covid_simple.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "covid_simple.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LCs2Q5vvGiDM",
        "colab_type": "text"
      },
      "source": [
        "#Подключаем библиотеки и загружаем данные"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tLz0ubq_yI_r",
        "colab_type": "code",
        "outputId": "588de6a9-4d0d-4250-88c8-55d853fc58c9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 323
        }
      },
      "source": [
        "!pip install catboost"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting catboost\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/b1/61/2b8106c8870601671d99ca94d8b8d180f2b740b7cdb95c930147508abcf9/catboost-0.23-cp36-none-manylinux1_x86_64.whl (64.7MB)\n",
            "\u001b[K     |████████████████████████████████| 64.8MB 60kB/s \n",
            "\u001b[?25hRequirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.18.3)\n",
            "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.0.3)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from catboost) (1.12.0)\n",
            "Requirement already satisfied: graphviz in /usr/local/lib/python3.6/dist-packages (from catboost) (0.10.1)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from catboost) (3.2.1)\n",
            "Requirement already satisfied: plotly in /usr/local/lib/python3.6/dist-packages (from catboost) (4.4.1)\n",
            "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from catboost) (1.4.1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
            "Requirement already satisfied: python-dateutil>=2.6.1 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2.8.1)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (1.2.0)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (0.10.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (2.4.7)\n",
            "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.6/dist-packages (from plotly->catboost) (1.3.3)\n",
            "Installing collected packages: catboost\n",
            "Successfully installed catboost-0.23\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2JMyOSZ7bLoa",
        "colab_type": "code",
        "outputId": "61531f39-5ed3-4be8-8451-f72ce8e1723e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "from sklearn import ensemble, model_selection, datasets, metrics, tree, linear_model, preprocessing \n",
        "from sklearn.neural_network import MLPRegressor\n",
        "\n",
        "from catboost import CatBoostRegressor\n",
        "\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torch.nn.functional as F\n",
        "from torch.autograd import Variable\n",
        "import torch.optim as optim\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "import lightgbm as lgb\n",
        "\n",
        "import xgboost as xgb\n",
        "\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "import numpy as np\n",
        "import random\n",
        "%matplotlib inline"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GBQ5OY9sVBkK",
        "colab_type": "text"
      },
      "source": [
        "## Данные по заражениям"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hjSHpNYZCUjY",
        "colab_type": "code",
        "outputId": "e92b575a-947c-4456-c92a-037148466b51",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/CSSEGISandData/COVID-19 -b web-data COVID-19-new"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'COVID-19-new'...\n",
            "remote: Enumerating objects: 35, done.\u001b[K\n",
            "remote: Counting objects:   2% (1/35)\u001b[K\rremote: Counting objects:   5% (2/35)\u001b[K\rremote: Counting objects:   8% (3/35)\u001b[K\rremote: Counting objects:  11% (4/35)\u001b[K\rremote: Counting objects:  14% (5/35)\u001b[K\rremote: Counting objects:  17% (6/35)\u001b[K\rremote: Counting objects:  20% (7/35)\u001b[K\rremote: Counting objects:  22% (8/35)\u001b[K\rremote: Counting objects:  25% (9/35)\u001b[K\rremote: Counting objects:  28% (10/35)\u001b[K\rremote: Counting objects:  31% (11/35)\u001b[K\rremote: Counting objects:  34% (12/35)\u001b[K\rremote: Counting objects:  37% (13/35)\u001b[K\rremote: Counting objects:  40% (14/35)\u001b[K\rremote: Counting objects:  42% (15/35)\u001b[K\rremote: Counting objects:  45% (16/35)\u001b[K\rremote: Counting objects:  48% (17/35)\u001b[K\rremote: Counting objects:  51% (18/35)\u001b[K\rremote: Counting objects:  54% (19/35)\u001b[K\rremote: Counting objects:  57% (20/35)\u001b[K\rremote: Counting objects:  60% (21/35)\u001b[K\rremote: Counting objects:  62% (22/35)\u001b[K\rremote: Counting objects:  65% (23/35)\u001b[K\rremote: Counting objects:  68% (24/35)\u001b[K\rremote: Counting objects:  71% (25/35)\u001b[K\rremote: Counting objects:  74% (26/35)\u001b[K\rremote: Counting objects:  77% (27/35)\u001b[K\rremote: Counting objects:  80% (28/35)\u001b[K\rremote: Counting objects:  82% (29/35)\u001b[K\rremote: Counting objects:  85% (30/35)\u001b[K\rremote: Counting objects:  88% (31/35)\u001b[K\rremote: Counting objects:  91% (32/35)\u001b[K\rremote: Counting objects:  94% (33/35)\u001b[K\rremote: Counting objects:  97% (34/35)\u001b[K\rremote: Counting objects: 100% (35/35)\u001b[K\rremote: Counting objects: 100% (35/35), done.\u001b[K\n",
            "remote: Compressing objects:   5% (1/18)\u001b[K\rremote: Compressing objects:  11% (2/18)\u001b[K\rremote: Compressing objects:  16% (3/18)\u001b[K\rremote: Compressing objects:  22% (4/18)\u001b[K\rremote: Compressing objects:  27% (5/18)\u001b[K\rremote: Compressing objects:  33% (6/18)\u001b[K\rremote: Compressing objects:  38% (7/18)\u001b[K\rremote: Compressing objects:  44% (8/18)\u001b[K\rremote: Compressing objects:  50% (9/18)\u001b[K\rremote: Compressing objects:  55% (10/18)\u001b[K\rremote: Compressing objects:  61% (11/18)\u001b[K\rremote: Compressing objects:  66% (12/18)\u001b[K\rremote: Compressing objects:  72% (13/18)\u001b[K\rremote: Compressing objects:  77% (14/18)\u001b[K\rremote: Compressing objects:  83% (15/18)\u001b[K\rremote: Compressing objects:  88% (16/18)\u001b[K\rremote: Compressing objects:  94% (17/18)\u001b[K\rremote: Compressing objects: 100% (18/18)\u001b[K\rremote: Compressing objects: 100% (18/18), done.\u001b[K\n",
            "Receiving objects:   0% (1/24642)   \rReceiving objects:   1% (247/24642)   \rReceiving objects:   2% (493/24642)   \rReceiving objects:   3% (740/24642)   \rReceiving objects:   4% (986/24642)   \rReceiving objects:   5% (1233/24642)   \rReceiving objects:   6% (1479/24642)   \rReceiving objects:   7% (1725/24642)   \rReceiving objects:   8% (1972/24642)   \rReceiving objects:   9% (2218/24642)   \rReceiving objects:  10% (2465/24642)   \rReceiving objects:  11% (2711/24642)   \rReceiving objects:  12% (2958/24642)   \rReceiving objects:  13% (3204/24642)   \rReceiving objects:  14% (3450/24642)   \rReceiving objects:  15% (3697/24642)   \rReceiving objects:  16% (3943/24642)   \rReceiving objects:  17% (4190/24642)   \rReceiving objects:  18% (4436/24642), 2.04 MiB | 4.01 MiB/s   \rReceiving objects:  19% (4682/24642), 2.04 MiB | 4.01 MiB/s   \rReceiving objects:  19% (4924/24642), 2.04 MiB | 4.01 MiB/s   \rReceiving objects:  20% (4929/24642), 16.19 MiB | 15.79 MiB/s   \rReceiving objects:  20% (4968/24642), 34.63 MiB | 22.70 MiB/s   \rReceiving objects:  21% (5175/24642), 56.12 MiB | 27.71 MiB/s   \rReceiving objects:  22% (5422/24642), 56.12 MiB | 27.71 MiB/s   \rReceiving objects:  23% (5668/24642), 56.12 MiB | 27.71 MiB/s   \rReceiving objects:  24% (5915/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  25% (6161/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  26% (6407/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  27% (6654/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  28% (6900/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  29% (7147/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  30% (7393/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  31% (7640/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  32% (7886/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  33% (8132/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  34% (8379/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  35% (8625/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  36% (8872/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  36% (8979/24642), 72.26 MiB | 28.61 MiB/s   \rReceiving objects:  37% (9118/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  38% (9364/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  39% (9611/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  40% (9857/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  41% (10104/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  42% (10350/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  43% (10597/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  44% (10843/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  45% (11089/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  46% (11336/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  47% (11582/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  48% (11829/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  49% (12075/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  50% (12321/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  51% (12568/24642), 88.12 MiB | 29.12 MiB/s   \rReceiving objects:  52% (12814/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  53% (13061/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  54% (13307/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  55% (13554/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  56% (13800/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  57% (14046/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  58% (14293/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  59% (14539/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  60% (14786/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  61% (15032/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  62% (15279/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  63% (15525/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  64% (15771/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  65% (16018/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  66% (16264/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  67% (16511/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  68% (16757/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  69% (17003/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  70% (17250/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  71% (17496/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  72% (17743/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  73% (17989/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  74% (18236/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  75% (18482/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  76% (18728/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  77% (18975/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  78% (19221/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  79% (19468/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  80% (19714/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  81% (19961/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  82% (20207/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  83% (20453/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  84% (20700/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  85% (20946/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  86% (21193/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  87% (21439/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  88% (21685/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  89% (21932/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  90% (22178/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  91% (22425/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  92% (22671/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  93% (22918/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  94% (23164/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  95% (23410/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  96% (23657/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  97% (23903/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  98% (24150/24642), 104.25 MiB | 29.57 MiB/s   \rReceiving objects:  99% (24396/24642), 104.25 MiB | 29.57 MiB/s   \rremote: Total 24642 (delta 20), reused 28 (delta 17), pack-reused 24607\u001b[K\n",
            "Receiving objects: 100% (24642/24642), 110.17 MiB | 29.65 MiB/s, done.\n",
            "Resolving deltas: 100% (13643/13643), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eEM4nwAcT-uj",
        "colab_type": "code",
        "outputId": "0ab4c97d-7dd7-4cf6-e88c-075859f66bd2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/CSSEGISandData/COVID-19.git\n"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'COVID-19'...\n",
            "remote: Enumerating objects: 35, done.\u001b[K\n",
            "remote: Counting objects: 100% (35/35), done.\u001b[K\n",
            "remote: Compressing objects: 100% (18/18), done.\u001b[K\n",
            "remote: Total 24642 (delta 20), reused 28 (delta 17), pack-reused 24607\u001b[K\n",
            "Receiving objects: 100% (24642/24642), 110.17 MiB | 31.67 MiB/s, done.\n",
            "Resolving deltas: 100% (13643/13643), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XI3hn-gjnTV5",
