYde+ZIFi9fzs8smk5LQP05tmaZJ/oFqNn66l3dW59LQ6MVhs5CZFktWWhwJcQ5Ky2t5afnhw3MbzFo2ed+lwjzAaMsMKndkY0wyuvXt3fB5cDRWYm+spkavYULtDmK8dcQaDUQ11hHtqyPK1xBIGI1Y8bVb36H31nDike08oo/myKRypPyH3+VS/DfBdWOjwXBQh5M6I/AT+LvWiKLaiKaeHt5/tx0b921rtzzJMomi6M/5oOx5BtdPw2oefi3Ut63j+PcrX08jPyYjnkEJTnKLqnj/8zyWrdlLksvJsPQ4hqbGYrMe/V7/wTcmB6cxIuQkiYSYYXPgPP4S7GNOon7VszSsfpZG/QlRp1yNdVDb050cOyqVn192LA+9uoU/Pf0FP5k/iVFD++ZV7aZpUlBSw9rtB/h8+wGKS2uxWgzSEqPIGhRHelJMiwkQj1ZlHmKj5x3c1DPZOodBlhHNZXu2fHTU+jbTQ5xZR6xZRwz1xJr+nxiznlizjmjczesWb4dTA3832mJotLtodCTQaIuhwRaFxxqN1xqNp+XfVic+qwOfYcNnsfGN2RP460NP48XS5ge7YZpY8GHFixWf/8f0YseDw/Rw6bmnsPTtpdjx4DQbceIm2mwgxVdBDA1HJTIPFqqNaKqNaGqMGKqJpmbH5yR4Kqm0xmEavTfc1umLZ1D9ZA5EbWR/9OcMrpvWYb9UQpyTyaOdTBieRN6BavYWVbFx5yG27C4lIzWW7PQ4klxOueZkAJAkEiaWhMFEn3sLnj2f07B6CbWv/Qn7+NNwTr8UI6r1kVjjhyfzuyun8c+XNnPPs+u57IzRzD4+s090Ypqmyb6SGtZpf+LYf6gWw4Bx2UnMmZ7FOSeP4IVlbZ82aVLiy2WLdzk2nEy3XUy84b8WxOJrpKE4l0xvMS6z9rCflkkC/B+4tUYUtURRaEmlxoiizoiiDgffufo7/OvZN6jHgcVmw+v1gRsOr8IH1AZ+jma9YAbeDqa5MQ0DbyCFNGvxMrmOPR3d1vNhmjjwEB1IhHHUEWfWNSfKdF8ZdrwUv3gP3wK8WKiwuSi3xlNuS6DMFvhtjcdjCc6pz2hvKul10zgQvYH90WtJr5+Gw9fxiEG7zcqooQmMHBJPaVUD+cXVFBysIe9ANXHRVBcyHwAAFUhJREFUdrLT4yJi5OFAJkkkjAzDwD7yBGyZk2hY9xqNX76HZ886HFMvxD7+jFZHcWWkxXHHtdNZ+OY2/vt+Dht2lHD1XMWQlNiQx9/o8bJtbzmbdh1k886DHKpswABUdiKzp2UyVQ0iIdZ/5X5CXPvTwZimyV7fZnJ8nzLIl8TJjZNIqdyBq/5jXPVFxLgPUaBNTgmsX4eDKiOG/ZZUqowY/r+9Ow+O5KoPOP7t7umZnhnNrWNPSXvp7eE9vPb6XkIZbIwJhQ2UHXOTkASKkEpVkuKPcISqBFMOgQCGMiQOCRgIODGUbcAHiW2MD2zveu097Oddr7WH7mM00twz3Z0/eqSVtJJ2V7vSSuv3qZoabb/uUb99mvnNe6/790a0EDnNIqcFKWFO20MILF1NQVvgQ4GaRhmTsmaSYYpJb9clQIW/+MAN3PPz+4lXh0lUMySqGVpLx9HHDaVl9RBDvihpX4y0L8aIk6CPCEX9zP8PLCfBksIOeqzddAefo7FwMZZzevcwaZpGKmqRilpctCpJx0COoz1ZDrSneeVImky+wvZ19Wxdk8JvLqIlJRUVRBYCzR/EuvI2zLZrKD39I0pP/4jy3kcIXHozvjVXnDRfErZMPvO+zfzu5S7+6/8O8YW7n+OtFy/nxitaSETmJmVtLB7C9Ol09GXZe6ifXa/2sudgH6WyTcBvsG1dA7dtbGLHxiXTXoocnrzmh+sSKA8RzHeQzTzDsnwnt5QhUu0FvDmSkhkjF1pKOrWZ7VdczA8efIoRLURVexP/6WoaJfxYy9t4LThxCFR3baJ2lng1Q6I67D3bw4jC6/jdKgx7+xX0AIO+OP0P9bMpNzAWZAq6NePci9+JeIEkuJvu4C4aShed8en7fDotTRFamiKM5Msc7cly8Fia3+/vxvIbXNLWwOWbmtjQksCYYa5QWRjexO/EhcdIrST4h5/FPr6X0nP3Unzse+i778e/7V341l2JNm71Pk3T2Ll1GVvW1vOLJw/z2O4OHn+xgx0bGrlm81LaVsannMA8E1XboaMvx+GuYdp7sjy3v5tSbRnfYMBgaTLEkmSQVMzC0HWOdmY42pmZ8rU+cdNFDL70AAl3xHs4I8TdEQJUAW8AKW0EGHATHDIipPUIaS1CWfNDESjCWzZeTfpXe8+qThc6RzMY8sUY8sVoH1/guoSdPA2MEC2mSVSHSFQzZPf9lp2lE8N2FSNIIdjoPaxG8rXnijlx+ceys5Fncw/Qp73M/a8+ius2zGp+IxLys2lVkj++aQtP7TrKMwd62CX7eGpfN9Gwn8vWN3L5piZWL42q+ZMFSgWRBUbTNHwrt2CsuIjq4Rco73mA4hN3o+36BebGazHFTvRgdGz/WNjPR29Yz41XtPDI88d4el83z+7vIeA32NCcYENrgmWpMI2JIKmoddKEtuO6FEs2g8NF3ujNcfDIIMd7sxztGaGjP4ddW9ArFbNYkgrRmAzRmAgRCZnTvql1u0yo0E0430ko30U430X7Hb3caFcAb85iSKuj3VePjBY5HtQxci1ECyoX1pma7b0mmdrjyusFX//qN4i6OWJOjpibJZrLEM920FQL8OBdQZbRwmS0Ooa1MAU9zDqtjldTUe556T5W6hch9KvQZjnBb+gaG1qTbGhN8uHr23j59UGePdDN43s6+c2u46SiFtvbGtjeVs+6FfEZL8xQ5pcKIguUpumYay7Dt3oH9rG9lF/6JeXn7qX8wn34Wi/BXHcVxopNaIY3cdoQD/LB69p4/1vXsO/wIPvbB9l3eIA9h/rHvSb4TQPT0DEMjXLFpliyT7oYNRIyaW6KcP3qJC1NEVqXRNi4rpHvfvtnUAAKMLpumukUidbWWonYaaL2ECFnZGwOuaL5GdbjLL/0nfx612HSWoQRLcSwleNI/RE0N0Rrfyvh8vzP6VwIprpabSaGoXsXE9Ro71jvXXSgWfToqRM7ui4WZS+4uDliTpaom2OF04tFBWq3yFS6DIrxONJ+mlzgEFHrSvKBJopmbNaXJJs+g0tEA5eIBvLFKrtf62P3a3089mIHj75wjEjIZNvaei4RDWxoSapEkOeZCiILnKZp+Jq34Gvegp3upPLK41QOPkX18HNgBvG1bMPXvAVj6Xr0cIKAeeINCJAeKdGbztOTLtCfKVCuOFSqDhXbwTINggEfwYCPZDTAutYUhuNM2ctw7QoRO02dnaHOzhBxvGfLLYztU9BCjBhxusxmRow4I0Z8bEnhHW+/lSN77gJgIDzA8cRxrKrFqr5V+O0zT5uvzDFNo0iAohagl+SEpegD7mhwyRJzcmyJx/EdP0hw5ChwFABbNylYDScNi5UCCZiitzLTnfMtKxPc/LY28sUKu2Uvz7zcxfOv9PDky10EAz4u3dDEpRsa2bymHqpnv+SwcmZUEFlEjMQyjKs+QODyW7A7DlB943kq7bupHnoGAC3WhNGwGj25AiO5Aj3aSDycoHXL8lNe8eLaVexCFiefo5LppTrUQ2Wol2q6h8pgB+2D3Vzpet9gHXSyepRBX+NYsBjW46dckMvBoTPeyUBkgEghQstAC4arrsRZbEqanz7NTx8JMOCGD36S++/8KQPmMTL+V2gummzIRomUs0QKkgZ3z9ixNgY5PULOiJLVo2PP+nXrJtywOJN42OTa7cvpHyrQNZjn9/u7eHJPB+AtrrV5dYotq1OsXhZVw17zQAWRRUgzfGO9k8DOj+MMHsXufJVq56vYXXIsqIzKBULkHT+27q3UiKahuQ6a62A4ZUw7j+mUTvo9Vc2k6E+QNxO0XX0VT+zqIGvEyOtnfnNb50gPh5oOUfAXaBhpYOnQUpWJ9wKTqqzE0SyeTBxgV9RmW+YKAk4An1smbI8QdjLU2cPUOcPEq/0sdY+OHfvGHb/hMl+SfKCBnL+eXKCBnD9FwZ/EmeJeF0PXaEqGaEqGcF2XTK5M89IYz+7r4pfPtPPg0+2ELR9tK+NcumkJK5MhljWEF8Q9VRcaFUQWOU3XMepbMepb8W+5AQC3lMNOd+CO9OPk0lhOjs6DxzGcMoZTAVxcdFzNwNZNKkaIihFEC0XJ2n6KZoyCmaBihMbGta/4g6307P3pGZ+fi0uX1c1nH7mdqlGlta+VWHFh3mmvnL2GcgNbM1vYG93HrviLbM1sJmyHyfhSZEhN2NdwK4TtEeqcYa7etoz+vfuJFjpoGt4/Yb+iL0renyRfCyr52qNgJnB1H5qmEa8LcOt1gmu3LSNXrLD/jUH2HR7k1aNpXjy4D4C6oMn65jjrWxKsXR5jRUOd6qmcAyqIXIC0QBjfkjZY0gZAfUOEn+dOPVQQDgfI5U7ukcxW2hzi9fBhhs1hNibW4b7kU/MfbwLJSoKLM1t5KbaXF+K72TSygfryyVfe2ZrJsC/JMElS197KwxXvDn7dLmMV+wiWBrCK/VilAaziAE0jBzDtE3NwLholf5yilaIYSJF+9iiheCPReAMrdyzhxp1r0DSNnsE8ew/1s7d2U+wLsg8Ay2/Q1pxAtCQQzQlES5KQ5VOJIM+QCiLKOeXg0Bfop8PqYMifwW/72TC8ni/c8hlu3/298316yhwx/RM/SlIkuTJ7OS+GX+Ll2D5WllbQVlyLb4aPnOGDU33RSXkPH1AHPqdE2MkScrKEnBHvOTdAavgI6f/9/YQjq7qfghmnYiUx9SgbzRitK+L0L4/SXQzQl4MjXcPsfb0ft3aJYn08yIr6MCsb62huqmNlYx318aAaBpuBCiLKWXNxyfqy9Ab66Ap0UzbKWLbF2uwalheWYWCgz2GCQOX86x7um3J7a6aV7ng3xyLH6fJ1sySzhEQugc7s/h6qeoCMHjhpaAzX5U8/8S7uve8prMoQVmWIYCWDVRkiVB4iWjqMz5mYY81BoxyuIxuJccRp4kg1RSnUwJGeAi+NCyyW32B5fZimZIglo49UiKZEENOnLgxZ8EFECNEG/CfeV5IB4CNSyoPn96zmTyweOie5hE5KOXKW+9nYDJkZBv2D9AX6KRpFNBeS5STLs8tJlZNq4lxBR2fZ0DJi+Rid8U6OJ4/THesmkU8QzUcJVULo7jn4gqFpGMEII8GljASXTigKhwPkskV8ThGrFlisygj+6giBapZAdYS26jE2u6/gz+fBhHLcoNuO01FN0GEn6elPcqAnytP2iZQ+Gi7JENTX+WiI+mlIBKmPWYSCAZLxCJFYHZGQ/4JP3bLggwhwF/BtKeU9QogPAd8Frj3P5zRv/KYx46JBp+MD71g/zVDBRKbfR6VcPWm7g0P3VUkGQgNkfMNkjGEyRgZHc9BcjVQ1yZrSKhqrDfjd2pyHmvpQxgmXw6ztXUs2kKU/0k9/XT99kT5wIVgO8r0XfsyRYAeWYxG0LSzbwnTNM/4iMt2XoHCdBVi4xCnQQmHKveC2t63h+IN346sUWV0psqpahGoGt9oH1TLFcpXeop+ekkVvJcxANUx6MMyB/jBDThhnUg9LwyWkl4n4KkR8NhG/S50fwn4Ny9QIBgws/+j9WiZBy8SyAgT8fnwBP35/AJ8VQPf5QfeB4fPSH+mG9/MC6OEv6CAihGgEtgPX1Tb9BLhTCNEgpZy6/3yCAczq6ouObBd7Xt/DcMMQwCmWF5qKd8RDrz2BrjPWLfZKZn41151Y7qbhaOb0gsh0r/ygPE66cfDUL2BAxS1j41DVqti6TUkvU9ZL7H98HzSAhkZdOUhjuZ5wKUSoHB77JmnjUKA47cvHolPfTHYmzuVrGIaGbZ95657r85jP15iqzvN5HnGirCgvo1qpUvDnKfiL5EN5Dg28Qa4xP5ofEvAy5xuuien4MB0Tn+tDdw0MNDRXx8BAc0+EmV+/9hjHMnLcG8ErMQwDx555obFRj7b3om8+dVLJBJBwXXTbxigX0SsltFKJYq5CJmNTLEKpBKWyRqmqU6gaFCoBcraP/qpJqWLO8ElQrT1OTPAbOJiaPe7ZxtQcdM1Bx8XQXDRcdM3F0CBV5/DeW24+rTpPNu4z87SGQLTJH1oLiRDiEuAHUspN47YdAD4kpdx9isOvAZ6cy/NTFEW5gO0EfneqnRZ0T+QsPY/3n9DFWKYfRVEU5RQMYCneZ+gpLfQgcgxYLoQwpJS2EMIAltW2n0qJ04iiiqIoykleP90dz/+szAyklL3AHuC22qbbgBdPYz5EURRFmQcLek4EQAixHu8S3wSQxrvEV57fs1IURVFgEQQRRVEUZeFa0MNZiqIoysKmgoiiKIoyayqIKIqiKLOmgoiiKIoyawv9PpF5t5gTPgoh2oFi7QHwWSnlw0KIK/ByjgWBdrw7/ntrx5zzsjms31eB9wGtwGYp5b7a9mnbbL7L5rHO7UzR1rWyRdveQogU8ENgDVAGDgJ/LqXsm+96LZA6u8BeYDRvy4ellHtrx70b+Ce8z/FdwMellPm5KpuO6omcbDThYxvwbbw/osXk/VLKbbXHw0IIHbgH+HStTr8FvgIwF2Vz7BfAW4Ajk7bP1GbzXXauTVdnmNTWMDdtOs/t7QJ3SCmFlHIz3k1vX5nvei2EOo8rv2pcO48GkDrgX4F3SynXAiPA38xV2UxUEBlnXMLHn9Q2/QTYLoRoOH9nddYuAYpSytG79+8CbpnDsjkjpfydlHJCtoKZ2my+y851fWHqOp/Com5vKeWglPLxcZueBVrm6NwXep1n8k7ghXE94LuAW+ewbFoqiEy0EuiQUtoAtefO2vbF4kdCiJeFEN8RQsSBZsZ9i5VS9gO6ECI5R2XzbaY2m++y+Ta5reECau9ab+BTwP3noV4Loc6jHhdC7BFC3C6EGM11P+H8gKOc+Buci7JpqSByYdkppdwK7MDLg33neT4fZe68Gdr6W0CWC7Nu05lc52Yp5aV4Q5obgc+frxObjgoiE40lfAQ4w4SP593osIeUsgR8B7ga79vEWNdYCFEPOFLKwTkqm28ztdl8l82badoaLpD2rl1QsA64VUrpnId6LYQ6j2/nYeDfmKad8XoRx+awbFoqiIwjF3HCRyFEWAgRq/2sAX+EV5ddQFAIcU1t108C99Z+nouyeTVTm8132dzVcqIZ2hougPYWQnwZb07iplqQnKtzX9B1FkIkhBDB2s8+4P2caOeHgB1CiHXjzu9nc1g2LZU7axKxSBM+CiFWA/+DtxaAARwA/lJK2SWEuArvCiKLE5cq9tSOO+dlc1jHbwLvBZYA/cCAlHLTTG0232XzUWfg3UzT1rVjFm17CyE2AfuA12BsFds3pJQ3z3e9znedgTtqv98FTOBp4K+klNnace+p7WMALwIfk1Lm5qpsOiqIKIqiKLOmhrMURVGUWVNBRFEURZk1FUQURVGUWVNBRFEURZk1FUQURVGUWVNBRFGmIYRoF0K8fRbHPS6E+MRcnJOiLDQqiCiKoiizpoKIoiiKMmtqUSpFmdmO2l3jS/HW9vgU3iJFPwQux3sPPQV8Ukp5fPLBQog1eGs0bMW78/hhvDUqhmrl7XjJ9j6Cl7foIeCjUspirfw9wJeA1UBf7diHamlPvgbciLdg0feBL45mF1aU+aJ6Iooysw8C78Bbda4N+Bze++b7eB/6zXipKqbLNKsBt+MladyAl1r77yftcwtwA7AK2AJ8DEAIcRnwA+BvgTheJtf22jH/AVSBtcDFwPWAmodR5p3qiSjKzO4czaQqhPhH4FtSys/h5a5i3PbHpjpYSnkIOFT7Z58Q4mvAFyft9k0pZWfttR4AttW2/wnw71LKR2v/7qjt04TXA4lLKQtATgjxdeDPWHwrcSqLnAoiijKz8amwjwDLhBAh4Ot4vYdErSwihDAmDyfVPvC/AewEIni9mPSk39E97uc8Xq8FvF7Lr6Y4pxa8hHxdQojRbTqLZMkC5cKigoiizGz8ym7NeKsY/jUggMullN1CiG14GU+1KY7/Mt5cyGYp5aAQ4iZOf5GlY3jDaFNtLwH1Usrqab6WoswJFUQUZWafFkI8iNdD+Dvgp3g9igIwVFsudfLw1HgRIANkhBDL8eY3TtfdwCO13/8Y3uR+REr5qhDiEeCfhRCfx1sJbxWwQkr5xJlVT1HOjppYV5SZ/Rh4BDgMvA78A/AveFdo9QPP4l1RNZ0vAdvxAskvgftO9xdLKZ8DPo43dJYBnuDEynMfAfx4a4mkgf/GCzKKMq/UeiKKoijKrKmeiKIoijJrKogoiqIos6aCiKIoijJrKogoiqIos6aCiKIoijJrKogoiqIos6aCiKIoijJrKogoiqIos6aCiKIoijJr/w+8NCooC3KYBAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.histplot(data=frame[frame['churn'] == 1], x='balance', kde=True, hue='country')\n",
    "\n",
    "plt.title('Баланс на момент оттока')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Выводы**\n",
    "\n",
    "Обработали имеющиеся данные: поправили названия, убрали лишние столбцы, заполнили пропуски медианным значением по возрастной группе.\n",
    "\n",
    "Выборка состоит:\n",
    "- 50% клиенты из Фарнции\n",
    "- 25% клиенты из Германии\n",
    "- 25% клиенты из Испании\n",
    "\n",
    "Интересно, что среди оттока немцев нет клиентов с нулевым балансом, а значительная часть клиентов в оттоке как раз имеет нулевой баланс. Меньше всего банк покинуло клиентов из Испании. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Обработаем признаки со страной и полом с помощью дамми-кодирования."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "frame = pd.get_dummies(frame, drop_first=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Также необходимо привести некоторые признаки к единой размрности для более корректной работы моделей.\n",
    "\n",
    "- credit_score\n",
    "- age\n",
    "- tenure\n",
    "- balance\n",
    "- num_of_products\n",
    "- salary"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Исследование задачи"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    0.7963\n",
       "1    0.2037\n",
       "Name: churn, dtype: float64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "frame.churn.value_counts(normalize=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Есть дисбаланс класса, который нужно предсказать, пропорция 80 на 20.\n",
    "\n",
    "Подготовим данные для обучения на простейшей модели и сравним её со случайной, пока без учёта дисбаланса.\n",
    "\n",
    "*Создадим обучающую, тестовую и валидационные выборки в соотношении 3:1:1*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "def post_normalize(frame):\n",
    "    to_norm = ['credit_score', 'age', 'tenure', 'balance', 'num_of_products', 'salary']\n",
    "    scaler = MinMaxScaler()\n",
    "    scaler.fit(frame[to_norm])\n",
    "    frame.loc[:, to_norm] = scaler.transform(frame[to_norm])\n",
    "    \n",
    "    return frame"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Размер тестовой выборки = (2000, 11)\n",
      "Размер валидационной выборки = (2000, 11)\n",
      "Размер обучающей выборки = (6000, 11)\n"
     ]
    }
   ],
   "source": [
    "pd.options.mode.chained_assignment = None\n",
    "\n",
    "target = frame['churn']\n",
    "features = frame.drop('churn', axis=1)\n",
    "\n",
    "def get_samples(target, features, size):\n",
    "    '''\n",
    "    Принимает на вход: целевой признак, отобранные фичи, доля тестовых данных.\n",
    "    \n",
    "    Возвращает: разделённые выборки в указанной пропорции.\n",
    "    '''\n",
    "    features_train, features_valid, target_train, target_valid = train_test_split(\n",
    "        features, target, test_size=size, random_state=42, stratify=target\n",
    "    )\n",
    "        \n",
    "    return post_normalize(features_train), post_normalize(features_valid), target_train, target_valid\n",
    "    \n",
    "features_train, features_valid, target_train, target_valid = get_samples(target, features, 0.4)\n",
    "features_test, features_valid, target_test, target_valid = get_samples(target_valid, features_valid, 0.5)\n",
    "\n",
    "print(f'Размер тестовой выборки = {features_test.shape}')\n",
    "print(f'Размер валидационной выборки = {features_valid.shape}')\n",
    "print(f'Размер обучающей выборки = {features_train.shape}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Для удобства сравнения моделей создадим фрейм:\n",
    "- 'model' - имя модели,\n",
    "- f1_valid'- f1 на валидационной выборке,\n",
    "- 'roc_auc' - ROC-AUC на выборке,\n",
    "- 'balanced' - сбалансирован ли класс."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = pd.DataFrame(columns=[\n",
    "                              'model',\n",
    "                              'f1_valid',\n",
    "                              'roc_auc',\n",
    "                              'balanced'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_actual_auc(model, target_valid, features_valid):\n",
    "    '''\n",
    "    Принимает на вход: обученную модель, валидационную выборку.\n",
    "    \n",
    "    Возвращает: \n",
    "    - площадь, ограниченную рабочей характеристикой приёмника,\n",
    "    - процент верно предсказанных среди положительных значений,\n",
    "    - процент неверно предсказанных среди отрицательных значений,\n",
    "    - пороговые значения.\n",
    "    '''\n",
    "    probabilities = model.predict_proba(features_valid)\n",
    "\n",
    "    fpr, tpr, thresholds = roc_curve(target_valid, probabilities[:,1])\n",
    "    auc_roc = roc_auc_score(target_valid, probabilities[:,1])\n",
    "    \n",
    "    return auc_roc, fpr, tpr, thresholds\n",
    "\n",
    "def show_roc(fpr, tpr, auc_roc):\n",
    "    plt.plot(fpr, tpr, color='darkorange',\n",
    "             label='ROC кривая (area = %0.2f)' % auc_roc)\n",
    "    plt.plot([0, 1], [0, 1], color='navy', linestyle='--')\n",
    "    plt.xlim([-0.1, 1.1])\n",
    "    plt.ylim([-0.1, 1.1])\n",
    "    plt.xlabel('False Positive Rate')\n",
    "    plt.ylabel('True Positive Rate')\n",
    "    plt.title('ROC-кривая')\n",
    "    plt.legend(loc=\"lower right\")\n",
    "    plt.grid(color = 'black')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Модель случайного леса**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_best_random_forest(features_train, target_train, features_valid, target_valid, class_weight='balanced'):\n",
    "    '''\n",
    "    Принимает на вход: обучающую и валидационную выборки, режим корректировки веса классов.\n",
    "    \n",
    "    Возвращает: \n",
    "    - площадь, ограниченную рабочей характеристикой приёмника,\n",
    "    - процент верно предсказанных среди положительных значений,\n",
    "    - процент неверно предсказанных среди отрицательных значений,\n",
    "    - пороговые значения.\n",
    "    '''\n",
    "    f1_dict = {a : [] for a in range(100, 136, 1)}\n",
    "    auc_dict = {a : [] for a in range(100, 136, 1)}\n",
    "    best_f1 = 0\n",
    "\n",
    "    for n_est in f1_dict:\n",
    "        for depth in range(15, 20):\n",
    "            model = RandomForestClassifier(random_state=42, max_depth=depth, n_estimators=n_est, class_weight=class_weight)\n",
    "            model.fit(features_train, target_train)\n",
    "            predictions = model.predict(features_valid)\n",
    "            actual_f1 = f1_score(target_valid, predictions)\n",
    "            \n",
    "            auc_roc, fpr, tpr, thresholds = get_actual_auc(model, target_valid, features_valid)\n",
    "            auc_dict[n_est].append(auc_roc)\n",
    "            \n",
    "            if actual_f1 > best_f1:\n",
    "                best_f1 = actual_f1\n",
    "                best_model = model\n",
    "\n",
    "            f1_dict[n_est].append(actual_f1)\n",
    "    \n",
    "    print(f'Лучшая точность на валидационной выборке: {best_f1:.4f}')\n",
    "    params = best_model.get_params()\n",
    "    print(f\"Параметры лучшей модели: max_depth = {params['max_depth']}, n_estimators = {params['n_estimators']}\")\n",
    "    \n",
    "    return best_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.5941\n",
      "Параметры лучшей модели: max_depth = 15, n_estimators = 120\n"
     ]
    }
   ],
   "source": [
    "best_model = get_best_random_forest(features_train, target_train, features_valid, target_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier',\n",
    "                          'f1_valid':0.5941,\n",
    "                          'roc_auc':0.85,\n",
    "                          'balanced':False},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Логистическая регрессия**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.3105\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def get_logistic_regression(features_train, target_train, features_valid, target_valid):\n",
    "    \n",
    "    best_model = LogisticRegression(random_state=42, solver='liblinear', max_iter=200)\n",
    "    \n",
    "    best_model.fit(features_train, target_train)\n",
    "    predictions = best_model.predict(features_valid)\n",
    "    \n",
    "    print(f'Лучшая точность на валидационной выборке: {f1_score(target_valid, predictions):.4f}')\n",
    "    return best_model\n",
    "    \n",
    "best_logist_model = get_logistic_regression(features_train, target_train, features_valid, target_valid)\n",
    "\n",
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_logist_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'LogisticRegression',\n",
    "                          'f1_valid':0.3105,\n",
    "                          'roc_auc':0.77,\n",
    "                          'balanced':False},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Модель логистической регрессии справилась хуже. Это видно по метрикам AUC и F1/\n",
    "\n",
    "Выяснили, что экземлпяров одного класса больше в 4 раза, чем другого класса. Перейдём к обучению с учётом дисбаланса классов. Также заметно, что модель случайного леса проявила себя лучше на обучении без учета дисбаланса."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Борьба с дисбалансом"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Увеличим выборку меньшего класса в 4 раза."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "'отношение количества элементов положительного и отрицательного класса равно единице.'"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def upsample(features, target, repeat):\n",
    "    features_zeros = features[target == 0]\n",
    "    features_ones = features[target == 1]\n",
    "    target_zeros = target[target == 0]\n",
    "    target_ones = target[target == 1]\n",
    "\n",
    "    features_upsampled = pd.concat([features_zeros] + [features_ones] * repeat)\n",
    "    target_upsampled = pd.concat([target_zeros] + [target_ones] * repeat)\n",
    "    \n",
    "    features_upsampled, target_upsampled = shuffle(\n",
    "        features_upsampled, target_upsampled, random_state=42)\n",
    "    \n",
    "    return features_upsampled, target_upsampled\n",
    "\n",
    "features_train, features_valid, target_train, target_valid = get_samples(target, features, 0.4)\n",
    "features_test, features_valid, target_test, target_valid = get_samples(target_valid, features_valid, 0.5)\n",
    "\n",
    "features_train_upsampled, target_train_upsampled = upsample(features_train, target_train, 3)\n",
    "\n",
    "display(len(features_train_upsampled) / len(target_train_upsampled), 'отношение количества элементов положительного и отрицательного класса равно единице.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.6068\n",
      "Параметры лучшей модели: max_depth = 16, n_estimators = 100\n"
     ]
    }
   ],
   "source": [
    "best_forest_model = get_best_random_forest(features_train_upsampled, target_train_upsampled, features_valid, target_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_forest_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Получилось увеличить точность F1 Метрики. ROC кривая практически такая же. Тем не менее попробуем уменьшить больший класс в 2 раза и увеличить в 2 раза меньший класс. Сохраним модель отдельно."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier',\n",