        "colab_type": "code",
        "outputId": "775079ff-91f1-4c63-9171-d5bec13fd36f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/grwlf/COVID-19_plus_Russia.git"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'COVID-19_plus_Russia'...\n",
            "remote: Enumerating objects: 624, done.\u001b[K\n",
            "remote: Counting objects:   0% (1/624)\u001b[K\rremote: Counting objects:   1% (7/624)\u001b[K\rremote: Counting objects:   2% (13/624)\u001b[K\rremote: Counting objects:   3% (19/624)\u001b[K\rremote: Counting objects:   4% (25/624)\u001b[K\rremote: Counting objects:   5% (32/624)\u001b[K\rremote: Counting objects:   6% (38/624)\u001b[K\rremote: Counting objects:   7% (44/624)\u001b[K\rremote: Counting objects:   8% (50/624)\u001b[K\rremote: Counting objects:   9% (57/624)\u001b[K\rremote: Counting objects:  10% (63/624)\u001b[K\rremote: Counting objects:  11% (69/624)\u001b[K\rremote: Counting objects:  12% (75/624)\u001b[K\rremote: Counting objects:  13% (82/624)\u001b[K\rremote: Counting objects:  14% (88/624)\u001b[K\rremote: Counting objects:  15% (94/624)\u001b[K\rremote: Counting objects:  16% (100/624)\u001b[K\rremote: Counting objects:  17% (107/624)\u001b[K\rremote: Counting objects:  18% (113/624)\u001b[K\rremote: Counting objects:  19% (119/624)\u001b[K\rremote: Counting objects:  20% (125/624)\u001b[K\rremote: Counting objects:  21% (132/624)\u001b[K\rremote: Counting objects:  22% (138/624)\u001b[K\rremote: Counting objects:  23% (144/624)\u001b[K\rremote: Counting objects:  24% (150/624)\u001b[K\rremote: Counting objects:  25% (156/624)\u001b[K\rremote: Counting objects:  26% (163/624)\u001b[K\rremote: Counting objects:  27% (169/624)\u001b[K\rremote: Counting objects:  28% (175/624)\u001b[K\rremote: Counting objects:  29% (181/624)\u001b[K\rremote: Counting objects:  30% (188/624)\u001b[K\rremote: Counting objects:  31% (194/624)\u001b[K\rremote: Counting objects:  32% (200/624)\u001b[K\rremote: Counting objects:  33% (206/624)\u001b[K\rremote: Counting objects:  34% (213/624)\u001b[K\rremote: Counting objects:  35% (219/624)\u001b[K\rremote: Counting objects:  36% (225/624)\u001b[K\rremote: Counting objects:  37% (231/624)\u001b[K\rremote: Counting objects:  38% (238/624)\u001b[K\rremote: Counting objects:  39% (244/624)\u001b[K\rremote: Counting objects:  40% (250/624)\u001b[K\rremote: Counting objects:  41% (256/624)\u001b[K\rremote: Counting objects:  42% (263/624)\u001b[K\rremote: Counting objects:  43% (269/624)\u001b[K\rremote: Counting objects:  44% (275/624)\u001b[K\rremote: Counting objects:  45% (281/624)\u001b[K\rremote: Counting objects:  46% (288/624)\u001b[K\rremote: Counting objects:  47% (294/624)\u001b[K\rremote: Counting objects:  48% (300/624)\u001b[K\rremote: Counting objects:  49% (306/624)\u001b[K\rremote: Counting objects:  50% (312/624)\u001b[K\rremote: Counting objects:  51% (319/624)\u001b[K\rremote: Counting objects:  52% (325/624)\u001b[K\rremote: Counting objects:  53% (331/624)\u001b[K\rremote: Counting objects:  54% (337/624)\u001b[K\rremote: Counting objects:  55% (344/624)\u001b[K\rremote: Counting objects:  56% (350/624)\u001b[K\rremote: Counting objects:  57% (356/624)\u001b[K\rremote: Counting objects:  58% (362/624)\u001b[K\rremote: Counting objects:  59% (369/624)\u001b[K\rremote: Counting objects:  60% (375/624)\u001b[K\rremote: Counting objects:  61% (381/624)\u001b[K\rremote: Counting objects:  62% (387/624)\u001b[K\rremote: Counting objects:  63% (394/624)\u001b[K\rremote: Counting objects:  64% (400/624)\u001b[K\rremote: Counting objects:  65% (406/624)\u001b[K\rremote: Counting objects:  66% (412/624)\u001b[K\rremote: Counting objects:  67% (419/624)\u001b[K\rremote: Counting objects:  68% (425/624)\u001b[K\rremote: Counting objects:  69% (431/624)\u001b[K\rremote: Counting objects:  70% (437/624)\u001b[K\rremote: Counting objects:  71% (444/624)\u001b[K\rremote: Counting objects:  72% (450/624)\u001b[K\rremote: Counting objects:  73% (456/624)\u001b[K\rremote: Counting objects:  74% (462/624)\u001b[K\rremote: Counting objects:  75% (468/624)\u001b[K\rremote: Counting objects:  76% (475/624)\u001b[K\rremote: Counting objects:  77% (481/624)\u001b[K\rremote: Counting objects:  78% (487/624)\u001b[K\rremote: Counting objects:  79% (493/624)\u001b[K\rremote: Counting objects:  80% (500/624)\u001b[K\rremote: Counting objects:  81% (506/624)\u001b[K\rremote: Counting objects:  82% (512/624)\u001b[K\rremote: Counting objects:  83% (518/624)\u001b[K\rremote: Counting objects:  84% (525/624)\u001b[K\rremote: Counting objects:  85% (531/624)\u001b[K\rremote: Counting objects:  86% (537/624)\u001b[K\rremote: Counting objects:  87% (543/624)\u001b[K\rremote: Counting objects:  88% (550/624)\u001b[K\rremote: Counting objects:  89% (556/624)\u001b[K\rremote: Counting objects:  90% (562/624)\u001b[K\rremote: Counting objects:  91% (568/624)\u001b[K\rremote: Counting objects:  92% (575/624)\u001b[K\rremote: Counting objects:  93% (581/624)\u001b[K\rremote: Counting objects:  94% (587/624)\u001b[K\rremote: Counting objects:  95% (593/624)\u001b[K\rremote: Counting objects:  96% (600/624)\u001b[K\rremote: Counting objects:  97% (606/624)\u001b[K\rremote: Counting objects:  98% (612/624)\u001b[K\rremote: Counting objects:  99% (618/624)\u001b[K\rremote: Counting objects: 100% (624/624)\u001b[K\rremote: Counting objects: 100% (624/624), done.\u001b[K\n",
            "remote: Compressing objects: 100% (252/252), done.\u001b[K\n",
            "remote: Total 18171 (delta 376), reused 615 (delta 370), pack-reused 17547\u001b[K\n",
            "Receiving objects: 100% (18171/18171), 97.87 MiB | 26.89 MiB/s, done.\n",
            "Resolving deltas: 100% (9285/9285), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Hcd2xqhUVNAB",
        "colab_type": "text"
      },
      "source": [
        "## Данные про страны"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xNntNTQ1UsKO",
        "colab_type": "code",
        "outputId": "678c397f-f33c-4971-f98a-f98c22975602",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/vlomme/sberbank-covid19-forecast-2020.git"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'sberbank-covid19-forecast-2020'...\n",
            "remote: Enumerating objects: 81, done.\u001b[K\n",
            "remote: Counting objects:   1% (1/81)\u001b[K\rremote: Counting objects:   2% (2/81)\u001b[K\rremote: Counting objects:   3% (3/81)\u001b[K\rremote: Counting objects:   4% (4/81)\u001b[K\rremote: Counting objects:   6% (5/81)\u001b[K\rremote: Counting objects:   7% (6/81)\u001b[K\rremote: Counting objects:   8% (7/81)\u001b[K\rremote: Counting objects:   9% (8/81)\u001b[K\rremote: Counting objects:  11% (9/81)\u001b[K\rremote: Counting objects:  12% (10/81)\u001b[K\rremote: Counting objects:  13% (11/81)\u001b[K\rremote: Counting objects:  14% (12/81)\u001b[K\rremote: Counting objects:  16% (13/81)\u001b[K\rremote: Counting objects:  17% (14/81)\u001b[K\rremote: Counting objects:  18% (15/81)\u001b[K\rremote: Counting objects:  19% (16/81)\u001b[K\rremote: Counting objects:  20% (17/81)\u001b[K\rremote: Counting objects:  22% (18/81)\u001b[K\rremote: Counting objects:  23% (19/81)\u001b[K\rremote: Counting objects:  24% (20/81)\u001b[K\rremote: Counting objects:  25% (21/81)\u001b[K\rremote: Counting objects:  27% (22/81)\u001b[K\rremote: Counting objects:  28% (23/81)\u001b[K\rremote: Counting objects:  29% (24/81)\u001b[K\rremote: Counting objects:  30% (25/81)\u001b[K\rremote: Counting objects:  32% (26/81)\u001b[K\rremote: Counting objects:  33% (27/81)\u001b[K\rremote: Counting objects:  34% (28/81)\u001b[K\rremote: Counting objects:  35% (29/81)\u001b[K\rremote: Counting objects:  37% (30/81)\u001b[K\rremote: Counting objects:  38% (31/81)\u001b[K\rremote: Counting objects:  39% (32/81)\u001b[K\rremote: Counting objects:  40% (33/81)\u001b[K\rremote: Counting objects:  41% (34/81)\u001b[K\rremote: Counting objects:  43% (35/81)\u001b[K\rremote: Counting objects:  44% (36/81)\u001b[K\rremote: Counting objects:  45% (37/81)\u001b[K\rremote: Counting objects:  46% (38/81)\u001b[K\rremote: Counting objects:  48% (39/81)\u001b[K\rremote: Counting objects:  49% (40/81)\u001b[K\rremote: Counting objects:  50% (41/81)\u001b[K\rremote: Counting objects:  51% (42/81)\u001b[K\rremote: Counting objects:  53% (43/81)\u001b[K\rremote: Counting objects:  54% (44/81)\u001b[K\rremote: Counting objects:  55% (45/81)\u001b[K\rremote: Counting objects:  56% (46/81)\u001b[K\rremote: Counting objects:  58% (47/81)\u001b[K\rremote: Counting objects:  59% (48/81)\u001b[K\rremote: Counting objects:  60% (49/81)\u001b[K\rremote: Counting objects:  61% (50/81)\u001b[K\rremote: Counting objects:  62% (51/81)\u001b[K\rremote: Counting objects:  64% (52/81)\u001b[K\rremote: Counting objects:  65% (53/81)\u001b[K\rremote: Counting objects:  66% (54/81)\u001b[K\rremote: Counting objects:  67% (55/81)\u001b[K\rremote: Counting objects:  69% (56/81)\u001b[K\rremote: Counting objects:  70% (57/81)\u001b[K\rremote: Counting objects:  71% (58/81)\u001b[K\rremote: Counting objects:  72% (59/81)\u001b[K\rremote: Counting objects:  74% (60/81)\u001b[K\rremote: Counting objects:  75% (61/81)\u001b[K\rremote: Counting objects:  76% (62/81)\u001b[K\rremote: Counting objects:  77% (63/81)\u001b[K\rremote: Counting objects:  79% (64/81)\u001b[K\rremote: Counting objects:  80% (65/81)\u001b[K\rremote: Counting objects:  81% (66/81)\u001b[K\rremote: Counting objects:  82% (67/81)\u001b[K\rremote: Counting objects:  83% (68/81)\u001b[K\rremote: Counting objects:  85% (69/81)\u001b[K\rremote: Counting objects:  86% (70/81)\u001b[K\rremote: Counting objects:  87% (71/81)\u001b[K\rremote: Counting objects:  88% (72/81)\u001b[K\rremote: Counting objects:  90% (73/81)\u001b[K\rremote: Counting objects:  91% (74/81)\u001b[K\rremote: Counting objects:  92% (75/81)\u001b[K\rremote: Counting objects:  93% (76/81)\u001b[K\rremote: Counting objects:  95% (77/81)\u001b[K\rremote: Counting objects:  96% (78/81)\u001b[K\rremote: Counting objects:  97% (79/81)\u001b[K\rremote: Counting objects:  98% (80/81)\u001b[K\rremote: Counting objects: 100% (81/81)\u001b[K\rremote: Counting objects: 100% (81/81), done.\u001b[K\n",
            "remote: Compressing objects:   1% (1/81)\u001b[K\rremote: Compressing objects:   2% (2/81)\u001b[K\rremote: Compressing objects:   3% (3/81)\u001b[K\rremote: Compressing objects:   4% (4/81)\u001b[K\rremote: Compressing objects:   6% (5/81)\u001b[K\rremote: Compressing objects:   7% (6/81)\u001b[K\rremote: Compressing objects:   8% (7/81)\u001b[K\rremote: Compressing objects:   9% (8/81)\u001b[K\rremote: Compressing objects:  11% (9/81)\u001b[K\rremote: Compressing objects:  12% (10/81)\u001b[K\rremote: Compressing objects:  13% (11/81)\u001b[K\rremote: Compressing objects:  14% (12/81)\u001b[K\rremote: Compressing objects:  16% (13/81)\u001b[K\rremote: Compressing objects:  17% (14/81)\u001b[K\rremote: Compressing objects:  18% (15/81)\u001b[K\rremote: Compressing objects:  19% (16/81)\u001b[K\rremote: Compressing objects:  20% (17/81)\u001b[K\rremote: Compressing objects:  22% (18/81)\u001b[K\rremote: Compressing objects:  23% (19/81)\u001b[K\rremote: Compressing objects:  24% (20/81)\u001b[K\rremote: Compressing objects:  25% (21/81)\u001b[K\rremote: Compressing objects:  27% (22/81)\u001b[K\rremote: Compressing objects:  28% (23/81)\u001b[K\rremote: Compressing objects:  29% (24/81)\u001b[K\rremote: Compressing objects:  30% (25/81)\u001b[K\rremote: Compressing objects:  32% (26/81)\u001b[K\rremote: Compressing objects:  33% (27/81)\u001b[K\rremote: Compressing objects:  34% (28/81)\u001b[K\rremote: Compressing objects:  35% (29/81)\u001b[K\rremote: Compressing objects:  37% (30/81)\u001b[K\rremote: Compressing objects:  38% (31/81)\u001b[K\rremote: Compressing objects:  39% (32/81)\u001b[K\rremote: Compressing objects:  40% (33/81)\u001b[K\rremote: Compressing objects:  41% (34/81)\u001b[K\rremote: Compressing objects:  43% (35/81)\u001b[K\rremote: Compressing objects:  44% (36/81)\u001b[K\rremote: Compressing objects:  45% (37/81)\u001b[K\rremote: Compressing objects:  46% (38/81)\u001b[K\rremote: Compressing objects:  48% (39/81)\u001b[K\rremote: Compressing objects:  49% (40/81)\u001b[K\rremote: Compressing objects:  50% (41/81)\u001b[K\rremote: Compressing objects:  51% (42/81)\u001b[K\rremote: Compressing objects:  53% (43/81)\u001b[K\rremote: Compressing objects:  54% (44/81)\u001b[K\rremote: Compressing objects:  55% (45/81)\u001b[K\rremote: Compressing objects:  56% (46/81)\u001b[K\rremote: Compressing objects:  58% (47/81)\u001b[K\rremote: Compressing objects:  59% (48/81)\u001b[K\rremote: Compressing objects:  60% (49/81)\u001b[K\rremote: Compressing objects:  61% (50/81)\u001b[K\rremote: Compressing objects:  62% (51/81)\u001b[K\rremote: Compressing objects:  64% (52/81)\u001b[K\rremote: Compressing objects:  65% (53/81)\u001b[K\rremote: Compressing objects:  66% (54/81)\u001b[K\rremote: Compressing objects:  67% (55/81)\u001b[K\rremote: Compressing objects:  69% (56/81)\u001b[K\rremote: Compressing objects:  70% (57/81)\u001b[K\rremote: Compressing objects:  71% (58/81)\u001b[K\rremote: Compressing objects:  72% (59/81)\u001b[K\rremote: Compressing objects:  74% (60/81)\u001b[K\rremote: Compressing objects:  75% (61/81)\u001b[K\rremote: Compressing objects:  76% (62/81)\u001b[K\rremote: Compressing objects:  77% (63/81)\u001b[K\rremote: Compressing objects:  79% (64/81)\u001b[K\rremote: Compressing objects:  80% (65/81)\u001b[K\rremote: Compressing objects:  81% (66/81)\u001b[K\rremote: Compressing objects:  82% (67/81)\u001b[K\rremote: Compressing objects:  83% (68/81)\u001b[K\rremote: Compressing objects:  85% (69/81)\u001b[K\rremote: Compressing objects:  86% (70/81)\u001b[K\rremote: Compressing objects:  87% (71/81)\u001b[K\rremote: Compressing objects:  88% (72/81)\u001b[K\rremote: Compressing objects:  90% (73/81)\u001b[K\rremote: Compressing objects:  91% (74/81)\u001b[K\rremote: Compressing objects:  92% (75/81)\u001b[K\rremote: Compressing objects:  93% (76/81)\u001b[K\rremote: Compressing objects:  95% (77/81)\u001b[K\rremote: Compressing objects:  96% (78/81)\u001b[K\rremote: Compressing objects:  97% (79/81)\u001b[K\rremote: Compressing objects:  98% (80/81)\u001b[K\rremote: Compressing objects: 100% (81/81)\u001b[K\rremote: Compressing objects: 100% (81/81), done.\u001b[K\n",