    "                          'f1_valid':0.6068,\n",
    "                          'roc_auc':0.84,\n",
    "                          'balanced':'Positive x 4'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "def downsample(features, target, fraction):\n",
    "    features_zeros = features[target == 0]\n",
    "    features_ones = features[target == 1]\n",
    "    target_zeros = target[target == 0]\n",
    "    target_ones = target[target == 1]\n",
    "\n",
    "    features_downsampled = pd.concat(\n",
    "        [features_zeros.sample(frac=fraction, random_state=42)] + [features_ones])\n",
    "    target_downsampled = pd.concat(\n",
    "        [target_zeros.sample(frac=fraction, random_state=42)] + [target_ones])\n",
    "    \n",
    "    features_downsampled, target_downsampled = shuffle(\n",
    "        features_downsampled, target_downsampled, random_state=42)\n",
    "    \n",
    "    return features_downsampled, target_downsampled"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "features_train, features_valid, target_train, target_valid = get_samples(target, features, 0.4)\n",
    "features_test, features_valid, target_test, target_valid = get_samples(target_valid, features_valid, 0.5)\n",
    "\n",
    "features_train_downsampled, target_train_downsampled = downsample(features_train, target_train, 0.5)\n",
    "features_train_combined, target_train_combined = upsample(features_train_downsampled, target_train_downsampled, 2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.6013\n",
      "Параметры лучшей модели: max_depth = 17, n_estimators = 128\n"
     ]
    }
   ],
   "source": [
    "best_forest_model = get_best_random_forest(features_train_combined, target_train_combined, features_valid, target_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_forest_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier',\n",
    "                          'f1_valid':0.6013,\n",
    "                          'roc_auc':0.84,\n",
    "                          'balanced':'Negative / 2, Positive x 2'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Попробуем уменьшить больший класс в четыре раза."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.5798\n",
      "Параметры лучшей модели: max_depth = 15, n_estimators = 132\n"
     ]
    }
   ],
   "source": [
    "features_train_downsampled, target_train_downsampled = downsample(features_train, target_train, 0.25)\n",
    "\n",
    "best_forest_model1 = get_best_random_forest(features_train_downsampled, target_train_downsampled, features_valid, target_valid)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_forest_model1, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier',\n",
    "                          'f1_valid':0.5798,\n",
    "                          'roc_auc':0.85,\n",
    "                          'balanced':'Negative / 4'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Лучшая точность была при использорвании комбинированного способа решения проблемы дисбаланса."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Лучшая точность на валидационной выборке: 0.4874\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "best_logist_model = get_logistic_regression(features_train_combined, target_train_combined, features_valid, target_valid)\n",
    "\n",
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_logist_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'LogisticRegression',\n",
    "                          'f1_valid':0.4874,\n",
    "                          'roc_auc':0.77,\n",
    "                          'balanced':'Negative / 2, Positive x 2'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Логистическая регрессия показала себя значительно хуже случайного леса, остановимся на нём."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "С лучшим порогом (0.5000000000000002) значение метрики F1 = 0.6012793176972282\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "probabilities_valid = best_forest_model.predict_proba(features_valid)\n",
    "probabilities_one_valid = probabilities_valid[:, 1]\n",
    "\n",
    "list_threshold = []\n",
    "best_threshold = 0\n",
    "actual_f1 = 0\n",
    "best_f1 = 0\n",
    "max_range = 0.9\n",
    "min_range = 0.2\n",
    "range_threshold = np.arange(min_range, max_range, 0.001)\n",
    "\n",
    "for threshold in range_threshold:\n",
    "    predictions = probabilities_one_valid > threshold\n",
    "    actual_f1 = f1_score(target_valid, predictions)\n",
    "    if actual_f1 > best_f1:\n",
    "        best_f1 = actual_f1\n",
    "        best_threshold = threshold\n",
    "    list_threshold.append(actual_f1)\n",
    "    \n",
    "print(f'С лучшим порогом ({best_threshold}) значение метрики F1 = {best_f1}')\n",
    "\n",
    "\n",
    "plt.plot(range_threshold.tolist(), list_threshold, color='darkorange')\n",
    "plt.plot([best_threshold, best_threshold], [0, 1.05], color='navy', linestyle='--')\n",
    "plt.grid(color='black')\n",
    "plt.xlim([min_range, max_range])\n",
    "plt.ylim([0.3, 1.05])\n",
    "plt.xlabel('Threshold')\n",
    "plt.ylabel('F1')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 = 0.6013\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "probabilities = best_forest_model.predict_proba(features_valid)\n",
    "probabilities_one_valid = probabilities_valid[:, 1]\n",
    "\n",
    "predictions = probabilities_one_valid > best_threshold\n",
    "\n",
    "print(f'F1 = {f1_score(target_valid, predictions):.4f}')\n",
    "\n",
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_forest_model, target_valid, features_valid)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier',\n",
    "                          'f1_valid':0.6013,\n",
    "                          'roc_auc':0.84,\n",
    "                          'balanced':'Negative / 2, Positive x 2, Best Threshold'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Результат удовлетворительный, переходим к тестированию."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Тестирование модели"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 = 0.5893\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "probabilities = best_forest_model.predict_proba(features_test)\n",
    "probabilities_one_valid = probabilities[:, 1]\n",
    "\n",
    "predictions = probabilities_one_valid > best_threshold\n",
    "\n",
    "print(f'F1 = {f1_score(target_test, predictions):.4f}')\n",
    "\n",
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(best_forest_model, target_test, features_test)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = results.append(\n",
    "                         {'model':'RandomForestClassifier (test)',\n",
    "                          'f1_valid':0.5893,\n",
    "                          'roc_auc':0.86,\n",
    "                          'balanced':'Negative / 2, Positive x 2, Best Threshold'},ignore_index=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "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>model</th>\n",
       "      <th>f1_valid</th>\n",
       "      <th>roc_auc</th>\n",
       "      <th>balanced</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>RandomForestClassifier</td>\n",
       "      <td>0.5941</td>\n",
       "      <td>0.85</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>LogisticRegression</td>\n",
       "      <td>0.3105</td>\n",
       "      <td>0.77</td>\n",
       "      <td>False</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>RandomForestClassifier</td>\n",
       "      <td>0.6068</td>\n",
       "      <td>0.84</td>\n",
       "      <td>Positive x 4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>RandomForestClassifier</td>\n",
       "      <td>0.6013</td>\n",
       "      <td>0.84</td>\n",
       "      <td>Negative / 2, Positive x 2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>RandomForestClassifier</td>\n",
       "      <td>0.5798</td>\n",
       "      <td>0.85</td>\n",
       "      <td>Negative / 4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>LogisticRegression</td>\n",
       "      <td>0.4874</td>\n",
       "      <td>0.77</td>\n",
       "      <td>Negative / 2, Positive x 2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>RandomForestClassifier</td>\n",
       "      <td>0.6013</td>\n",
       "      <td>0.84</td>\n",
       "      <td>Negative / 2, Positive x 2, Best Threshold</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>RandomForestClassifier (test)</td>\n",
       "      <td>0.5893</td>\n",
       "      <td>0.86</td>\n",
       "      <td>Negative / 2, Positive x 2, Best Threshold</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                           model  f1_valid  roc_auc  \\\n",
       "0         RandomForestClassifier    0.5941     0.85   \n",
       "1             LogisticRegression    0.3105     0.77   \n",
       "2         RandomForestClassifier    0.6068     0.84   \n",
       "3         RandomForestClassifier    0.6013     0.84   \n",
       "4         RandomForestClassifier    0.5798     0.85   \n",
       "5             LogisticRegression    0.4874     0.77   \n",
       "6         RandomForestClassifier    0.6013     0.84   \n",
       "7  RandomForestClassifier (test)    0.5893     0.86   \n",
       "\n",
       "                                     balanced  \n",
       "0                                       False  \n",
       "1                                       False  \n",
       "2                                Positive x 4  \n",
       "3                  Negative / 2, Positive x 2  \n",
       "4                                Negative / 4  \n",
       "5                  Negative / 2, Positive x 2  \n",
       "6  Negative / 2, Positive x 2, Best Threshold  \n",