            "Receiving objects:   0% (1/274)   \rReceiving objects:   1% (3/274)   \rReceiving objects:   2% (6/274)   \rReceiving objects:   3% (9/274)   \rReceiving objects:   4% (11/274)   \rReceiving objects:   5% (14/274)   \rReceiving objects:   6% (17/274)   \rReceiving objects:   7% (20/274)   \rReceiving objects:   8% (22/274)   \rReceiving objects:   9% (25/274)   \rReceiving objects:  10% (28/274)   \rReceiving objects:  11% (31/274)   \rReceiving objects:  12% (33/274)   \rReceiving objects:  13% (36/274)   \rReceiving objects:  14% (39/274)   \rReceiving objects:  15% (42/274)   \rReceiving objects:  16% (44/274)   \rReceiving objects:  17% (47/274)   \rReceiving objects:  18% (50/274)   \rReceiving objects:  19% (53/274)   \rReceiving objects:  20% (55/274)   \rReceiving objects:  21% (58/274)   \rReceiving objects:  22% (61/274)   \rReceiving objects:  23% (64/274)   \rReceiving objects:  24% (66/274)   \rReceiving objects:  25% (69/274)   \rReceiving objects:  26% (72/274)   \rReceiving objects:  27% (74/274)   \rReceiving objects:  28% (77/274)   \rReceiving objects:  29% (80/274)   \rReceiving objects:  30% (83/274)   \rReceiving objects:  31% (85/274)   \rReceiving objects:  32% (88/274)   \rReceiving objects:  33% (91/274)   \rReceiving objects:  34% (94/274)   \rReceiving objects:  35% (96/274)   \rReceiving objects:  36% (99/274)   \rReceiving objects:  37% (102/274)   \rReceiving objects:  38% (105/274)   \rReceiving objects:  39% (107/274)   \rReceiving objects:  40% (110/274)   \rReceiving objects:  41% (113/274)   \rReceiving objects:  42% (116/274)   \rReceiving objects:  43% (118/274)   \rReceiving objects:  44% (121/274)   \rReceiving objects:  45% (124/274)   \rReceiving objects:  46% (127/274)   \rReceiving objects:  47% (129/274)   \rReceiving objects:  48% (132/274)   \rReceiving objects:  49% (135/274)   \rReceiving objects:  50% (137/274)   \rReceiving objects:  51% (140/274)   \rReceiving objects:  52% (143/274)   \rReceiving objects:  53% (146/274)   \rReceiving objects:  54% (148/274)   \rReceiving objects:  55% (151/274)   \rReceiving objects:  56% (154/274)   \rReceiving objects:  57% (157/274)   \rReceiving objects:  58% (159/274)   \rReceiving objects:  59% (162/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  60% (165/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  61% (168/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  62% (170/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  63% (173/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  64% (176/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  65% (179/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  66% (181/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  67% (184/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  68% (187/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  69% (190/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  70% (192/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  71% (195/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  72% (198/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  73% (201/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  74% (203/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  75% (206/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  76% (209/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  77% (211/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  78% (214/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  79% (217/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  80% (220/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  81% (222/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  82% (225/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  83% (228/274), 4.67 MiB | 9.22 MiB/s   \rReceiving objects:  83% (228/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  84% (231/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  85% (233/274), 11.12 MiB | 10.65 MiB/s   \rremote: Total 274 (delta 40), reused 0 (delta 0), pack-reused 193\u001b[K\n",
            "Receiving objects:  86% (236/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  87% (239/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  88% (242/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  89% (244/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  90% (247/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  91% (250/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  92% (253/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  93% (255/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  94% (258/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  95% (261/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  96% (264/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  97% (266/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  98% (269/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects:  99% (272/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects: 100% (274/274), 11.12 MiB | 10.65 MiB/s   \rReceiving objects: 100% (274/274), 12.21 MiB | 8.90 MiB/s, done.\n",
            "Resolving deltas:   0% (0/134)   \rResolving deltas:   1% (2/134)   \rResolving deltas:   8% (12/134)   \rResolving deltas:  12% (17/134)   \rResolving deltas:  33% (45/134)   \rResolving deltas:  41% (56/134)   \rResolving deltas:  64% (87/134)   \rResolving deltas:  67% (91/134)   \rResolving deltas:  81% (109/134)   \rResolving deltas:  88% (119/134)   \rResolving deltas:  90% (121/134)   \rResolving deltas:  91% (122/134)   \rResolving deltas:  97% (130/134)   \rResolving deltas:  98% (132/134)   \rResolving deltas:  99% (133/134)   \rResolving deltas: 100% (134/134)   \rResolving deltas: 100% (134/134), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "73CcvjEpY2OX",
        "colab_type": "text"
      },
      "source": [
        "## Данные по карантинам"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "V7qgyyfjYva5",
        "colab_type": "code",
        "outputId": "d0c3601c-900d-46d3-b876-16252944fdba",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 136
        }
      },
      "source": [
        "!git clone https://github.com/tyz910/sberbank-covid19.git"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Cloning into 'sberbank-covid19'...\n",
            "remote: Enumerating objects: 216, done.\u001b[K\n",
            "remote: Counting objects:   0% (1/216)\u001b[K\rremote: Counting objects:   1% (3/216)\u001b[K\rremote: Counting objects:   2% (5/216)\u001b[K\rremote: Counting objects:   3% (7/216)\u001b[K\rremote: Counting objects:   4% (9/216)\u001b[K\rremote: Counting objects:   5% (11/216)\u001b[K\rremote: Counting objects:   6% (13/216)\u001b[K\rremote: Counting objects:   7% (16/216)\u001b[K\rremote: Counting objects:   8% (18/216)\u001b[K\rremote: Counting objects:   9% (20/216)\u001b[K\rremote: Counting objects:  10% (22/216)\u001b[K\rremote: Counting objects:  11% (24/216)\u001b[K\rremote: Counting objects:  12% (26/216)\u001b[K\rremote: Counting objects:  13% (29/216)\u001b[K\rremote: Counting objects:  14% (31/216)\u001b[K\rremote: Counting objects:  15% (33/216)\u001b[K\rremote: Counting objects:  16% (35/216)\u001b[K\rremote: Counting objects:  17% (37/216)\u001b[K\rremote: Counting objects:  18% (39/216)\u001b[K\rremote: Counting objects:  19% (42/216)\u001b[K\rremote: Counting objects:  20% (44/216)\u001b[K\rremote: Counting objects:  21% (46/216)\u001b[K\rremote: Counting objects:  22% (48/216)\u001b[K\rremote: Counting objects:  23% (50/216)\u001b[K\rremote: Counting objects:  24% (52/216)\u001b[K\rremote: Counting objects:  25% (54/216)\u001b[K\rremote: Counting objects:  26% (57/216)\u001b[K\rremote: Counting objects:  27% (59/216)\u001b[K\rremote: Counting objects:  28% (61/216)\u001b[K\rremote: Counting objects:  29% (63/216)\u001b[K\rremote: Counting objects:  30% (65/216)\u001b[K\rremote: Counting objects:  31% (67/216)\u001b[K\rremote: Counting objects:  32% (70/216)\u001b[K\rremote: Counting objects:  33% (72/216)\u001b[K\rremote: Counting objects:  34% (74/216)\u001b[K\rremote: Counting objects:  35% (76/216)\u001b[K\rremote: Counting objects:  36% (78/216)\u001b[K\rremote: Counting objects:  37% (80/216)\u001b[K\rremote: Counting objects:  38% (83/216)\u001b[K\rremote: Counting objects:  39% (85/216)\u001b[K\rremote: Counting objects:  40% (87/216)\u001b[K\rremote: Counting objects:  41% (89/216)\u001b[K\rremote: Counting objects:  42% (91/216)\u001b[K\rremote: Counting objects:  43% (93/216)\u001b[K\rremote: Counting objects:  44% (96/216)\u001b[K\rremote: Counting objects:  45% (98/216)\u001b[K\rremote: Counting objects:  46% (100/216)\u001b[K\rremote: Counting objects:  47% (102/216)\u001b[K\rremote: Counting objects:  48% (104/216)\u001b[K\rremote: Counting objects:  49% (106/216)\u001b[K\rremote: Counting objects:  50% (108/216)\u001b[K\rremote: Counting objects:  51% (111/216)\u001b[K\rremote: Counting objects:  52% (113/216)\u001b[K\rremote: Counting objects:  53% (115/216)\u001b[K\rremote: Counting objects:  54% (117/216)\u001b[K\rremote: Counting objects:  55% (119/216)\u001b[K\rremote: Counting objects:  56% (121/216)\u001b[K\rremote: Counting objects:  57% (124/216)\u001b[K\rremote: Counting objects:  58% (126/216)\u001b[K\rremote: Counting objects:  59% (128/216)\u001b[K\rremote: Counting objects:  60% (130/216)\u001b[K\rremote: Counting objects:  61% (132/216)\u001b[K\rremote: Counting objects:  62% (134/216)\u001b[K\rremote: Counting objects:  63% (137/216)\u001b[K\rremote: Counting objects:  64% (139/216)\u001b[K\rremote: Counting objects:  65% (141/216)\u001b[K\rremote: Counting objects:  66% (143/216)\u001b[K\rremote: Counting objects:  67% (145/216)\u001b[K\rremote: Counting objects:  68% (147/216)\u001b[K\rremote: Counting objects:  69% (150/216)\u001b[K\rremote: Counting objects:  70% (152/216)\u001b[K\rremote: Counting objects:  71% (154/216)\u001b[K\rremote: Counting objects:  72% (156/216)\u001b[K\rremote: Counting objects:  73% (158/216)\u001b[K\rremote: Counting objects:  74% (160/216)\u001b[K\rremote: Counting objects:  75% (162/216)\u001b[K\rremote: Counting objects:  76% (165/216)\u001b[K\rremote: Counting objects:  77% (167/216)\u001b[K\rremote: Counting objects:  78% (169/216)\u001b[K\rremote: Counting objects:  79% (171/216)\u001b[K\rremote: Counting objects:  80% (173/216)\u001b[K\rremote: Counting objects:  81% (175/216)\u001b[K\rremote: Counting objects:  82% (178/216)\u001b[K\rremote: Counting objects:  83% (180/216)\u001b[K\rremote: Counting objects:  84% (182/216)\u001b[K\rremote: Counting objects:  85% (184/216)\u001b[K\rremote: Counting objects:  86% (186/216)\u001b[K\rremote: Counting objects:  87% (188/216)\u001b[K\rremote: Counting objects:  88% (191/216)\u001b[K\rremote: Counting objects:  89% (193/216)\u001b[K\rremote: Counting objects:  90% (195/216)\u001b[K\rremote: Counting objects:  91% (197/216)\u001b[K\rremote: Counting objects:  92% (199/216)\u001b[K\rremote: Counting objects:  93% (201/216)\u001b[K\rremote: Counting objects:  94% (204/216)\u001b[K\rremote: Counting objects:  95% (206/216)\u001b[K\rremote: Counting objects:  96% (208/216)\u001b[K\rremote: Counting objects:  97% (210/216)\u001b[K\rremote: Counting objects:  98% (212/216)\u001b[K\rremote: Counting objects:  99% (214/216)\u001b[K\rremote: Counting objects: 100% (216/216)\u001b[K\rremote: Counting objects: 100% (216/216), done.\u001b[K\n",