       "7  Negative / 2, Positive x 2, Best Threshold  "
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 8000 entries, 6851 to 7634\n",
      "Data columns (total 11 columns):\n",
      "credit_score       8000 non-null float64\n",
      "age                8000 non-null float64\n",
      "tenure             8000 non-null float64\n",
      "balance            8000 non-null float64\n",
      "num_of_products    8000 non-null float64\n",
      "has_cr_card        8000 non-null int64\n",
      "is_active          8000 non-null int64\n",
      "salary             8000 non-null float64\n",
      "country_Germany    8000 non-null uint8\n",
      "country_Spain      8000 non-null uint8\n",
      "gender_Male        8000 non-null uint8\n",
      "dtypes: float64(6), int64(2), uint8(3)\n",
      "memory usage: 585.9 KB\n"
     ]
    }
   ],
   "source": [
    "features_all = features_train.append(features_valid)\n",
    "features_all.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_all = RandomForestClassifier(random_state=42, max_depth=17, n_estimators=128, class_weight='balanced')\n",
    "model_all.fit(features_train_combined, target_train_combined)\n",
    "predictions = model_all.predict(features_valid)\n",
    "actual_f1 = f1_score(target_valid, predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "F1 = 0.5893\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "probabilities = model_all.predict_proba(features_test)\n",
    "probabilities_all = probabilities[:, 1]\n",
    "\n",
    "predictions = probabilities_all > best_threshold\n",
    "\n",
    "print(f'F1 = {f1_score(target_test, predictions):.4f}')\n",
    "\n",
    "auc_roc, fpr, tpr, thresholds = get_actual_auc(model_all, target_test, features_test)\n",
    "show_roc(fpr, tpr, auc_roc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "feat_importances1 = pd.Series(best_forest_model.feature_importances_, index=features_train_combined.columns)\n",
    "feat_importances1.nlargest(20).plot(kind='barh')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Общий вывод**\n",
    "\n",
    "Наш итоговый результат: F1 = 0.5893, AUC = 0.86\n",
    "\n",
    "Мы смогли добиться этого, уменьшив больший класс вдвое, и увеличив меньший класс также вдвое. Подобрали оптимальное значение порога. Наилучим образом отработала модель случайного леса.\n",
    "\n",
    "Если учитывать показатели метрики F-меры, то модель, с учетом балансировки, делает прогнозы лучше, чем без учета. На метрику ROC-AUC балансировка не повлияла.\n",
    "\n",
    "Для более адекватной оценки модели, необходим отдельный тестовый набор данных. Службе маркетинга можем рекомендовать взять данную модель для пронозирования оттока на тестирование на реальных данных. "
   ]
  }
 ],
 "metadata": {
  "ExecuteTimeLog": [
   {
    "duration": 1970,
    "start_time": "2022-01-18T15:35:17.300Z"
   },
   {
    "duration": 2538,
    "start_time": "2022-01-18T15:35:30.538Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-18T15:44:23.363Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-18T15:44:46.281Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-18T15:45:05.816Z"
   },
   {
    "duration": 272,
    "start_time": "2022-01-18T15:51:15.859Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-18T15:51:20.271Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-18T15:58:55.147Z"
   },
   {
    "duration": 48,
    "start_time": "2022-01-18T15:59:33.467Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-18T16:00:35.154Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-18T17:03:03.751Z"
   },
   {
    "duration": 264,
    "start_time": "2022-01-18T17:09:04.122Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-18T17:10:16.697Z"
   },
   {
    "duration": 30,
    "start_time": "2022-01-18T17:10:16.707Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-18T17:10:16.739Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-18T17:10:16.752Z"
   },
   {
    "duration": 67,
    "start_time": "2022-01-18T17:10:16.767Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-18T17:10:16.835Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-18T17:10:30.665Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-18T17:10:39.434Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-18T17:10:41.354Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-18T17:12:00.706Z"
   },
   {
    "duration": 268,
    "start_time": "2022-01-18T17:14:46.997Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-18T17:14:55.157Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-18T17:16:20.806Z"
   },
   {
    "duration": 21,
    "start_time": "2022-01-18T17:17:24.315Z"
   },
   {
    "duration": 1912,
    "start_time": "2022-01-19T10:39:50.076Z"
   },
   {
    "duration": 39,
    "start_time": "2022-01-19T10:39:51.991Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-19T10:39:52.033Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-19T10:39:52.815Z"
   },
   {
    "duration": 66,
    "start_time": "2022-01-19T10:39:55.366Z"
   },
   {
    "duration": 18,
    "start_time": "2022-01-19T10:39:57.321Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-19T10:39:58.528Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-19T10:39:59.237Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-19T10:40:00.437Z"
   },
   {
    "duration": 22,
    "start_time": "2022-01-19T10:40:00.902Z"
   },
   {
    "duration": 9857,
    "start_time": "2022-01-19T10:40:23.785Z"
   },
   {
    "duration": 10554,
    "start_time": "2022-01-19T10:41:06.589Z"
   },
   {
    "duration": 10494,
    "start_time": "2022-01-19T10:41:41.102Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-19T10:42:28.582Z"
   },
   {
    "duration": 19,
    "start_time": "2022-01-19T10:46:20.820Z"
   },
   {
    "duration": 394,
    "start_time": "2022-01-19T10:49:32.981Z"
   },
   {
    "duration": 10125,
    "start_time": "2022-01-19T10:49:48.433Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-19T10:50:07.802Z"
   },
   {
    "duration": 374,
    "start_time": "2022-01-19T10:55:42.045Z"
   },
   {
    "duration": 387,
    "start_time": "2022-01-19T11:11:44.365Z"
   },
   {
    "duration": 359,
    "start_time": "2022-01-19T11:11:53.627Z"
   },
   {
    "duration": 386,
    "start_time": "2022-01-19T11:12:27.568Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-19T13:56:37.394Z"
   },
   {
    "duration": 1734,
    "start_time": "2022-01-19T13:56:39.960Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-19T13:56:41.696Z"
   },
   {
    "duration": 19,
    "start_time": "2022-01-19T13:56:43.287Z"
   },
   {
    "duration": 63,
    "start_time": "2022-01-19T13:56:46.924Z"
   },
   {
    "duration": 3238,
    "start_time": "2022-01-19T13:56:48.544Z"
   },
   {
    "duration": 6988,
    "start_time": "2022-01-19T13:56:51.785Z"
   },
   {
    "duration": 2997,
    "start_time": "2022-01-19T13:56:58.776Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-19T13:57:01.775Z"
   },
   {
    "duration": 36,
    "start_time": "2022-01-19T13:57:01.793Z"
   },
   {
    "duration": 479,
    "start_time": "2022-01-19T14:00:00.628Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-19T14:00:13.830Z"
   },
   {
    "duration": 21,
    "start_time": "2022-01-19T14:04:11.130Z"
   },
   {
    "duration": 66,
    "start_time": "2022-01-19T14:09:18.060Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-19T14:10:12.228Z"
   },
   {
    "duration": 1247,
    "start_time": "2022-01-19T14:27:13.832Z"
   },
   {
    "duration": 1443,
    "start_time": "2022-01-19T14:27:30.904Z"
   },
   {
    "duration": 1518,
    "start_time": "2022-01-19T14:27:47.186Z"
   },
   {
    "duration": 43,
    "start_time": "2022-01-19T14:38:26.173Z"
   },
   {
    "duration": 21,
    "start_time": "2022-01-19T14:43:02.855Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-19T14:43:28.445Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-19T14:44:14.660Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-19T14:44:29.026Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-19T14:44:44.890Z"
   },
   {
    "duration": 19,
    "start_time": "2022-01-19T14:51:24.997Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-19T14:51:39.675Z"
   },
   {
    "duration": 251,
    "start_time": "2022-01-19T14:53:25.173Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-19T14:54:08.983Z"
   },
   {
    "duration": 2025,
    "start_time": "2022-01-19T14:54:13.013Z"
   },
   {
    "duration": 393,
    "start_time": "2022-01-19T14:54:19.972Z"
   },
   {
    "duration": 590,
    "start_time": "2022-01-19T14:56:36.221Z"
   },
   {
    "duration": 398,
    "start_time": "2022-01-19T15:03:15.761Z"
   },
   {
    "duration": 4813,
    "start_time": "2022-01-19T15:03:49.505Z"
   },
   {
    "duration": 371,
    "start_time": "2022-01-19T15:04:07.491Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-19T15:04:32.524Z"
   },
   {
    "duration": 39,
    "start_time": "2022-01-19T15:04:32.868Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-19T15:04:33.294Z"
   },
   {
    "duration": 19,
    "start_time": "2022-01-19T15:04:33.990Z"
   },
   {
    "duration": 64,
    "start_time": "2022-01-19T15:04:34.603Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-19T15:04:35.718Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-19T15:04:37.343Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-19T15:04:37.421Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-19T15:04:38.315Z"
   },
   {
    "duration": 21,
    "start_time": "2022-01-19T15:04:39.481Z"
   },
   {
    "duration": 42,
    "start_time": "2022-01-19T15:04:49.628Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-19T15:04:53.425Z"
   },
   {
    "duration": 2061,
    "start_time": "2022-01-19T15:04:57.763Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-19T15:05:03.171Z"
   },
   {
    "duration": 318,
    "start_time": "2022-01-19T15:05:03.649Z"
   },
   {
    "duration": 421,
    "start_time": "2022-01-19T15:05:04.598Z"
   },
   {
    "duration": 1527,
    "start_time": "2022-01-19T15:06:42.891Z"
   },
   {
    "duration": 41,
    "start_time": "2022-01-19T15:06:44.421Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-19T15:06:44.468Z"
   },
   {
    "duration": 25,
    "start_time": "2022-01-19T15:06:44.485Z"
   },
   {
    "duration": 106,
    "start_time": "2022-01-19T15:06:44.512Z"
   },
   {
    "duration": 52,
    "start_time": "2022-01-19T15:06:44.621Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-19T15:06:44.679Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-19T15:06:44.689Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-19T15:06:44.699Z"
   },
   {
    "duration": 67,
    "start_time": "2022-01-19T15:06:44.707Z"
   },
   {
    "duration": 42,
    "start_time": "2022-01-19T15:06:44.776Z"
   },
   {
    "duration": 40,
    "start_time": "2022-01-19T15:06:44.820Z"
   },
   {
    "duration": 2114,
    "start_time": "2022-01-19T15:06:44.863Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-19T15:06:46.979Z"
   },
   {
    "duration": 382,
    "start_time": "2022-01-19T15:06:46.996Z"
   },
   {
    "duration": 506,
    "start_time": "2022-01-19T15:06:47.381Z"
   },
   {
    "duration": 929,
    "start_time": "2022-01-19T15:06:46.960Z"
   },
   {
    "duration": 30,
    "start_time": "2022-01-19T15:08:32.999Z"
   },
   {
    "duration": 855,
    "start_time": "2022-01-19T15:08:47.008Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-19T15:17:18.742Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-19T15:20:18.759Z"
   },
   {
    "duration": 432,
    "start_time": "2022-01-19T15:21:31.089Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-19T15:21:36.273Z"
   },
   {
    "duration": 25,
    "start_time": "2022-01-19T15:35:48.674Z"
   },
   {
    "duration": 22,
    "start_time": "2022-01-19T15:37:59.357Z"
   },
   {
    "duration": 26,
    "start_time": "2022-01-19T15:38:29.177Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-19T15:40:08.659Z"
   },
   {
    "duration": 1915,
    "start_time": "2022-01-19T17:29:10.273Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-19T17:29:19.612Z"
   },
   {
    "duration": 1809,
    "start_time": "2022-01-19T17:29:23.922Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-19T17:29:59.556Z"
   },
   {
    "duration": 2571,
    "start_time": "2022-01-19T17:30:05.901Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-19T17:30:27.899Z"
   },
   {
    "duration": 18888,
    "start_time": "2022-01-19T17:30:33.294Z"
   },
   {
    "duration": 128427,
    "start_time": "2022-01-19T17:33:18.746Z"
   },
   {
    "duration": 480,
    "start_time": "2022-01-19T17:43:57.554Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-19T17:46:20.859Z"
   },
   {
    "duration": 498,
    "start_time": "2022-01-19T17:46:25.262Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-19T17:47:22.864Z"
   },
   {
    "duration": 491,
    "start_time": "2022-01-19T17:47:27.420Z"
   },
   {
    "duration": 1252,
    "start_time": "2022-01-20T14:26:09.201Z"
   },
   {
    "duration": 28,
    "start_time": "2022-01-20T14:26:11.114Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T14:26:11.846Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-20T14:26:12.850Z"
   },
   {
    "duration": 44,
    "start_time": "2022-01-20T14:26:13.668Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-20T14:26:15.077Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T14:26:17.719Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T14:26:17.725Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T14:26:18.418Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-20T14:26:21.877Z"
   },
   {
    "duration": 30,
    "start_time": "2022-01-20T14:26:22.181Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-20T14:26:25.136Z"
   },
   {
    "duration": 1480,
    "start_time": "2022-01-20T14:26:25.149Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-20T14:26:30.476Z"
   },
   {
    "duration": 260,
    "start_time": "2022-01-20T14:26:31.235Z"
   },
   {
    "duration": 406,
    "start_time": "2022-01-20T14:26:34.307Z"
   },
   {
    "duration": 83,
    "start_time": "2022-01-20T14:28:06.623Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T14:28:17.853Z"
   },
   {
    "duration": 100615,
    "start_time": "2022-01-20T14:28:34.298Z"
   },
   {
    "duration": 663,
    "start_time": "2022-01-20T14:30:14.915Z"
   },
   {
    "duration": 2479,
    "start_time": "2022-01-20T14:32:29.926Z"
   },
   {
    "duration": 1167,
    "start_time": "2022-01-20T14:33:20.731Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-20T14:33:21.900Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-20T14:33:21.931Z"
   },
   {
    "duration": 23,
    "start_time": "2022-01-20T14:33:21.947Z"
   },
   {
    "duration": 55,
    "start_time": "2022-01-20T14:33:21.972Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-20T14:33:22.028Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T14:33:22.041Z"
   },
   {
    "duration": 24,
    "start_time": "2022-01-20T14:33:22.048Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-20T14:33:22.074Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-20T14:33:22.088Z"
   },
   {
    "duration": 28,
    "start_time": "2022-01-20T14:33:22.118Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T14:33:22.147Z"
   },
   {
    "duration": 1470,
    "start_time": "2022-01-20T14:33:22.159Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T14:33:23.631Z"
   },
   {
    "duration": 273,
    "start_time": "2022-01-20T14:33:23.641Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-20T14:33:23.916Z"
   },
   {
    "duration": 593,
    "start_time": "2022-01-20T14:33:23.920Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-20T14:33:24.515Z"
   },
   {
    "duration": 31,
    "start_time": "2022-01-20T14:33:24.532Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-20T14:33:24.565Z"
   },
   {
    "duration": 37,
    "start_time": "2022-01-20T14:33:24.575Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T14:33:24.614Z"
   },
   {
    "duration": 86814,
    "start_time": "2022-01-20T14:33:24.628Z"
   },
   {
    "duration": 352,
    "start_time": "2022-01-20T14:34:51.443Z"
   },
   {
    "duration": 530,
    "start_time": "2022-01-20T14:38:50.460Z"
   },
   {
    "duration": 384,
    "start_time": "2022-01-20T15:00:06.198Z"
   },
   {
    "duration": 491,
    "start_time": "2022-01-20T15:02:09.614Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T15:02:52.532Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-20T15:03:00.308Z"
   },
   {
    "duration": 26,
    "start_time": "2022-01-20T15:05:05.722Z"
   },
   {
    "duration": 144,
    "start_time": "2022-01-20T15:05:23.853Z"
   },
   {
    "duration": 26,
    "start_time": "2022-01-20T15:05:52.750Z"
   },
   {
    "duration": 87623,
    "start_time": "2022-01-20T15:07:11.353Z"
   },
   {
    "duration": -375,
    "start_time": "2022-01-20T15:08:39.353Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-20T15:08:42.896Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T15:08:55.580Z"
   },
   {
    "duration": 92090,
    "start_time": "2022-01-20T15:09:12.661Z"
   },
   {
    "duration": 122845,
    "start_time": "2022-01-20T15:12:28.399Z"
   },
   {
    "duration": 339,
    "start_time": "2022-01-20T15:21:47.541Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T15:24:07.855Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-20T15:25:31.938Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T15:26:24.125Z"
   },
   {
    "duration": 75677,
    "start_time": "2022-01-20T15:26:52.826Z"
   },
   {
    "duration": 344,
    "start_time": "2022-01-20T15:28:45.517Z"
   },
   {
    "duration": 48416,
    "start_time": "2022-01-20T15:30:30.354Z"
   },
   {
    "duration": 511,
    "start_time": "2022-01-20T15:37:06.681Z"
   },
   {
    "duration": 82,
    "start_time": "2022-01-20T15:37:52.978Z"
   },
   {
    "duration": 1314,
    "start_time": "2022-01-20T15:39:31.326Z"
   },
   {
    "duration": 1301,
    "start_time": "2022-01-20T15:40:51.443Z"
   },
   {
    "duration": 387,
    "start_time": "2022-01-20T15:42:00.075Z"
   },
   {
    "duration": 90,
    "start_time": "2022-01-20T15:43:23.810Z"
   },
   {
    "duration": 371,
    "start_time": "2022-01-20T15:43:37.212Z"
   },
   {
    "duration": 297,
    "start_time": "2022-01-20T17:34:20.071Z"
   },
   {
    "duration": 272,
    "start_time": "2022-01-20T17:34:26.855Z"
   },
   {
    "duration": 1162,
    "start_time": "2022-01-20T17:34:33.620Z"
   },
   {
    "duration": 1302,
    "start_time": "2022-01-20T17:38:04.604Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-20T17:38:05.908Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T17:38:05.939Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-20T17:38:05.950Z"
   },
   {
    "duration": 60,
    "start_time": "2022-01-20T17:38:05.968Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-20T17:38:06.030Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T17:38:06.045Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T17:38:06.052Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-20T17:38:06.062Z"
   },
   {
    "duration": 41,
    "start_time": "2022-01-20T17:38:06.072Z"
   },
   {
    "duration": 545,
    "start_time": "2022-01-20T17:38:06.115Z"
   },
   {
    "duration": -84,
    "start_time": "2022-01-20T17:38:06.746Z"
   },
   {
    "duration": -85,
    "start_time": "2022-01-20T17:38:06.749Z"
   },
   {
    "duration": -87,
    "start_time": "2022-01-20T17:38:06.752Z"
   },
   {
    "duration": -91,
    "start_time": "2022-01-20T17:38:06.757Z"
   },
   {
    "duration": -94,
    "start_time": "2022-01-20T17:38:06.761Z"
   },
   {
    "duration": -96,
    "start_time": "2022-01-20T17:38:06.764Z"
   },
   {
    "duration": -99,
    "start_time": "2022-01-20T17:38:06.768Z"