            "remote: Compressing objects:   0% (1/133)\u001b[K\rremote: Compressing objects:   1% (2/133)\u001b[K\rremote: Compressing objects:   2% (3/133)\u001b[K\rremote: Compressing objects:   3% (4/133)\u001b[K\rremote: Compressing objects:   4% (6/133)\u001b[K\rremote: Compressing objects:   5% (7/133)\u001b[K\rremote: Compressing objects:   6% (8/133)\u001b[K\rremote: Compressing objects:   7% (10/133)\u001b[K\rremote: Compressing objects:   8% (11/133)\u001b[K\rremote: Compressing objects:   9% (12/133)\u001b[K\rremote: Compressing objects:  10% (14/133)\u001b[K\rremote: Compressing objects:  11% (15/133)\u001b[K\rremote: Compressing objects:  12% (16/133)\u001b[K\rremote: Compressing objects:  13% (18/133)\u001b[K\rremote: Compressing objects:  14% (19/133)\u001b[K\rremote: Compressing objects:  15% (20/133)\u001b[K\rremote: Compressing objects:  16% (22/133)\u001b[K\rremote: Compressing objects:  17% (23/133)\u001b[K\rremote: Compressing objects:  18% (24/133)\u001b[K\rremote: Compressing objects:  19% (26/133)\u001b[K\rremote: Compressing objects:  20% (27/133)\u001b[K\rremote: Compressing objects:  21% (28/133)\u001b[K\rremote: Compressing objects:  22% (30/133)\u001b[K\rremote: Compressing objects:  23% (31/133)\u001b[K\rremote: Compressing objects:  24% (32/133)\u001b[K\rremote: Compressing objects:  25% (34/133)\u001b[K\rremote: Compressing objects:  26% (35/133)\u001b[K\rremote: Compressing objects:  27% (36/133)\u001b[K\rremote: Compressing objects:  28% (38/133)\u001b[K\rremote: Compressing objects:  29% (39/133)\u001b[K\rremote: Compressing objects:  30% (40/133)\u001b[K\rremote: Compressing objects:  31% (42/133)\u001b[K\rremote: Compressing objects:  32% (43/133)\u001b[K\rremote: Compressing objects:  33% (44/133)\u001b[K\rremote: Compressing objects:  34% (46/133)\u001b[K\rremote: Compressing objects:  35% (47/133)\u001b[K\rremote: Compressing objects:  36% (48/133)\u001b[K\rremote: Compressing objects:  37% (50/133)\u001b[K\rremote: Compressing objects:  38% (51/133)\u001b[K\rremote: Compressing objects:  39% (52/133)\u001b[K\rremote: Compressing objects:  40% (54/133)\u001b[K\rremote: Compressing objects:  41% (55/133)\u001b[K\rremote: Compressing objects:  42% (56/133)\u001b[K\rremote: Compressing objects:  43% (58/133)\u001b[K\rremote: Compressing objects:  44% (59/133)\u001b[K\rremote: Compressing objects:  45% (60/133)\u001b[K\rremote: Compressing objects:  46% (62/133)\u001b[K\rremote: Compressing objects:  47% (63/133)\u001b[K\rremote: Compressing objects:  48% (64/133)\u001b[K\rremote: Compressing objects:  49% (66/133)\u001b[K\rremote: Compressing objects:  50% (67/133)\u001b[K\rremote: Compressing objects:  51% (68/133)\u001b[K\rremote: Compressing objects:  52% (70/133)\u001b[K\rremote: Compressing objects:  53% (71/133)\u001b[K\rremote: Compressing objects:  54% (72/133)\u001b[K\rremote: Compressing objects:  55% (74/133)\u001b[K\rremote: Compressing objects:  56% (75/133)\u001b[K\rremote: Compressing objects:  57% (76/133)\u001b[K\rremote: Compressing objects:  58% (78/133)\u001b[K\rremote: Compressing objects:  59% (79/133)\u001b[K\rremote: Compressing objects:  60% (80/133)\u001b[K\rremote: Compressing objects:  61% (82/133)\u001b[K\rremote: Compressing objects:  62% (83/133)\u001b[K\rremote: Compressing objects:  63% (84/133)\u001b[K\rremote: Compressing objects:  64% (86/133)\u001b[K\rremote: Compressing objects:  65% (87/133)\u001b[K\rremote: Compressing objects:  66% (88/133)\u001b[K\rremote: Compressing objects:  67% (90/133)\u001b[K\rremote: Compressing objects:  68% (91/133)\u001b[K\rremote: Compressing objects:  69% (92/133)\u001b[K\rremote: Compressing objects:  70% (94/133)\u001b[K\rremote: Compressing objects:  71% (95/133)\u001b[K\rremote: Compressing objects:  72% (96/133)\u001b[K\rremote: Compressing objects:  73% (98/133)\u001b[K\rremote: Compressing objects:  74% (99/133)\u001b[K\rremote: Compressing objects:  75% (100/133)\u001b[K\rremote: Compressing objects:  76% (102/133)\u001b[K\rremote: Compressing objects:  77% (103/133)\u001b[K\rremote: Compressing objects:  78% (104/133)\u001b[K\rremote: Compressing objects:  79% (106/133)\u001b[K\rremote: Compressing objects:  80% (107/133)\u001b[K\rremote: Compressing objects:  81% (108/133)\u001b[K\rremote: Compressing objects:  82% (110/133)\u001b[K\rremote: Compressing objects:  83% (111/133)\u001b[K\rremote: Compressing objects:  84% (112/133)\u001b[K\rremote: Compressing objects:  85% (114/133)\u001b[K\rremote: Compressing objects:  86% (115/133)\u001b[K\rremote: Compressing objects:  87% (116/133)\u001b[K\rremote: Compressing objects:  88% (118/133)\u001b[K\rremote: Compressing objects:  89% (119/133)\u001b[K\rremote: Compressing objects:  90% (120/133)\u001b[K\rremote: Compressing objects:  91% (122/133)\u001b[K\rremote: Compressing objects:  92% (123/133)\u001b[K\rremote: Compressing objects:  93% (124/133)\u001b[K\rremote: Compressing objects:  94% (126/133)\u001b[K\rremote: Compressing objects:  95% (127/133)\u001b[K\rremote: Compressing objects:  96% (128/133)\u001b[K\rremote: Compressing objects:  97% (130/133)\u001b[K\rremote: Compressing objects:  98% (131/133)\u001b[K\rremote: Compressing objects:  99% (132/133)\u001b[K\rremote: Compressing objects: 100% (133/133)\u001b[K\rremote: Compressing objects: 100% (133/133), done.\u001b[K\n",
            "Receiving objects:   0% (1/216)   \rReceiving objects:   1% (3/216)   \rReceiving objects:   2% (5/216)   \rReceiving objects:   3% (7/216)   \rReceiving objects:   4% (9/216)   \rReceiving objects:   5% (11/216)   \rReceiving objects:   6% (13/216)   \rReceiving objects:   7% (16/216)   \rReceiving objects:   8% (18/216)   \rReceiving objects:   9% (20/216)   \rReceiving objects:  10% (22/216)   \rReceiving objects:  11% (24/216)   \rReceiving objects:  12% (26/216)   \rReceiving objects:  13% (29/216)   \rReceiving objects:  14% (31/216)   \rReceiving objects:  15% (33/216)   \rReceiving objects:  16% (35/216)   \rReceiving objects:  17% (37/216), 100.00 KiB | 186.00 KiB/s   \rReceiving objects:  18% (39/216), 100.00 KiB | 186.00 KiB/s   \rReceiving objects:  19% (42/216), 100.00 KiB | 186.00 KiB/s   \rReceiving objects:  20% (44/216), 100.00 KiB | 186.00 KiB/s   \rReceiving objects:  20% (45/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  21% (46/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  22% (48/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  23% (50/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  24% (52/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  25% (54/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  26% (57/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  27% (59/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  28% (61/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  29% (63/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  30% (65/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  31% (67/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  32% (70/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  33% (72/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  34% (74/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  35% (76/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  36% (78/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  37% (80/216), 228.00 KiB | 216.00 KiB/s   \rReceiving objects:  37% (81/216), 636.00 KiB | 299.00 KiB/s   \rReceiving objects:  38% (83/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  39% (85/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  40% (87/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  41% (89/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  42% (91/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  43% (93/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  44% (96/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  45% (98/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  46% (100/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  47% (102/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  48% (104/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  49% (106/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  50% (108/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  51% (111/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  52% (113/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  53% (115/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  54% (117/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  55% (119/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  56% (121/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  57% (124/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  58% (126/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  59% (128/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  60% (130/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  61% (132/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  62% (134/216), 948.00 KiB | 358.00 KiB/s   \rReceiving objects:  63% (137/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  64% (139/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  65% (141/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  66% (143/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  67% (145/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  68% (147/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  69% (150/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  70% (152/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  71% (154/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  72% (156/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  73% (158/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  74% (160/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  75% (162/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  76% (165/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  77% (167/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  78% (169/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  79% (171/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  80% (173/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  81% (175/216), 1.34 MiB | 433.00 KiB/s   \rReceiving objects:  82% (178/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  83% (180/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  84% (182/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  85% (184/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  86% (186/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  87% (188/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  88% (191/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  89% (193/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  90% (195/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  91% (197/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  92% (199/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  93% (201/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  94% (204/216), 1.90 MiB | 529.00 KiB/s   \rremote: Total 216 (delta 100), reused 193 (delta 77), pack-reused 0\u001b[K\n",
            "Receiving objects:  95% (206/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  96% (208/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  97% (210/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  98% (212/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects:  99% (214/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects: 100% (216/216), 1.90 MiB | 529.00 KiB/s   \rReceiving objects: 100% (216/216), 2.04 MiB | 555.00 KiB/s, done.\n",
            "Resolving deltas:   0% (0/100)   \rResolving deltas:   3% (3/100)   \rResolving deltas:   8% (8/100)   \rResolving deltas:   9% (9/100)   \rResolving deltas:  11% (11/100)   \rResolving deltas:  15% (15/100)   \rResolving deltas:  17% (17/100)   \rResolving deltas:  22% (22/100)   \rResolving deltas:  23% (23/100)   \rResolving deltas:  38% (38/100)   \rResolving deltas:  39% (39/100)   \rResolving deltas:  42% (42/100)   \rResolving deltas:  44% (44/100)   \rResolving deltas:  45% (45/100)   \rResolving deltas:  53% (53/100)   \rResolving deltas:  56% (56/100)   \rResolving deltas:  59% (59/100)   \rResolving deltas:  61% (61/100)   \rResolving deltas:  62% (62/100)   \rResolving deltas:  63% (63/100)   \rResolving deltas:  64% (64/100)   \rResolving deltas:  70% (70/100)   \rResolving deltas:  75% (75/100)   \rResolving deltas:  76% (76/100)   \rResolving deltas:  78% (78/100)   \rResolving deltas:  81% (81/100)   \rResolving deltas:  83% (83/100)   \rResolving deltas:  86% (86/100)   \rResolving deltas:  87% (87/100)   \rResolving deltas:  89% (89/100)   \rResolving deltas:  92% (92/100)   \rResolving deltas:  93% (93/100)   \rResolving deltas:  94% (94/100)   \rResolving deltas:  97% (97/100)   \rResolving deltas:  99% (99/100)   \rResolving deltas: 100% (100/100)   \rResolving deltas: 100% (100/100), done.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cwR6lCCZev3a",
        "colab_type": "text"
      },
      "source": [
        "## Загружаем данные"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6z9w3muR2NA1",
        "colab_type": "code",
        "outputId": "901dc462-da0f-490c-bb4c-0f1475b8e6ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "today = pd.to_datetime('today').strftime(\"%Y-%m-%d\")\n",
        "#Загрузить список стран и удалить лишние атрибуты \n",
        "countries = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/countries.csv\")\n",
        "countries.drop(['iso_alpha2','iso_numeric','official_name','name'], axis='columns',inplace=True)\n",
        "\n",
        "#Загрузить список регионов\n",
        "regions = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/russia_regions.csv\")\n",
        "regions = regions[[\"iso_code\",\"csse_province_state\",\"csse_province_state2\",\"population\",\"population_urban\"]]\n",
        "regions.population_urban = regions.population_urban/regions.population\n",
        "regions.rename(columns={\"iso_code\": \"iso_alpha3\",\"csse_province_state\":\"ccse_name\",\"csse_province_state2\":\"ccse_name2\",\"population_urban\":\"urban_pop_rate\"},inplace=True)\n",
        "\n",
        "regions_info = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/regions-info.csv\", decimal=\",\")\n",
        "regions_info = regions_info[[\"Region_eng\",\"Density_pop_sqkm\",\"Area\"]]\n",
        "regions_info.rename(columns={\"Region_eng\": \"ccse_name2\",\"Area\": \"land_area\",\"Density_pop_sqkm\": \"density\"},inplace=True)\n",
        "regions= regions.merge(regions_info, how = 'left', left_on='ccse_name2', right_on='ccse_name2')\n",
        "\n",
        "#Загрузить данные заражений по России\n",
        "data_Ru = pd.read_csv(\"/content/COVID-19_plus_Russia/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_RU.csv\").groupby(\"Province_State\").sum()\n",
        "data_Ru = data_Ru.iloc[:,15:].stack().reset_index()\n",
        "data_Ru.columns = [\"countries\",\"date\",\"confirmed\"]\n",
        "#Загрузить данные смертей по России\n",
        "data_deaths_Ru = pd.read_csv(\"/content/COVID-19_plus_Russia/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_RU.csv\").groupby(\"Province_State\").sum()\n",
        "data_deaths_Ru = data_deaths_Ru.iloc[:,15:].stack().reset_index()\n",
        "data_deaths_Ru.columns = [\"countries\",\"date\",\"deaths\"]\n",
        "data_Ru = data_Ru.merge(data_deaths_Ru, how = 'left', left_on=('countries','date'), right_on=('countries','date'))\n",
        "data_Ru[\"recovered\"] = 0\n",
        "\n",
        "data_Ru_today = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/covid19-russia-cases.csv\")\n",
        "data_Ru_today = data_Ru_today[data_Ru_today['Region/City-Eng']!=\"Diamond Princess\"]\n",
        "data_Ru_today = data_Ru_today[[\"Region/City-Eng\",\"Date\",\"Confirmed\",\t\"Deaths\",\t\"Recovered\"]]\n",
        "data_Ru_today.columns = [\"countries\",\"date\",\"confirmed\",\"deaths\",\"recovered\"]\n",
        "data_Ru_today = data_Ru_today[data_Ru_today.date == today]\n",
        "data_Ru_today = data_Ru_today.replace(regions.ccse_name2.to_list(), regions.ccse_name.to_list())\n",
        "data_Ru = pd.concat([data_Ru, data_Ru_today])\n",
        "regions.drop(['ccse_name2'], axis='columns',inplace=True)\n",
        "countries = pd.concat([countries, regions])\n",
        "\n",
        "#Загрузить USA\n",
        "data_USA = pd.read_csv(\"/content/COVID-19/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_US.csv\")\n",
        "data_USA = data_USA.groupby(\"Province_State\").sum()\n",
        "data_USA = data_USA.iloc[:,5:].stack().reset_index()\n",
        "data_USA.columns = [\"countries\",\"date\",\"confirmed\"]\n",
        "\n",
        "data_USA_deaths = pd.read_csv(\"/content/COVID-19/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_US.csv\")\n",
        "data_USA_deaths = data_USA_deaths.groupby(\"Province_State\").sum()\n",
        "data_USA_deaths = data_USA_deaths.iloc[:,5:].stack().reset_index()\n",
        "data_USA_deaths.columns = [\"countries\",\"date\",\"deaths\"]\n",
        "data_USA = data_USA.merge(data_USA_deaths, how = 'left', left_on=('countries','date'), right_on=('countries','date'))\n",
        "data_USA[\"recovered\"] = 0\n",
        "data_USA.loc[data_USA.countries ==\"Georgia\",\"countries\"] = \"Georgia2\"\n",
        "USA = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/USA.csv\")\n",
        "USA[\"iso_alpha3\"] = \"USA_\" + USA[\"ccse_name\"]\n",
        "USA[\"urban_pop_rate\"] = USA[\"urban_pop_rate\"]/100\n",
        "countries = pd.concat([countries, USA])\n",
        "\n",
        "\n",
        "#Загрузить список стран c датами\n",
        "data_ans = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/sample_submission.csv\").iloc[:,:-1]\n",
        "data_ans[\"confirmed\"] = -1\n",
        "data_ans = data_ans.merge(countries, how = 'left', left_on='region', right_on='iso_alpha3')\n",
        "data_ans = data_ans[[\"date\",\"ccse_name\",\"confirmed\"]]\n",
        "data_ans = data_ans.rename(columns={\"ccse_name\": \"countries\"})\n",
        "data_ans[\"deaths\"] = 0\n",
        "data_ans[\"recovered\"] = 0\n",
        "data_ans = data_ans[data_ans.date >= today]\n",
        "\n",
        "#Загрузить данные заражений по миру\n",
        "data = pd.read_csv(\"/content/COVID-19/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_confirmed_global.csv\")\n",
        "data = data.groupby(\"Country/Region\").sum()\n",
        "\n",
        "#Загрузить данные смертей по миру\n",
        "data_deaths = pd.read_csv(\"/content/COVID-19/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv\")\n",
        "data_deaths = data_deaths.groupby(\"Country/Region\").sum()\n",
        "\n",
        "\n",
        "#Каждый день на новой строке\n",
        "data_deaths = data_deaths.iloc[:,4:].stack().reset_index()\n",
        "data_deaths.columns = [\"countries\",\"date\",\"deaths\"]\n",
        "\n",
        "#Загрузить данные смертей по миру\n",
        "data_recovered = pd.read_csv(\"/content/COVID-19/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_recovered_global.csv\")\n",
        "data_recovered = data_recovered.groupby(\"Country/Region\").sum()\n",
        "data_recovered = data_recovered.iloc[:,4:].stack().reset_index()\n",
        "data_recovered.columns = [\"countries\",\"date\",\"recovered\"]\n",
        "data_recovered[\"recovered\"] = np.log10(data_recovered[\"recovered\"]+1)\n",
        "\n",
        "#Карантин\n",
        "data_quarantine = pd.read_csv(\"/content/sberbank-covid19/data/quarantine.csv\")\n",
        "data_quarantine = data_quarantine.groupby(\"country\").max().loc[:,\"date\"]\n",
        "countries = countries.set_index(\"iso_alpha3\", drop=False)\n",
        "countries[\"Quarantine\"] = data_quarantine\n",
        "countries = countries.rename(columns={\"ccse_name\": \"countries\"})\n",
        "dic_countries = dict(zip(countries.iso_alpha3.to_list(), countries.countries.to_list()))\n",
        "\n",
        "# Сегоднящние данные\n",
        "countries_today = pd.read_csv(\"/content/COVID-19-new/data/cases_country.csv\")\n",
        "countries_today[\"Last_Update\"] = pd.to_datetime(countries_today[\"Last_Update\"]).apply(lambda x: pd.Series(x.strftime(\"%Y-%m-%d\")))\n",
        "countries_today = countries_today[[\"Country_Region\",\"Last_Update\",\"Confirmed\",\"Deaths\",\"Recovered\"]]\n",
        "countries_today.columns = ['countries', 'date', 'confirmed', 'deaths', 'recovered']\n",
        "countries_today.recovered =  np.log10(countries_today[\"recovered\"]+1)\n",
        "countries_today = countries_today[countries_today.countries.isin(countries.countries.to_list())]\n",
        "countries_today = countries_today.sort_values(by='countries')\n",
        "\n",
        "#Здравохранение\n",
        "Health_systems = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/2.12_Health_systems.csv\").groupby(\"Country_Region\").mean()\n",
        "Health_systems = Health_systems.fillna(Health_systems.mean())\n",
        "Health_systems = Health_systems[[\"Health_exp_pct_GDP_2016\",\"Health_exp_per_capita_USD_2016\",\"Physicians_per_1000_2009-18\",\"Nurse_midwife_per_1000_2009-18\"]]\n",
        "Health_systems.columns = [\"Health_GDP\",\"Health_USD\",\"Physicians\",\"Nurse\"]\n",
        "countries = countries.merge(Health_systems, how = 'left', left_on='countries', right_on='Country_Region')\n",
        "\n",
        "#Возраст\n",
        "WorldPop = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/WorldPopulationByAge2020.csv\")\n",
        "WorldPop = WorldPop.fillna(WorldPop.mean())\n",
        "WorldPop2 = (WorldPop[WorldPop[\"AgeGrp\"]==\"60+\"][\"PopTotal\"].to_numpy() + WorldPop[WorldPop[\"AgeGrp\"]==\"40-59\"][\"PopTotal\"].to_numpy())/(WorldPop[WorldPop[\"AgeGrp\"]==\"0-19\"][\"PopTotal\"].to_numpy()+WorldPop[WorldPop[\"AgeGrp\"]==\"20-39\"][\"PopTotal\"].to_numpy())\n",
        "WorldPop = WorldPop.groupby(\"Location\").mean()\n",
        "WorldPop[\"Age_old/new\"] = WorldPop2\n",
        "WorldPop = WorldPop[[\"Age_old/new\"]].reset_index()\n",
        "countries = countries.merge(WorldPop, how = 'left', left_on='countries', right_on='Location')\n",
        "\n",
        "#Курение\n",
        "smoke = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/share-of-adults-who-smoke.csv\")\n",
        "smoke = smoke[smoke.Year > 2015].groupby(\"Entity\").mean()\n",
        "smoke = smoke[[\"Smoking prevalence, total (ages 15+) (% of adults)\"]]\n",
        "smoke.columns = [\"Smoking\"]\n",
        "smoke = smoke.reset_index()\n",
        "countries= countries.merge(smoke, how = 'left', left_on='countries', right_on='Entity')\n",
        "countries.drop(['Entity','Location'], axis='columns',inplace=True)\n",
        "\n",
        "#Еще инфа о странах\n",
        "countries2 = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/data/covid19countryinfo.csv\")\n",
        "countries2 = countries2.fillna(0).groupby(\"country\").sum()\n",
        "countries2 = countries2[[\"tests\",\"testpop\",\"gatheringlimit\",\"hospibed\",\"healthperpop\"]]\n",
        "countries = countries.merge(countries2, how = 'left', left_on='countries', right_on='country')\n",
        "\n",
        "\n",
        "countries[\"region\"] = 0\n",
        "countries.loc[countries.iso_alpha3.str.contains('-'),\"region\"] = 1\n",
        "countries.loc[countries.iso_alpha3.str.contains('USA_'),\"region\"] = 2\n",
        "\n",
        "\"\"\"#mobility-google\n",
        "mobility = pd.read_csv(\"/content/sberbank-covid19/data/mobility-google.csv\")\n",
        "mobility.columns = [\"country\",\"date\",\"Parks\",\"Residential\",\"Retail & recreation\",\"Transit stations\",\"Workplaces\",\"Grocery & pharmacy\"]\n",
        "mobility[\"date\"] = pd.to_datetime(mobility.date)\n",
        "\n",
        "mobility_apple = pd.read_csv(\"/content/sberbank-covid19/data/mobility-apple.csv\")\n",
        "mobility_apple.drop(['apl-transit'], axis='columns',inplace=True)\n",
        "mobility_apple.columns = [\"date\", \"country\", \"driving\", \"walking\"]\n",
        "mobility_apple[\"date\"] = pd.to_datetime(mobility_apple.date)\n",
        "\n",
        "mobility_yandex = pd.read_csv(\"/content/sberbank-covid19/data/mobility-yandex.csv\")\n",
        "mobility_yandex[\"date\"] = pd.to_datetime(mobility_yandex.date)\"\"\""
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'#mobility-google\\nmobility = pd.read_csv(\"/content/sberbank-covid19/data/mobility-google.csv\")\\nmobility.columns = [\"country\",\"date\",\"Parks\",\"Residential\",\"Retail & recreation\",\"Transit stations\",\"Workplaces\",\"Grocery & pharmacy\"]\\nmobility[\"date\"] = pd.to_datetime(mobility.date)\\n\\nmobility_apple = pd.read_csv(\"/content/sberbank-covid19/data/mobility-apple.csv\")\\nmobility_apple.drop([\\'apl-transit\\'], axis=\\'columns\\',inplace=True)\\nmobility_apple.columns = [\"date\", \"country\", \"driving\", \"walking\"]\\nmobility_apple[\"date\"] = pd.to_datetime(mobility_apple.date)\\n\\nmobility_yandex = pd.read_csv(\"/content/sberbank-covid19/data/mobility-yandex.csv\")\\nmobility_yandex[\"date\"] = pd.to_datetime(mobility_yandex.date)'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lwl1Gki0kcPO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Избавляемся от Nan\n",
        "countries_mean = countries.mean()\n",
        "countries[countries.region==2] = countries[countries.region==2].fillna(countries[countries.iso_alpha3 == \"USA\"].to_dict('records')[0])\n",
        "countries[countries.region==1] = countries[countries.region==1].fillna(countries[countries.iso_alpha3 == \"RUS\"].to_dict('records')[0])\n",
        "countries.loc[countries.tests == 0,\"tests\"] = countries.tests.mean()\n",
        "countries.loc[countries.testpop == 0,\"testpop\"] = countries.testpop.mean()\n",
        "countries.loc[countries.gatheringlimit == 0,\"gatheringlimit\"] = countries.gatheringlimit.mean()\n",
        "countries.loc[countries.hospibed == 0,\"hospibed\"] = countries.hospibed.mean()\n",
        "countries['Quarantine'] = countries['Quarantine'].fillna(\"2020-04-10\")\n",
        "countries = countries.fillna(countries_mean.to_dict())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "v2gXwGZ-Js5u",
        "colab_type": "text"
      },
      "source": [
        "## Смотрим на первые страны"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TMVUPrvb_JYP",
        "colab_type": "code",
        "outputId": "e6bee96a-5ec5-4466-ab40-1f72602d3f91",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 282
        }
      },
      "source": [
        "\n",
        "data.iloc[:8].T.iloc[50:].plot()"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f6c58366fd0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZQjmbQDZz6NH",
        "colab_type": "text"
      },
      "source": [
        "# Обрабатываем данные"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aWqa7oXAHPmJ",
        "colab_type": "text"
      },
      "source": [
        "## Представляем данные в виде, где каждой дате соответствует свой столбец"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aZyMgfcUSUIB",
        "colab_type": "code",
        "outputId": "454ccc50-add1-4511-c54e-cfa6129fe903",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "data2 = data.iloc[:,4:].stack().reset_index()\n",
        "data2.columns = [\"countries\",\"date\",\"confirmed\"]\n",
        "\n",
        "#Соединим смерти и заражения\n",
        "data2 = data2.merge(data_deaths, how = 'left', left_on=('countries','date'), right_on=('countries','date'))\n",
        "\n",
        "# Добавим выздоровивших\n",
        "data2 = data2.merge(data_recovered, how = 'left', left_on=('countries','date'), right_on=('countries','date'))\n",
        "data2 = data2[data2.countries.isin(countries.countries.to_list())]\n",
        "\n",
        "#Объединим с сегодня\n",
        "last_confirmed = data2.loc[pd.to_datetime(data2.date) == pd.to_datetime(today) - pd.offsets.Day(1)].sort_values(by='countries')\n",
        "countries_today_new = countries_today[countries_today.confirmed.values > last_confirmed.confirmed.values]\n",
        "data2 = pd.concat([data2, countries_today_new],ignore_index=True)\n",
        "\n",
        "#Добавим Россию\n",
        "data2 = pd.concat([data2, data_Ru],ignore_index=True)\n",
        "\n",
        "#Добавим USA\n",
        "data2 = pd.concat([data2, data_USA],ignore_index=True)\n",
        "\n",
        "#Объединим с будущим\n",
        "data_ans = data_ans[(data_ans.date != today)|(~(data_ans.countries.isin(countries_today_new.countries.values))&~(data_ans.countries.isin(data_Ru_today.countries.values)))]\n",