   },
   {
    "duration": -100,
    "start_time": "2022-01-20T17:38:06.771Z"
   },
   {
    "duration": -104,
    "start_time": "2022-01-20T17:38:06.776Z"
   },
   {
    "duration": -114,
    "start_time": "2022-01-20T17:38:06.787Z"
   },
   {
    "duration": -117,
    "start_time": "2022-01-20T17:38:06.791Z"
   },
   {
    "duration": -119,
    "start_time": "2022-01-20T17:38:06.794Z"
   },
   {
    "duration": -121,
    "start_time": "2022-01-20T17:38:06.797Z"
   },
   {
    "duration": -122,
    "start_time": "2022-01-20T17:38:06.799Z"
   },
   {
    "duration": -124,
    "start_time": "2022-01-20T17:38:06.802Z"
   },
   {
    "duration": -127,
    "start_time": "2022-01-20T17:38:06.806Z"
   },
   {
    "duration": -129,
    "start_time": "2022-01-20T17:38:06.809Z"
   },
   {
    "duration": -131,
    "start_time": "2022-01-20T17:38:06.812Z"
   },
   {
    "duration": -137,
    "start_time": "2022-01-20T17:38:06.819Z"
   },
   {
    "duration": -139,
    "start_time": "2022-01-20T17:38:06.822Z"
   },
   {
    "duration": -143,
    "start_time": "2022-01-20T17:38:06.827Z"
   },
   {
    "duration": -147,
    "start_time": "2022-01-20T17:38:06.832Z"
   },
   {
    "duration": -150,
    "start_time": "2022-01-20T17:38:06.836Z"
   },
   {
    "duration": -152,
    "start_time": "2022-01-20T17:38:06.840Z"
   },
   {
    "duration": -156,
    "start_time": "2022-01-20T17:38:06.844Z"
   },
   {
    "duration": -159,
    "start_time": "2022-01-20T17:38:06.848Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T17:43:22.360Z"
   },
   {
    "duration": 42,
    "start_time": "2022-01-20T17:43:22.372Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-20T17:43:22.417Z"
   },
   {
    "duration": 17,
    "start_time": "2022-01-20T17:43:22.431Z"
   },
   {
    "duration": 90,
    "start_time": "2022-01-20T17:43:22.450Z"
   },
   {
    "duration": 18,
    "start_time": "2022-01-20T17:43:22.542Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T17:43:22.562Z"
   },
   {
    "duration": 32,
    "start_time": "2022-01-20T17:43:22.571Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T17:43:22.606Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-20T17:43:22.618Z"
   },
   {
    "duration": 57,
    "start_time": "2022-01-20T17:43:22.641Z"
   },
   {
    "duration": 24,
    "start_time": "2022-01-20T17:43:22.701Z"
   },
   {
    "duration": 1617,
    "start_time": "2022-01-20T17:43:22.728Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T17:43:24.347Z"
   },
   {
    "duration": 235,
    "start_time": "2022-01-20T17:43:24.359Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-20T17:43:24.598Z"
   },
   {
    "duration": 614,
    "start_time": "2022-01-20T17:43:24.602Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-20T17:43:25.219Z"
   },
   {
    "duration": 31,
    "start_time": "2022-01-20T17:43:25.237Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T17:43:25.270Z"
   },
   {
    "duration": 43,
    "start_time": "2022-01-20T17:43:25.277Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-20T17:43:25.322Z"
   },
   {
    "duration": 88138,
    "start_time": "2022-01-20T17:43:25.338Z"
   },
   {
    "duration": 376,
    "start_time": "2022-01-20T17:44:53.478Z"
   },
   {
    "duration": 539,
    "start_time": "2022-01-20T17:44:53.856Z"
   },
   {
    "duration": 26,
    "start_time": "2022-01-20T17:44:54.398Z"
   },
   {
    "duration": 116414,
    "start_time": "2022-01-20T17:44:54.425Z"
   },
   {
    "duration": 340,
    "start_time": "2022-01-20T17:46:50.841Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T17:46:51.182Z"
   },
   {
    "duration": 39,
    "start_time": "2022-01-20T17:46:51.188Z"
   },
   {
    "duration": 75709,
    "start_time": "2022-01-20T17:46:51.228Z"
   },
   {
    "duration": 354,
    "start_time": "2022-01-20T17:48:06.938Z"
   },
   {
    "duration": 47527,
    "start_time": "2022-01-20T17:48:07.293Z"
   },
   {
    "duration": 545,
    "start_time": "2022-01-20T17:48:54.821Z"
   },
   {
    "duration": 1300,
    "start_time": "2022-01-20T17:48:55.367Z"
   },
   {
    "duration": 394,
    "start_time": "2022-01-20T17:48:56.669Z"
   },
   {
    "duration": 387,
    "start_time": "2022-01-20T17:48:57.065Z"
   },
   {
    "duration": 1177,
    "start_time": "2022-01-20T18:10:39.517Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-20T18:10:40.696Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T18:10:40.728Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-20T18:10:40.739Z"
   },
   {
    "duration": 56,
    "start_time": "2022-01-20T18:10:40.757Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-20T18:10:40.815Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-20T18:10:40.828Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T18:10:40.833Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-20T18:10:40.842Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-20T18:10:43.857Z"
   },
   {
    "duration": 520,
    "start_time": "2022-01-20T18:10:45.644Z"
   },
   {
    "duration": 39,
    "start_time": "2022-01-20T18:11:24.559Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-20T18:21:19.118Z"
   },
   {
    "duration": 45,
    "start_time": "2022-01-20T18:21:20.550Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-20T18:21:21.928Z"
   },
   {
    "duration": 1501,
    "start_time": "2022-01-20T18:21:28.392Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-20T18:21:32.358Z"
   },
   {
    "duration": 293,
    "start_time": "2022-01-20T18:21:32.895Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-20T18:21:36.713Z"
   },
   {
    "duration": 576,
    "start_time": "2022-01-20T18:21:38.681Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-20T18:21:43.972Z"
   },
   {
    "duration": 17,
    "start_time": "2022-01-20T18:21:46.581Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T18:38:55.950Z"
   },
   {
    "duration": 19,
    "start_time": "2022-01-20T18:38:59.062Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-20T19:02:08.958Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T19:08:04.838Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-20T19:16:37.759Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T19:25:02.228Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T19:31:02.251Z"
   },
   {
    "duration": 146063,
    "start_time": "2022-01-20T19:33:08.386Z"
   },
   {
    "duration": 341,
    "start_time": "2022-01-20T19:35:34.451Z"
   },
   {
    "duration": 255,
    "start_time": "2022-01-20T20:53:36.775Z"
   },
   {
    "duration": 275,
    "start_time": "2022-01-20T20:58:36.416Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T20:59:02.755Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-20T20:59:12.912Z"
   },
   {
    "duration": 498,
    "start_time": "2022-01-20T21:02:06.455Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T21:02:25.395Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-20T21:02:34.705Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T21:02:58.705Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T21:03:02.095Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-20T21:33:27.026Z"
   },
   {
    "duration": 26,
    "start_time": "2022-01-20T21:48:28.636Z"
   },
   {
    "duration": 195531,
    "start_time": "2022-01-20T21:48:31.824Z"
   },
   {
    "duration": 338,
    "start_time": "2022-01-20T21:52:38.570Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T21:55:11.636Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T21:55:16.136Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-20T21:55:25.257Z"
   },
   {
    "duration": 128440,
    "start_time": "2022-01-20T21:55:37.669Z"
   },
   {
    "duration": 326,
    "start_time": "2022-01-20T21:58:13.868Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T21:59:12.865Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T22:03:29.487Z"
   },
   {
    "duration": 278,
    "start_time": "2022-01-20T22:04:33.108Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-20T22:04:53.430Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-20T22:05:12.274Z"
   },
   {
    "duration": 79809,
    "start_time": "2022-01-20T22:05:33.848Z"
   },
   {
    "duration": 493,
    "start_time": "2022-01-20T22:08:19.296Z"
   },
   {
    "duration": 326,
    "start_time": "2022-01-20T22:10:27.600Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-20T22:11:47.004Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T22:13:07.729Z"
   },
   {
    "duration": 1295,
    "start_time": "2022-01-20T22:13:18.220Z"
   },
   {
    "duration": 396,
    "start_time": "2022-01-20T22:13:35.871Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-20T22:14:35.872Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T22:15:08.146Z"
   },
   {
    "duration": 125234,
    "start_time": "2022-01-20T22:17:10.656Z"
   },
   {
    "duration": 364,
    "start_time": "2022-01-20T22:19:22.530Z"
   },
   {
    "duration": 1267,
    "start_time": "2022-01-20T22:19:45.924Z"
   },
   {
    "duration": 393,
    "start_time": "2022-01-20T22:20:00.389Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T22:22:58.379Z"
   },
   {
    "duration": 384,
    "start_time": "2022-01-20T22:23:05.343Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-20T22:23:55.739Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T22:24:11.234Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-20T22:24:48.341Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-20T22:25:01.709Z"
   },
   {
    "duration": 1207,
    "start_time": "2022-01-21T06:52:46.178Z"
   },
   {
    "duration": 28,
    "start_time": "2022-01-21T06:52:47.386Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-21T06:52:47.416Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-21T06:52:48.725Z"
   },
   {
    "duration": 36,