        "data2 = pd.concat([data2, data_ans],ignore_index=True)\n",
        "\n",
        "\n",
        "#Добавим информацию о стране(плотность, население и тд)\n",
        "data2 = data2.merge(countries, how = 'right', left_on='countries', right_on='countries')\n",
        "\n",
        "#Удалим дни без больных\n",
        "data2 = data2[data2.confirmed != 0]\n",
        "data2.loc[data2.confirmed == -1,\"confirmed\"] = 0\n",
        "\n",
        "\n",
        "data2[\"date\"] = pd.to_datetime(data2.date)\n",
        "\n",
        "\"\"\"\n",
        "data2 = data2.merge(mobility_yandex, how = 'left', left_on=('iso_alpha3','date'), right_on=('country','date'))\n",
        "mobility_min = mobility_yandex.groupby(\"country\").min().drop(['date'], axis='columns')\n",
        "mobility_min.columns = ['min_isolation']\n",
        "data2 = data2.merge(mobility_min, how = 'left', left_on='iso_alpha3', right_on='country')\n",
        "data2.loc[(data2.isolation.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\n",
        "          'isolation'] = data2.loc[(data2.isolation.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),'min_isolation'].to_numpy()\n",
        "data2.drop(['country', 'min_isolation'], axis='columns',inplace=True)\n",
        "#data2['isolation'] = data2['isolation'].fillna(data2['isolation'].mean())\n",
        "data2['isolation'] = (data2['isolation'])/5\n",
        "\n",
        "\n",
        "data2 = data2.merge(mobility_apple, how = 'left', left_on=('iso_alpha3','date'), right_on=('country','date'))\n",
        "mobility_min = mobility_apple.groupby(\"country\").min().drop(['date'], axis='columns')\n",
        "mobility_min.columns = ['min_driving', 'min_walking']\n",
        "data2 = data2.merge(mobility_min, how = 'left', left_on='iso_alpha3', right_on='country')\n",
        "data2.loc[(data2.driving.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\n",
        "          ['driving', 'walking']] = data2.loc[(data2.driving.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),['min_driving', 'min_walking']].to_numpy()\n",
        "data2.drop(['country', 'min_driving', 'min_walking'], axis='columns',inplace=True)\n",
        "#data2.loc[data2.region==2,['driving', 'walking']] = data2.loc[data2.region==2,['driving', 'walking']].fillna(data2.loc[data2.iso_alpha3 == \"USA\",['driving', 'walking']].to_dict('records')[0])\n",
        "#data2.loc[data2.region==1,['driving', 'walking']] = data2.loc[data2.region==1,['driving', 'walking']].fillna(data2.loc[data2.iso_alpha3 == \"RUS\",['driving', 'walking']].to_dict('records')[0])\n",
        "#data2[['driving', 'walking']] = data2[['driving', 'walking']].fillna(data2[['driving', 'walking']].mean())\n",
        "data2.loc[data2.region != 1,'isolation'] = (data2['driving'] +  data2['walking'])/100\n",
        "\n",
        "\n",
        "data2 = data2.merge(mobility, how = 'left', left_on=('iso_alpha3','date'), right_on=('country','date'))\n",
        "mobility_min = mobility.groupby(\"country\").min().drop(['date'], axis='columns')\n",
        "mobility_min.columns = ['min_Retail & recreation', 'min_Grocery & pharmacy', 'min_Parks',\n",
        "       'min_Transit stations', 'min_Workplaces', 'min_Residential']\n",
        "data2 = data2.merge(mobility_min, how = 'left', left_on='iso_alpha3', right_on='country')\n",
        "data2.loc[(data2.Parks.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\n",
        "          ['Retail & recreation', 'Grocery & pharmacy', 'Parks', 'Transit stations', 'Workplaces', \n",
        "           'Residential']] = data2.loc[(data2.Parks.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\n",
        "                                       ['min_Retail & recreation', 'min_Grocery & pharmacy', 'min_Parks',\n",
        "       'min_Transit stations', 'min_Workplaces', 'min_Residential']].to_numpy()\n",
        "data2.drop(['country', 'min_Retail & recreation', 'min_Grocery & pharmacy', 'min_Parks',\n",
        "       'min_Transit stations', 'min_Workplaces', 'min_Residential'], axis='columns',inplace=True)\n",
        "data2.loc[data2.isolation.isnull(),'isolation'] = (data2['Retail & recreation'] +data2['Grocery & pharmacy'] +data2[ 'Parks'] \n",
        "                                                   +data2['Transit stations'] +data2[ 'Workplaces'] +data2[ 'Residential']+500)/1000     \n",
        "\n",
        "data2['isolation'] = data2['isolation'].fillna(data2['isolation'].mean())\"\"\""
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\ndata2 = data2.merge(mobility_yandex, how = \\'left\\', left_on=(\\'iso_alpha3\\',\\'date\\'), right_on=(\\'country\\',\\'date\\'))\\nmobility_min = mobility_yandex.groupby(\"country\").min().drop([\\'date\\'], axis=\\'columns\\')\\nmobility_min.columns = [\\'min_isolation\\']\\ndata2 = data2.merge(mobility_min, how = \\'left\\', left_on=\\'iso_alpha3\\', right_on=\\'country\\')\\ndata2.loc[(data2.isolation.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\\n          \\'isolation\\'] = data2.loc[(data2.isolation.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\\'min_isolation\\'].to_numpy()\\ndata2.drop([\\'country\\', \\'min_isolation\\'], axis=\\'columns\\',inplace=True)\\n#data2[\\'isolation\\'] = data2[\\'isolation\\'].fillna(data2[\\'isolation\\'].mean())\\ndata2[\\'isolation\\'] = (data2[\\'isolation\\'])/5\\n\\n\\ndata2 = data2.merge(mobility_apple, how = \\'left\\', left_on=(\\'iso_alpha3\\',\\'date\\'), right_on=(\\'country\\',\\'date\\'))\\nmobility_min = mobility_apple.groupby(\"country\").min().drop([\\'date\\'], axis=\\'columns\\')\\nmobility_min.columns = [\\'min_driving\\', \\'min_walking\\']\\ndata2 = data2.merge(mobility_min, how = \\'left\\', left_on=\\'iso_alpha3\\', right_on=\\'country\\')\\ndata2.loc[(data2.driving.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\\n          [\\'driving\\', \\'walking\\']] = data2.loc[(data2.driving.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),[\\'min_driving\\', \\'min_walking\\']].to_numpy()\\ndata2.drop([\\'country\\', \\'min_driving\\', \\'min_walking\\'], axis=\\'columns\\',inplace=True)\\n#data2.loc[data2.region==2,[\\'driving\\', \\'walking\\']] = data2.loc[data2.region==2,[\\'driving\\', \\'walking\\']].fillna(data2.loc[data2.iso_alpha3 == \"USA\",[\\'driving\\', \\'walking\\']].to_dict(\\'records\\')[0])\\n#data2.loc[data2.region==1,[\\'driving\\', \\'walking\\']] = data2.loc[data2.region==1,[\\'driving\\', \\'walking\\']].fillna(data2.loc[data2.iso_alpha3 == \"RUS\",[\\'driving\\', \\'walking\\']].to_dict(\\'records\\')[0])\\n#data2[[\\'driving\\', \\'walking\\']] = data2[[\\'driving\\', \\'walking\\']].fillna(data2[[\\'driving\\', \\'walking\\']].mean())\\ndata2.loc[data2.region != 1,\\'isolation\\'] = (data2[\\'driving\\'] +  data2[\\'walking\\'])/100\\n\\n\\ndata2 = data2.merge(mobility, how = \\'left\\', left_on=(\\'iso_alpha3\\',\\'date\\'), right_on=(\\'country\\',\\'date\\'))\\nmobility_min = mobility.groupby(\"country\").min().drop([\\'date\\'], axis=\\'columns\\')\\nmobility_min.columns = [\\'min_Retail & recreation\\', \\'min_Grocery & pharmacy\\', \\'min_Parks\\',\\n       \\'min_Transit stations\\', \\'min_Workplaces\\', \\'min_Residential\\']\\ndata2 = data2.merge(mobility_min, how = \\'left\\', left_on=\\'iso_alpha3\\', right_on=\\'country\\')\\ndata2.loc[(data2.Parks.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\\n          [\\'Retail & recreation\\', \\'Grocery & pharmacy\\', \\'Parks\\', \\'Transit stations\\', \\'Workplaces\\', \\n           \\'Residential\\']] = data2.loc[(data2.Parks.isnull())&(data2.date>pd.to_datetime(\"2020-04-01\")),\\n                                       [\\'min_Retail & recreation\\', \\'min_Grocery & pharmacy\\', \\'min_Parks\\',\\n       \\'min_Transit stations\\', \\'min_Workplaces\\', \\'min_Residential\\']].to_numpy()\\ndata2.drop([\\'country\\', \\'min_Retail & recreation\\', \\'min_Grocery & pharmacy\\', \\'min_Parks\\',\\n       \\'min_Transit stations\\', \\'min_Workplaces\\', \\'min_Residential\\'], axis=\\'columns\\',inplace=True)\\ndata2.loc[data2.isolation.isnull(),\\'isolation\\'] = (data2[\\'Retail & recreation\\'] +data2[\\'Grocery & pharmacy\\'] +data2[ \\'Parks\\'] \\n                                                   +data2[\\'Transit stations\\'] +data2[ \\'Workplaces\\'] +data2[ \\'Residential\\']+500)/1000     \\n\\ndata2[\\'isolation\\'] = data2[\\'isolation\\'].fillna(data2[\\'isolation\\'].mean())'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "v_bWNQR45Jsx",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "log_labels = True\n",
        "\n",
        "if log_labels:\n",
        "  data2.confirmed = np.log10(data2.confirmed+1)\n",
        "  data2.deaths = np.log10(data2.deaths+1)\n",
        "else:\n",
        "  data2.confirmed = data2.confirmed/10**4\n",
        "  data2.deaths = data2.deaths/10**3"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rgQAinJNj4HU",
        "colab_type": "text"
      },
      "source": [
        "## Добавить данные за прошлый день"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "okBJhPd2jyGB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "old_con = data2[\"confirmed\"].iloc[:-1]\n",
        "old_con2 = data2[\"deaths\"].iloc[:-1]\n",
        "old_recovered = data2[\"recovered\"].iloc[:-1]\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"pred_conf\"] = old_con.values\n",
        "data2[\"pred_deaths\"] = old_con2.values\n",
        "data2[\"pred_recovered\"] = old_recovered.values\n",
        "\n",
        "if log_labels:\n",
        "  data2[\"delta_conf\"] = data2[\"confirmed\"] - data2[\"pred_conf\"]\n",
        "  data2[\"delta_deaths\"] = data2[\"deaths\"] - data2[\"pred_deaths\"]\n",
        "else:\n",
        "  data2[\"delta_conf\"] = data2.confirmed/(data2[\"pred_conf\"]+0.0001) - 1\n",
        "  data2[\"delta_deaths\"] = data2.deaths/(data2[\"pred_deaths\"]+0.001) - 1\n",
        "  \n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf1\"] = old_con.values[1:] - old_con.values[:-1]\n",
        "data2[\"delta_deaths1\"] = old_con2.values[1:] - old_con2.values[:-1]\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf2\"] = old_con.values[1:-1] - old_con.values[:-2]\n",
        "data2[\"delta_deaths2\"] = old_con2.values[1:-1] - old_con2.values[:-2]\n",
        "\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf3\"] = old_con.values[1:-2] - old_con.values[:-3]\n",
        "data2[\"delta_deaths3\"] = old_con2.values[1:-2] - old_con2.values[:-3]\n",
        "\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf4\"] = old_con.values[1:-3] - old_con.values[:-4]\n",
        "data2[\"delta_deaths4\"] = old_con2.values[1:-3] - old_con2.values[:-4]\n",
        "\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf5\"] = old_con.values[1:-4] - old_con.values[:-5]\n",
        "data2[\"delta_deaths5\"] = old_con2.values[1:-4] - old_con2.values[:-5]\n",
        "\n",
        "data2 = data2.iloc[1:]\n",
        "data2[\"delta_conf6\"] = old_con.values[1:-5] - old_con.values[:-6]\n",
        "data2[\"delta_deaths6\"] = old_con2.values[1:-5] - old_con2.values[:-6]\n",
        "\n",
        "data2[\"delta_conf_sum\"] = (data2[\"delta_conf1\"] + data2[\"delta_conf2\"] + data2[\"delta_conf3\"] + data2[\"delta_conf4\"] + data2[\"delta_conf5\"] +data2[\"delta_conf6\"] ) /6\n",
        "data2[\"delta_conf_sum3\"] = (data2[\"delta_deaths1\"] + data2[\"delta_deaths2\"] + data2[\"delta_deaths3\"]) /3\n",
        "\n",
        "\n",
        "data2[\"delta_conf_count\"] = 0\n",
        "data2.loc[(data2[\"delta_conf1\"] < data2[\"delta_conf2\"]) | ((data2[\"delta_conf1\"] <0.02) & (data2[\"delta_conf1\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "data2.loc[(data2[\"delta_conf2\"] < data2[\"delta_conf3\"]) | ((data2[\"delta_conf2\"] <0.02) & (data2[\"delta_conf2\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "data2.loc[(data2[\"delta_conf3\"] < data2[\"delta_conf4\"]) | ((data2[\"delta_conf3\"] <0.02) & (data2[\"delta_conf3\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "data2.loc[(data2[\"delta_conf4\"] < data2[\"delta_conf5\"]) | ((data2[\"delta_conf4\"] <0.02) & (data2[\"delta_conf4\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "data2.loc[(data2[\"delta_conf5\"] < data2[\"delta_conf6\"]) | ((data2[\"delta_conf5\"] <0.02) & (data2[\"delta_conf5\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "\n",
        "\n",
        "\n",
        "\"\"\"data2[\"Retail_Grocery_Parks_Transit_Workplaces\"] = data2['Retail & recreation'] + data2['Grocery & pharmacy'] + data2['Parks'] + data2['Transit stations'] + data2['Workplaces']\n",
        "data2[\"Retail_Grocery_Parks_Transit_Workplaces\"] = data2[\"Retail_Grocery_Parks_Transit_Workplaces\"]/500\"\"\"\n",
        "\n",
        "data2['population_log'] = np.log10(data2.population+1)\n",
        "data2['pred_conf-population_log'] = data2['pred_conf'] - data2['population_log'] + 3"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cP_K0n4zHt6z",
        "colab_type": "text"
      },
      "source": [
        "## Добавляем признак, количество дней от первого зараженного"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gF73ZobwHs-Z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Считаем дни от первого больного\n",
        "end_date = \"2020-05-01\"\n",
        "#data2 = data2[(data2[\"date\"]<pd.to_datetime(\"2020-05-1\")) & ((data2[\"date\"]>pd.to_datetime(\"2020-04-1\")) | (data2.confirmed !=-1))]\n",
        "data4 = data2[[\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_min\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2[\"days\"] = (data2.date - data2.Date_min).dt.days\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 0.5, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_5\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_5'] = data2['Date_5'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_5\"] = (data2.date - data2.Date_5).dt.days\n",
        "#data2.loc[data2[\"days_10\"]<0,\"days_10\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 1, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_10\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_10'] = data2['Date_10'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_10\"] = (data2.date - data2.Date_10).dt.days\n",
        "#data2.loc[data2[\"days_10\"]<0,\"days_10\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 1.5, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_50\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_50'] = data2['Date_50'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_50\"] = (data2.date - data2.Date_50).dt.days\n",
        "#data2.loc[data2[\"days_10\"]<0,\"days_10\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 2, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_100\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_100'] = data2['Date_100'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_100\"] = (data2.date - data2.Date_100).dt.days\n",
        "#data2.loc[data2[\"days_100\"]<0,\"days_100\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 2.5, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_500\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_500'] = data2['Date_500'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_500\"] = (data2.date - data2.Date_500).dt.days\n",
        "#data2.loc[data2[\"days_10\"]<0,\"days_10\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2.confirmed > 3, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_1000\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_1000'] = data2['Date_1000'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_1000\"] = (data2.date - data2.Date_1000).dt.days\n",
        "#data2.loc[data2[\"days_1000\"]<0,\"days_1000\"] = 0\n",
        "\n",
        "data2[\"days_mart\"] = (data2.date - pd.to_datetime(\"2020-03-1\")).dt.days\n",
        "data2[\"days_after_Quarantine\"] = (data2.date - pd.to_datetime(data2.Quarantine)).dt.days - 14\n",
        "data2.loc[data2[\"days_after_Quarantine\"]<0,\"days_after_Quarantine\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2['pred_conf-population_log'] > 0, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_1_1000\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_1_1000'] = data2['Date_1_1000'].fillna(pd.to_datetime(\"2020-04-01\"))\n",
        "data2[\"days_1_1000\"] = (data2.date - data2[\"Date_1_1000\"]).dt.days\n",
        "#data2.loc[data2[\"days_1_1000\"]<0,\"days_1_1000\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2['pred_conf-population_log'] > -1, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_1_100\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_1_100'] = data2['Date_1_100'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_1_100\"] = (data2.date - data2[\"Date_1_100\"]).dt.days\n",
        "#data2.loc[data2[\"days_1_100\"]<0,\"days_1_100\"] = 0\n",
        "\n",
        "data4 = data2.loc[data2['pred_conf-population_log'] > -2, [\"countries\",\"date\"]].groupby(\"countries\").min()\n",
        "data4.columns = [\"Date_1_10000\"]\n",
        "data2 = data2.merge(data4, how = 'left', left_on='countries', right_on='countries')\n",
        "data2['Date_1_10000'] = data2['Date_1_10000'].fillna(pd.to_datetime(end_date))\n",
        "data2[\"days_1_10000\"] = (data2.date - data2[\"Date_1_10000\"]).dt.days\n",
        "#data2.loc[data2[\"days_1_10000\"]<0,\"days_1_10000\"] = 0\n",
        "\n",
        "data2.drop(['Date_min','Date_10','Date_100','Date_1000', 'Date_5','Date_50','Date_500', 'Date_1_1000', 'Date_1_100', 'Date_1_10000'], axis='columns',inplace=True)\n",
        "data2.date = data2[\"date\"].apply(lambda x: pd.Series(x.strftime(\"%m-%d\")))\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DdugXNiT0_UW",
        "colab_type": "code",
        "outputId": "3caad823-f53d-4860-c6bf-f5207c41ed78",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 326
        }
      },
      "source": [
        "Scale = True\n",
        "\n",
        "if Scale:\n",
        "  data2.density = data2.density/data2.density.max()\n",
        "  data2.fertility_rate = data2.fertility_rate/data2.fertility_rate.max()\n",
        "  data2.land_area = np.log10(data2.land_area+1)\n",
        "  data2.land_area = data2.land_area/data2.land_area.max()\n",
        "  data2.median_age = data2.median_age/data2.median_age.max()\n",
        "  data2.migrants = data2.migrants - data2.migrants.min()\n",
        "  data2.migrants = np.log10(data2.migrants+1)\n",
        "  data2.migrants = data2.migrants/data2.migrants.max()\n",
        "  data2.population = data2.population/data2.population.max()\n",
        "  data2.Health_GDP = data2.Health_GDP/data2.Health_GDP.max()\n",
        "  data2.Health_USD = data2.Health_USD/data2.Health_USD.max()\n",
        "  data2.Physicians = data2.Physicians/data2.Physicians.max()\n",
        "  data2.tests = data2.tests/data2.tests.max()\n",
        "  data2.testpop = data2.testpop/data2.testpop.max()\n",
        "  data2.gatheringlimit = data2.gatheringlimit/data2.gatheringlimit.max()\n",
        "  data2.hospibed = data2.hospibed/data2.hospibed.max()\n",
        "  data2.healthperpop = data2.healthperpop/data2.healthperpop.max()\n",
        "  data2.Nurse = data2.Nurse/data2.Nurse.max()\n",
        "  data2[\"Age_old/new\"] = data2[\"Age_old/new\"]/data2[\"Age_old/new\"].max()\n",
        "  data2.Smoking = data2.Smoking/50\n",
        "  data2.days = data2.days/100\n",
        "  data2.days_10 = data2.days_10/100\n",
        "  data2.days_100 = data2.days_100/100\n",
        "  data2.days_1000 = data2.days_1000/100\n",
        "  data2.days_5 = data2.days_5/100\n",
        "  data2.days_50 = data2.days_50/100\n",
        "  data2.days_500 = data2.days_500/100\n",
        "  data2.days_1_100 = data2.days_1_100/100\n",
        "  data2.days_1_1000 = data2.days_1_1000/100 \n",
        "  data2.days_1_10000 = data2.days_1_10000/100    \n",
        "  data2.days_mart = data2.days_mart/100\n",
        "\n",
        "  data2.days_after_Quarantine = data2.days_after_Quarantine/100\n",
        "  sc_days = 100\n",
        "else:\n",
        "  sc_days = 1\n",
        "data2[data2.iso_alpha3 == \"RUS\"].iloc[70:75]"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>countries</th>\n",
              "      <th>date</th>\n",
              "      <th>confirmed</th>\n",
              "      <th>deaths</th>\n",
              "      <th>recovered</th>\n",
              "      <th>iso_alpha3</th>\n",
              "      <th>density</th>\n",
              "      <th>fertility_rate</th>\n",
              "      <th>land_area</th>\n",
              "      <th>median_age</th>\n",
              "      <th>migrants</th>\n",
              "      <th>population</th>\n",
              "      <th>urban_pop_rate</th>\n",
              "      <th>world_share</th>\n",
              "      <th>Quarantine</th>\n",
              "      <th>Health_GDP</th>\n",
              "      <th>Health_USD</th>\n",
              "      <th>Physicians</th>\n",
              "      <th>Nurse</th>\n",
              "      <th>Age_old/new</th>\n",
              "      <th>Smoking</th>\n",
              "      <th>tests</th>\n",
              "      <th>testpop</th>\n",
              "      <th>gatheringlimit</th>\n",
              "      <th>hospibed</th>\n",
              "      <th>healthperpop</th>\n",
              "      <th>region</th>\n",
              "      <th>pred_conf</th>\n",
              "      <th>pred_deaths</th>\n",
              "      <th>pred_recovered</th>\n",
              "      <th>delta_conf</th>\n",
              "      <th>delta_deaths</th>\n",
              "      <th>delta_conf1</th>\n",
              "      <th>delta_deaths1</th>\n",
              "      <th>delta_conf2</th>\n",
              "      <th>delta_deaths2</th>\n",
              "      <th>delta_conf3</th>\n",
              "      <th>delta_deaths3</th>\n",
              "      <th>delta_conf4</th>\n",
              "      <th>delta_deaths4</th>\n",
              "      <th>delta_conf5</th>\n",
              "      <th>delta_deaths5</th>\n",
              "      <th>delta_conf6</th>\n",
              "      <th>delta_deaths6</th>\n",
              "      <th>delta_conf_sum</th>\n",
              "      <th>delta_conf_sum3</th>\n",
              "      <th>delta_conf_count</th>\n",
              "      <th>population_log</th>\n",
              "      <th>pred_conf-population_log</th>\n",
              "      <th>days</th>\n",
              "      <th>days_5</th>\n",
              "      <th>days_10</th>\n",
              "      <th>days_50</th>\n",
              "      <th>days_100</th>\n",
              "      <th>days_500</th>\n",
              "      <th>days_1000</th>\n",
              "      <th>days_mart</th>\n",
              "      <th>days_after_Quarantine</th>\n",
              "      <th>days_1_1000</th>\n",
              "      <th>days_1_100</th>\n",
              "      <th>days_1_10000</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>17969</th>\n",
              "      <td>Russia</td>\n",
              "      <td>04-10</td>\n",
              "      <td>4.076203</td>\n",
              "      <td>1.977724</td>\n",
              "      <td>2.900913</td>\n",
              "      <td>RUS</td>\n",
              "      <td>0.000342</td>\n",
              "      <td>0.257143</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.833333</td>\n",
              "      <td>0.9542</td>\n",
              "      <td>0.101391</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.0187</td>\n",
              "      <td>2020-03-30</td>\n",
              "      <td>0.309942</td>\n",
              "      <td>0.047529</td>\n",
              "      <td>0.487805</td>\n",
              "      <td>0.423645</td>\n",
              "      <td>0.391229</td>\n",
              "      <td>0.786</td>\n",
              "      <td>0.356813</td>\n",
              "      <td>0.003023</td>\n",
              "      <td>0.070348</td>\n",
              "      <td>0.288732</td>\n",
              "      <td>0.000089</td>\n",
              "      <td>0</td>\n",
              "      <td>4.005695</td>\n",
              "      <td>1.886491</td>\n",
              "      <td>2.844477</td>\n",
              "      <td>0.070508</td>\n",
              "      <td>0.091233</td>\n",
              "      <td>0.067526</td>\n",
              "      <td>0.080311</td>\n",
              "      <td>0.063224</td>\n",
              "      <td>0.035328</td>\n",
              "      <td>0.072582</td>\n",
              "      <td>0.089611</td>\n",
              "      <td>0.070774</td>\n",
              "      <td>0.018483</td>\n",
              "      <td>0.056544</td>\n",
              "      <td>0.019305</td>\n",
              "      <td>0.056997</td>\n",
              "      <td>0.099385</td>\n",
              "      <td>0.064608</td>\n",
              "      <td>0.068416</td>\n",
              "      <td>0.4</td>\n",
              "      <td>8.164158</td>\n",
              "      <td>-1.158463</td>\n",
              "      <td>0.70</td>\n",
              "      <td>0.39</td>\n",
              "      <td>0.35</td>\n",
              "      <td>0.28</td>\n",
              "      <td>0.24</td>\n",
              "      <td>0.19</td>\n",
              "      <td>0.14</td>\n",
              "      <td>0.40</td>\n",
              "      <td>0.00</td>\n",
              "      <td>-0.26</td>\n",
              "      <td>-0.03</td>\n",
              "      <td>0.11</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17970</th>\n",
              "      <td>Russia</td>\n",
              "      <td>04-11</td>\n",
              "      <td>4.133060</td>\n",
              "      <td>2.029384</td>\n",
              "      <td>3.019532</td>\n",
              "      <td>RUS</td>\n",
              "      <td>0.000342</td>\n",
              "      <td>0.257143</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.833333</td>\n",
              "      <td>0.9542</td>\n",
              "      <td>0.101391</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.0187</td>\n",
              "      <td>2020-03-30</td>\n",
              "      <td>0.309942</td>\n",
              "      <td>0.047529</td>\n",
              "      <td>0.487805</td>\n",
              "      <td>0.423645</td>\n",
              "      <td>0.391229</td>\n",
              "      <td>0.786</td>\n",
              "      <td>0.356813</td>\n",
              "      <td>0.003023</td>\n",
              "      <td>0.070348</td>\n",
              "      <td>0.288732</td>\n",
              "      <td>0.000089</td>\n",
              "      <td>0</td>\n",
              "      <td>4.076203</td>\n",
              "      <td>1.977724</td>\n",
              "      <td>2.900913</td>\n",
              "      <td>0.056856</td>\n",
              "      <td>0.051660</td>\n",
              "      <td>0.070508</td>\n",
              "      <td>0.091233</td>\n",
              "      <td>0.067526</td>\n",
              "      <td>0.080311</td>\n",
              "      <td>0.063224</td>\n",
              "      <td>0.035328</td>\n",