    "start_time": "2022-01-21T06:52:49.363Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-21T06:53:42.441Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-21T06:53:44.067Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-21T06:54:05.387Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-21T06:54:06.018Z"
   },
   {
    "duration": 33,
    "start_time": "2022-01-21T06:54:23.315Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-21T06:54:31.867Z"
   },
   {
    "duration": 24,
    "start_time": "2022-01-21T06:54:34.791Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-21T06:54:36.063Z"
   },
   {
    "duration": 1099,
    "start_time": "2022-01-21T06:54:45.056Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-21T06:54:47.960Z"
   },
   {
    "duration": 226,
    "start_time": "2022-01-21T06:54:48.219Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-21T06:54:48.962Z"
   },
   {
    "duration": 345,
    "start_time": "2022-01-21T06:54:49.495Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-21T07:00:27.345Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-21T07:00:29.143Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:02:21.154Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-21T07:02:22.942Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:03:51.108Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-21T07:03:51.976Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-21T07:04:01.700Z"
   },
   {
    "duration": 131272,
    "start_time": "2022-01-21T07:04:02.373Z"
   },
   {
    "duration": 282,
    "start_time": "2022-01-21T07:06:13.647Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-21T07:06:13.931Z"
   },
   {
    "duration": 404,
    "start_time": "2022-01-21T07:06:22.161Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-21T07:06:23.054Z"
   },
   {
    "duration": 23,
    "start_time": "2022-01-21T07:06:33.240Z"
   },
   {
    "duration": 179122,
    "start_time": "2022-01-21T07:06:54.639Z"
   },
   {
    "duration": 283,
    "start_time": "2022-01-21T07:09:53.763Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-21T07:09:54.048Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-21T07:09:54.057Z"
   },
   {
    "duration": 35,
    "start_time": "2022-01-21T07:09:54.069Z"
   },
   {
    "duration": 117151,
    "start_time": "2022-01-21T07:09:54.106Z"
   },
   {
    "duration": 301,
    "start_time": "2022-01-21T07:11:51.259Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:11:51.562Z"
   },
   {
    "duration": 75662,
    "start_time": "2022-01-21T07:11:51.570Z"
   },
   {
    "duration": 273,
    "start_time": "2022-01-21T07:13:07.234Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-21T07:13:07.509Z"
   },
   {
    "duration": 434,
    "start_time": "2022-01-21T07:13:07.516Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:13:07.951Z"
   },
   {
    "duration": 1143,
    "start_time": "2022-01-21T07:13:07.959Z"
   },
   {
    "duration": 406,
    "start_time": "2022-01-21T07:13:09.103Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:13:09.510Z"
   },
   {
    "duration": 310,
    "start_time": "2022-01-21T07:13:34.373Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-21T07:13:35.094Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-21T07:13:35.390Z"
   },
   {
    "duration": 1176,
    "start_time": "2022-01-22T11:47:14.696Z"
   },
   {
    "duration": 28,
    "start_time": "2022-01-22T11:47:17.803Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-22T11:47:19.110Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-22T11:47:23.158Z"
   },
   {
    "duration": 35,
    "start_time": "2022-01-22T11:47:26.613Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T11:47:37.223Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T11:47:38.820Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T11:47:46.327Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T11:47:50.085Z"
   },
   {
    "duration": 33,
    "start_time": "2022-01-22T11:47:51.934Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-22T11:47:58.083Z"
   },
   {
    "duration": 452,
    "start_time": "2022-01-22T11:49:47.691Z"
   },
   {
    "duration": 339,
    "start_time": "2022-01-22T11:49:56.886Z"
   },
   {
    "duration": 369,
    "start_time": "2022-01-22T11:50:44.055Z"
   },
   {
    "duration": 18,
    "start_time": "2022-01-22T11:50:48.965Z"
   },
   {
    "duration": 1166,
    "start_time": "2022-01-22T11:51:39.005Z"
   },
   {
    "duration": 29,
    "start_time": "2022-01-22T11:51:40.173Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-22T11:51:40.204Z"
   },
   {
    "duration": 18,
    "start_time": "2022-01-22T11:51:40.215Z"
   },
   {
    "duration": 34,
    "start_time": "2022-01-22T11:51:40.235Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-22T11:51:40.270Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T11:51:40.281Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T11:51:40.286Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T11:51:40.293Z"
   },
   {
    "duration": 50,
    "start_time": "2022-01-22T11:51:40.299Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T11:51:40.351Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-22T11:51:45.982Z"
   },
   {
    "duration": 17,
    "start_time": "2022-01-22T11:51:48.429Z"
   },
   {
    "duration": 17,
    "start_time": "2022-01-22T11:51:59.576Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T11:52:47.333Z"
   },
   {
    "duration": 24,
    "start_time": "2022-01-22T11:53:00.772Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T11:53:18.702Z"
   },
   {
    "duration": 1341,
    "start_time": "2022-01-22T11:53:23.629Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-22T11:53:33.057Z"
   },
   {
    "duration": 157,
    "start_time": "2022-01-22T11:53:33.724Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-22T11:53:37.467Z"
   },
   {
    "duration": 441,
    "start_time": "2022-01-22T11:53:38.706Z"
   },
   {
    "duration": 3304,
    "start_time": "2022-01-22T11:53:44.359Z"
   },
   {
    "duration": 1208,
    "start_time": "2022-01-22T11:55:58.400Z"
   },
   {
    "duration": 32,
    "start_time": "2022-01-22T11:55:59.610Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-22T11:55:59.644Z"
   },
   {
    "duration": 17,
    "start_time": "2022-01-22T11:55:59.653Z"
   },
   {
    "duration": 33,
    "start_time": "2022-01-22T11:55:59.671Z"
   },
   {
    "duration": 30,
    "start_time": "2022-01-22T11:55:59.706Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T11:55:59.738Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T11:55:59.744Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T11:55:59.751Z"
   },
   {
    "duration": 32,
    "start_time": "2022-01-22T11:55:59.757Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T11:55:59.790Z"
   },
   {
    "duration": 31,
    "start_time": "2022-01-22T11:55:59.802Z"
   },
   {
    "duration": 23,
    "start_time": "2022-01-22T11:55:59.835Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-22T11:55:59.859Z"
   },
   {
    "duration": 1131,
    "start_time": "2022-01-22T11:55:59.869Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-22T11:56:01.002Z"
   },
   {
    "duration": 236,
    "start_time": "2022-01-22T11:56:01.012Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-22T11:56:01.250Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-22T11:56:03.718Z"
   },
   {
    "duration": 449,
    "start_time": "2022-01-22T11:56:04.363Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T11:56:22.984Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-22T11:56:37.334Z"
   },
   {
    "duration": 1271,
    "start_time": "2022-01-22T12:08:50.601Z"
   },
   {
    "duration": 28,
    "start_time": "2022-01-22T12:08:51.873Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-22T12:08:51.903Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-22T12:08:51.912Z"
   },
   {
    "duration": 35,
    "start_time": "2022-01-22T12:08:51.928Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T12:08:51.965Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T12:08:51.977Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T12:08:51.984Z"
   },
   {
    "duration": 37,
    "start_time": "2022-01-22T12:08:51.991Z"
   },
   {
    "duration": 31,
    "start_time": "2022-01-22T12:08:52.030Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-22T12:08:52.063Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T12:08:52.077Z"
   },
   {
    "duration": 53,
    "start_time": "2022-01-22T12:08:52.090Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-22T12:08:52.145Z"
   },
   {
    "duration": 1187,
    "start_time": "2022-01-22T12:08:52.156Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-22T12:08:53.345Z"
   },
   {
    "duration": 174,
    "start_time": "2022-01-22T12:08:53.356Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-22T12:08:53.532Z"
   },
   {
    "duration": 461,
    "start_time": "2022-01-22T12:08:53.536Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T12:08:53.998Z"
   },
   {
    "duration": 16,
    "start_time": "2022-01-22T12:08:54.012Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T12:08:54.029Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T12:08:57.311Z"
   },
   {
    "duration": 43,
    "start_time": "2022-01-22T12:08:58.359Z"
   },
   {
    "duration": 42,
    "start_time": "2022-01-22T12:10:04.559Z"
   },
   {
    "duration": 367,
    "start_time": "2022-01-22T12:10:36.610Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T12:10:56.284Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-22T12:11:32.857Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-22T12:11:47.004Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T12:12:00.019Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T12:12:27.930Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-22T12:13:44.529Z"
   },
   {
    "duration": 1140,
    "start_time": "2022-01-22T12:16:25.380Z"
   },
   {
    "duration": 31,
    "start_time": "2022-01-22T12:16:26.522Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-22T12:16:26.555Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T12:16:26.564Z"
   },
   {
    "duration": 35,
    "start_time": "2022-01-22T12:16:26.579Z"
   },
   {
    "duration": 20,
    "start_time": "2022-01-22T12:16:26.615Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T12:16:26.637Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:16:26.642Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T12:16:26.650Z"
   },
   {
    "duration": 32,
    "start_time": "2022-01-22T12:16:26.655Z"
   },
   {
    "duration": 46,
    "start_time": "2022-01-22T12:16:26.688Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T12:16:26.736Z"
   },
   {
    "duration": 24,
    "start_time": "2022-01-22T12:16:26.749Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-22T12:16:26.774Z"
   },
   {
    "duration": 1129,
    "start_time": "2022-01-22T12:16:26.787Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-22T12:16:27.917Z"
   },
   {
    "duration": 219,
    "start_time": "2022-01-22T12:16:27.934Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-22T12:16:28.155Z"
   },
   {
    "duration": 471,
    "start_time": "2022-01-22T12:16:28.159Z"
   },
   {
    "duration": 12,
    "start_time": "2022-01-22T12:16:28.631Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T12:16:28.645Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-22T12:16:28.651Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T12:16:30.276Z"
   },
   {
    "duration": 48,
    "start_time": "2022-01-22T12:16:32.116Z"
   },
   {
    "duration": 41,
    "start_time": "2022-01-22T12:16:44.080Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T12:17:55.765Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-22T12:17:57.766Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:18:01.417Z"
   },
   {
    "duration": 131806,
    "start_time": "2022-01-22T12:18:03.094Z"
   },
   {
    "duration": 271,
    "start_time": "2022-01-22T12:43:03.230Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:43:06.856Z"
   },
   {
    "duration": 453,
    "start_time": "2022-01-22T12:43:14.991Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:43:29.449Z"
   },
   {
    "duration": 46,
    "start_time": "2022-01-22T12:43:37.056Z"
   },
   {
    "duration": 174665,
    "start_time": "2022-01-22T12:43:44.158Z"
   },
   {
    "duration": 275,
    "start_time": "2022-01-22T12:46:59.865Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:47:08.128Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-22T12:47:14.743Z"
   },
   {
    "duration": 48,
    "start_time": "2022-01-22T12:47:16.775Z"
   },
   {
    "duration": 113200,
    "start_time": "2022-01-22T12:47:18.536Z"
   },
   {
    "duration": 271,
    "start_time": "2022-01-22T12:49:23.733Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-22T12:49:30.686Z"
   },
   {
    "duration": 71906,
    "start_time": "2022-01-22T12:49:57.499Z"
   },
   {
    "duration": 342,
    "start_time": "2022-01-22T12:52:19.350Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-22T12:52:26.876Z"
   },
   {
    "duration": 452,
    "start_time": "2022-01-22T12:52:40.304Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-22T12:52:56.323Z"
   },
   {
    "duration": 1235,
    "start_time": "2022-01-22T12:53:01.457Z"
   },
   {
    "duration": 339,
    "start_time": "2022-01-22T12:53:44.738Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-22T12:53:59.245Z"
   },
   {
    "duration": 316,
    "start_time": "2022-01-22T12:58:38.696Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-22T12:59:30.652Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-22T12:59:55.939Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-23T09:04:11.021Z"
   },
   {
    "duration": 1162,
    "start_time": "2022-01-23T09:04:30.906Z"
   },
   {
    "duration": 27,
    "start_time": "2022-01-23T09:04:32.965Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-23T09:04:34.142Z"
   },
   {
    "duration": 15,
    "start_time": "2022-01-23T09:04:35.866Z"
   },
   {
    "duration": 39,
    "start_time": "2022-01-23T09:04:36.578Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-23T09:04:38.316Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-23T09:04:39.277Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-23T09:04:42.725Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-23T09:04:43.385Z"
   },
   {
    "duration": 33,
    "start_time": "2022-01-23T09:04:45.220Z"
   },
   {
    "duration": 14,
    "start_time": "2022-01-23T09:04:46.889Z"
   },
   {
    "duration": 11,
    "start_time": "2022-01-23T09:04:47.842Z"
   },
   {
    "duration": 25,
    "start_time": "2022-01-23T09:04:48.563Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-23T09:04:51.136Z"
   },
   {
    "duration": 1313,
    "start_time": "2022-01-23T09:04:56.583Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-23T09:05:09.749Z"
   },
   {
    "duration": 242,
    "start_time": "2022-01-23T09:05:11.311Z"
   },
   {
    "duration": 2,
    "start_time": "2022-01-23T09:05:14.323Z"
   },
   {
    "duration": 290,
    "start_time": "2022-01-23T09:05:15.467Z"
   },
   {
    "duration": 13,
    "start_time": "2022-01-23T09:05:22.728Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-23T09:05:26.516Z"
   },
   {
    "duration": 3,
    "start_time": "2022-01-23T09:05:29.178Z"
   },
   {
    "duration": 43,
    "start_time": "2022-01-23T09:05:29.588Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-23T09:05:31.713Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-23T09:05:34.293Z"
   },
   {
    "duration": 5,
    "start_time": "2022-01-23T09:05:39.432Z"
   },
   {
    "duration": 135132,
    "start_time": "2022-01-23T09:05:40.502Z"
   },
   {
    "duration": 315,
    "start_time": "2022-01-23T09:08:02.814Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-23T09:08:06.821Z"
   },
   {
    "duration": 461,
    "start_time": "2022-01-23T09:08:11.520Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-23T09:08:15.056Z"
   },
   {
    "duration": 47,
    "start_time": "2022-01-23T09:08:20.619Z"
   },
   {
    "duration": 180012,
    "start_time": "2022-01-23T09:08:24.018Z"
   },
   {
    "duration": 306,
    "start_time": "2022-01-23T09:12:18.669Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-23T09:12:21.377Z"
   },
   {
    "duration": 4,
    "start_time": "2022-01-23T09:12:22.234Z"
   },
   {
    "duration": 62,
    "start_time": "2022-01-23T09:12:23.996Z"
   },
   {
    "duration": 117045,
    "start_time": "2022-01-23T09:12:25.017Z"
   },
   {
    "duration": 318,
    "start_time": "2022-01-23T09:22:33.536Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-23T09:22:34.738Z"
   },
   {
    "duration": 75764,
    "start_time": "2022-01-23T09:22:40.250Z"
   },
   {
    "duration": 309,
    "start_time": "2022-01-23T09:24:49.479Z"
   },
   {
    "duration": 6,
    "start_time": "2022-01-23T09:24:50.531Z"
   },
   {
    "duration": 480,
    "start_time": "2022-01-23T09:24:54.603Z"
   },
   {
    "duration": 8,
    "start_time": "2022-01-23T09:24:57.917Z"
   },
   {
    "duration": 1229,
    "start_time": "2022-01-23T09:25:01.518Z"
   },
   {
    "duration": 458,
    "start_time": "2022-01-23T09:25:02.749Z"
   },
   {
    "duration": 7,
    "start_time": "2022-01-23T09:25:11.783Z"
   },
   {
    "duration": 388,
    "start_time": "2022-01-23T09:25:20.191Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-23T09:25:26.225Z"
   },
   {
    "duration": 10,
    "start_time": "2022-01-23T09:25:32.602Z"
   },
   {
    "duration": 9,
    "start_time": "2022-01-23T09:25:48.349Z"
   },
   {
    "duration": 686,
    "start_time": "2022-01-23T09:25:51.692Z"
   },
   {
    "duration": 404,
    "start_time": "2022-01-23T09:25:55.010Z"
   },
   {
    "duration": 366,
    "start_time": "2022-01-23T09:34:34.258Z"
   },
   {
    "duration": 281,
    "start_time": "2022-01-23T09:42:29.784Z"
   },
   {
    "duration": 267,
    "start_time": "2022-01-23T09:44:30.505Z"
   },
   {
    "duration": 4468,
    "start_time": "2022-02-09T08:11:34.407Z"
   },
   {
    "duration": 317,
    "start_time": "2022-02-09T08:11:42.269Z"
   },
   {
    "duration": 2464,
    "start_time": "2022-02-09T08:11:52.094Z"
   },
   {
    "duration": 31,
    "start_time": "2022-02-09T08:11:54.560Z"
   },
   {
    "duration": 13,
    "start_time": "2022-02-09T08:11:54.594Z"
   },
   {
    "duration": 59,
    "start_time": "2022-02-09T08:11:54.610Z"
   },
   {
    "duration": 63,
    "start_time": "2022-02-09T08:11:54.672Z"
   },
   {
    "duration": 49,
    "start_time": "2022-02-09T08:11:54.738Z"
   },
   {
    "duration": 6,
    "start_time": "2022-02-09T08:11:54.790Z"
   },
   {
    "duration": 13,
    "start_time": "2022-02-09T08:11:54.800Z"
   },
   {
    "duration": 7,
    "start_time": "2022-02-09T08:11:54.815Z"
   },
   {
    "duration": 103,
    "start_time": "2022-02-09T08:11:54.824Z"
   },
   {
    "duration": 52,
    "start_time": "2022-02-09T08:11:54.929Z"
   },
   {
    "duration": 18,
    "start_time": "2022-02-09T08:11:54.983Z"
   },
   {
    "duration": 72,
    "start_time": "2022-02-09T08:11:55.003Z"
   },
   {
    "duration": 15,
    "start_time": "2022-02-09T08:11:55.077Z"
   },
   {
    "duration": 2242,
    "start_time": "2022-02-09T08:11:55.094Z"
   },
   {
    "duration": 32,
    "start_time": "2022-02-09T08:11:57.339Z"
   },
   {
    "duration": 423,
    "start_time": "2022-02-09T08:11:57.373Z"
   },
   {
    "duration": 3479,
    "start_time": "2022-02-09T08:11:57.799Z"
   },
   {
    "duration": 867,
    "start_time": "2022-02-09T08:12:11.691Z"
   }
  ],
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.8"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": true,
   "title_cell": "Содержание",
   "title_sidebar": "Contents",
   "toc_cell": true,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "165px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}