              "      <td>0.072582</td>\n",
              "      <td>0.089611</td>\n",
              "      <td>0.070774</td>\n",
              "      <td>0.018483</td>\n",
              "      <td>0.056544</td>\n",
              "      <td>0.019305</td>\n",
              "      <td>0.066860</td>\n",
              "      <td>0.068957</td>\n",
              "      <td>0.2</td>\n",
              "      <td>8.164158</td>\n",
              "      <td>-1.087954</td>\n",
              "      <td>0.71</td>\n",
              "      <td>0.40</td>\n",
              "      <td>0.36</td>\n",
              "      <td>0.29</td>\n",
              "      <td>0.25</td>\n",
              "      <td>0.20</td>\n",
              "      <td>0.15</td>\n",
              "      <td>0.41</td>\n",
              "      <td>0.00</td>\n",
              "      <td>-0.25</td>\n",
              "      <td>-0.02</td>\n",
              "      <td>0.12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17971</th>\n",
              "      <td>Russia</td>\n",
              "      <td>04-12</td>\n",
              "      <td>4.197859</td>\n",
              "      <td>2.117271</td>\n",
              "      <td>3.111263</td>\n",
              "      <td>RUS</td>\n",
              "      <td>0.000342</td>\n",
              "      <td>0.257143</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.833333</td>\n",
              "      <td>0.9542</td>\n",
              "      <td>0.101391</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.0187</td>\n",
              "      <td>2020-03-30</td>\n",
              "      <td>0.309942</td>\n",
              "      <td>0.047529</td>\n",
              "      <td>0.487805</td>\n",
              "      <td>0.423645</td>\n",
              "      <td>0.391229</td>\n",
              "      <td>0.786</td>\n",
              "      <td>0.356813</td>\n",
              "      <td>0.003023</td>\n",
              "      <td>0.070348</td>\n",
              "      <td>0.288732</td>\n",
              "      <td>0.000089</td>\n",
              "      <td>0</td>\n",
              "      <td>4.133060</td>\n",
              "      <td>2.029384</td>\n",
              "      <td>3.019532</td>\n",
              "      <td>0.064800</td>\n",
              "      <td>0.087888</td>\n",
              "      <td>0.056856</td>\n",
              "      <td>0.051660</td>\n",
              "      <td>0.070508</td>\n",
              "      <td>0.091233</td>\n",
              "      <td>0.067526</td>\n",
              "      <td>0.080311</td>\n",
              "      <td>0.063224</td>\n",
              "      <td>0.035328</td>\n",
              "      <td>0.072582</td>\n",
              "      <td>0.089611</td>\n",
              "      <td>0.070774</td>\n",
              "      <td>0.018483</td>\n",
              "      <td>0.066912</td>\n",
              "      <td>0.074401</td>\n",
              "      <td>0.4</td>\n",
              "      <td>8.164158</td>\n",
              "      <td>-1.031098</td>\n",
              "      <td>0.72</td>\n",
              "      <td>0.41</td>\n",
              "      <td>0.37</td>\n",
              "      <td>0.30</td>\n",
              "      <td>0.26</td>\n",
              "      <td>0.21</td>\n",
              "      <td>0.16</td>\n",
              "      <td>0.42</td>\n",
              "      <td>0.00</td>\n",
              "      <td>-0.24</td>\n",
              "      <td>-0.01</td>\n",
              "      <td>0.13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17972</th>\n",
              "      <td>Russia</td>\n",
              "      <td>04-13</td>\n",
              "      <td>4.263139</td>\n",
              "      <td>2.173186</td>\n",
              "      <td>3.167613</td>\n",
              "      <td>RUS</td>\n",
              "      <td>0.000342</td>\n",
              "      <td>0.257143</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.833333</td>\n",
              "      <td>0.9542</td>\n",
              "      <td>0.101391</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.0187</td>\n",
              "      <td>2020-03-30</td>\n",
              "      <td>0.309942</td>\n",
              "      <td>0.047529</td>\n",
              "      <td>0.487805</td>\n",
              "      <td>0.423645</td>\n",
              "      <td>0.391229</td>\n",
              "      <td>0.786</td>\n",
              "      <td>0.356813</td>\n",
              "      <td>0.003023</td>\n",
              "      <td>0.070348</td>\n",
              "      <td>0.288732</td>\n",
              "      <td>0.000089</td>\n",
              "      <td>0</td>\n",
              "      <td>4.197859</td>\n",
              "      <td>2.117271</td>\n",
              "      <td>3.111263</td>\n",
              "      <td>0.065280</td>\n",
              "      <td>0.055915</td>\n",
              "      <td>0.064800</td>\n",
              "      <td>0.087888</td>\n",
              "      <td>0.056856</td>\n",
              "      <td>0.051660</td>\n",
              "      <td>0.070508</td>\n",
              "      <td>0.091233</td>\n",
              "      <td>0.067526</td>\n",
              "      <td>0.080311</td>\n",
              "      <td>0.063224</td>\n",
              "      <td>0.035328</td>\n",
              "      <td>0.072582</td>\n",
              "      <td>0.089611</td>\n",
              "      <td>0.065916</td>\n",
              "      <td>0.076927</td>\n",
              "      <td>0.4</td>\n",
              "      <td>8.164158</td>\n",
              "      <td>-0.966299</td>\n",
              "      <td>0.73</td>\n",
              "      <td>0.42</td>\n",
              "      <td>0.38</td>\n",
              "      <td>0.31</td>\n",
              "      <td>0.27</td>\n",
              "      <td>0.22</td>\n",
              "      <td>0.17</td>\n",
              "      <td>0.43</td>\n",
              "      <td>0.00</td>\n",
              "      <td>-0.23</td>\n",
              "      <td>0.00</td>\n",
              "      <td>0.14</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17973</th>\n",
              "      <td>Russia</td>\n",
              "      <td>04-14</td>\n",
              "      <td>4.324344</td>\n",
              "      <td>2.232996</td>\n",
              "      <td>3.229170</td>\n",
              "      <td>RUS</td>\n",
              "      <td>0.000342</td>\n",
              "      <td>0.257143</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.833333</td>\n",
              "      <td>0.9542</td>\n",
              "      <td>0.101391</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.0187</td>\n",
              "      <td>2020-03-30</td>\n",
              "      <td>0.309942</td>\n",
              "      <td>0.047529</td>\n",
              "      <td>0.487805</td>\n",
              "      <td>0.423645</td>\n",
              "      <td>0.391229</td>\n",
              "      <td>0.786</td>\n",
              "      <td>0.356813</td>\n",
              "      <td>0.003023</td>\n",
              "      <td>0.070348</td>\n",
              "      <td>0.288732</td>\n",
              "      <td>0.000089</td>\n",
              "      <td>0</td>\n",
              "      <td>4.263139</td>\n",
              "      <td>2.173186</td>\n",
              "      <td>3.167613</td>\n",
              "      <td>0.061205</td>\n",
              "      <td>0.059810</td>\n",
              "      <td>0.065280</td>\n",
              "      <td>0.055915</td>\n",
              "      <td>0.064800</td>\n",
              "      <td>0.087888</td>\n",
              "      <td>0.056856</td>\n",
              "      <td>0.051660</td>\n",
              "      <td>0.070508</td>\n",
              "      <td>0.091233</td>\n",
              "      <td>0.067526</td>\n",
              "      <td>0.080311</td>\n",
              "      <td>0.063224</td>\n",
              "      <td>0.035328</td>\n",
              "      <td>0.064699</td>\n",
              "      <td>0.065154</td>\n",
              "      <td>0.2</td>\n",
              "      <td>8.164158</td>\n",
              "      <td>-0.901019</td>\n",
              "      <td>0.74</td>\n",
              "      <td>0.43</td>\n",
              "      <td>0.39</td>\n",
              "      <td>0.32</td>\n",
              "      <td>0.28</td>\n",
              "      <td>0.23</td>\n",
              "      <td>0.18</td>\n",
              "      <td>0.44</td>\n",
              "      <td>0.01</td>\n",
              "      <td>-0.22</td>\n",
              "      <td>0.01</td>\n",
              "      <td>0.15</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "      countries   date  confirmed  ...  days_1_1000  days_1_100 days_1_10000\n",
              "17969    Russia  04-10   4.076203  ...        -0.26       -0.03         0.11\n",
              "17970    Russia  04-11   4.133060  ...        -0.25       -0.02         0.12\n",
              "17971    Russia  04-12   4.197859  ...        -0.24       -0.01         0.13\n",
              "17972    Russia  04-13   4.263139  ...        -0.23        0.00         0.14\n",
              "17973    Russia  04-14   4.324344  ...        -0.22        0.01         0.15\n",
              "\n",
              "[5 rows x 61 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LDR-u7ggH5Gz",
        "colab_type": "text"
      },
      "source": [
        "# Отделяем данные на трейн и тест"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eu0gPjRaKzj5",
        "colab_type": "code",
        "outputId": "4f2eddd7-cb5b-4550-b76f-29ffa2c3f0d6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 456
        }
      },
      "source": [
        "#Выберите день начала приватных данных начиная с 1 марта\n",
        "days_x = 56 #39\n",
        "iterat = 10\n",
        "#type_ans_delta - Предсказывать общее число, или изменение за день\n",
        "type_ans_delta = True\n",
        "\n",
        "#ВЫберите данные(только регионы России, только страны, всё)\n",
        "type_reg = \"all\"\n",
        "#type_reg = \"country\"\n",
        "#type_reg = \"region\"\n",
        "#type_reg = \"no_USA\"\n",
        "\n",
        "param_list =['countries', 'confirmed', 'deaths', 'iso_alpha3', 'delta_conf', 'delta_deaths', 'region',\n",
        "              'pred_conf', 'pred_deaths', 'delta_conf1', 'delta_deaths1',\n",
        "             'delta_conf_count',\n",
        "             'delta_deaths2','delta_deaths3','delta_deaths4','delta_deaths5','delta_deaths6',\n",
        "\n",
        "             'urban_pop_rate',             \n",
        "             'pred_conf-population_log',              \n",
        "             'population_log',\n",
        "             'land_area',\n",
        "\n",
        "             'days',\n",
        "             'days_10',\n",
        "             'days_100',\n",
        "             'days_1000',\n",
        "             'days_5',\n",
        "             'days_50',\n",
        "             'days_500',\n",
        "             'days_1_100', \n",
        "             'days_1_1000',              \n",
        "             'days_1_10000', \n",
        "\n",
        "             'delta_conf2', \n",
        "             'delta_conf3', \n",
        "             'delta_conf4', \n",
        "             'delta_conf5', \n",
        "             'delta_conf6', \n",
        "             \"delta_conf_sum\",\n",
        "            'delta_conf_sum3',\n",
        "             'days_mart']\n",
        "\n",
        "data3 = data2[param_list]\n",
        "\n",
        "\n",
        "if type_reg == \"region\":\n",
        "  data3 = data3[data3['region'] == 1]\n",
        "elif  type_reg == \"country\":\n",
        "  data3 = data3[data3['region'] == 0]\n",
        "elif  type_reg == \"no_USA\":\n",
        "  data3 = data3[(data3['region'] == 0)|(data3['region'] == 1)]  \n",
        "else:\n",
        "  data3 = data3\n",
        "\n",
        "model_Confirmed, model_Death = {}, {}\n",
        "\n",
        "#Представляем страны в one_hot\n",
        "#one_hot = pd.get_dummies(data3['iso_alpha3'])\n",
        "#data3 = data3.join(one_hot)\n",
        "\n",
        "#Приватные данные\n",
        "type_reg_mean = \"no_USA\"\n",
        "\n",
        "new_data = data3[data3.days_mart >= (days_x)/sc_days]\n",
        "if type_reg_mean == \"region\":\n",
        "  new_data = new_data[new_data['region'] == 1]\n",
        "elif  type_reg_mean == \"country\":\n",
        "  new_data = new_data[new_data['region'] == 0]\n",
        "elif  type_reg_mean == \"no_USA\":\n",
        "  new_data = new_data[(new_data['region'] == 0)|(new_data['region'] == 1)]  \n",
        "\n",
        "data_x = new_data[new_data.days_mart == (days_x)/sc_days]\n",
        "old_data = data3[(data3.days_mart <= (days_x)/sc_days) & (data3.confirmed > 0)]\n",
        "#old_data = old_data[(old_data.countries != 'Russia')&(old_data.countries != 'Korea, South')]\n",
        "#old_data = old_data[old_data.delta_conf>0]\n",
        "\n",
        "if type_ans_delta:\n",
        "  #train_labels = old_data.delta_conf - old_data.delta_conf_sum\n",
        "  train_labels = old_data.delta_conf\n",
        "  train_death = old_data.delta_deaths\n",
        "else:\n",
        "  train_labels = old_data.confirmed\n",
        "  train_death = old_data.deaths\n",
        "\n",
        "#Убираем ответ из данных\n",
        "train_data = old_data.drop(['iso_alpha3','region','confirmed','countries','deaths',\"delta_conf\",\"delta_deaths\"], axis='columns')\n",
        "old_data"
      ],
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>countries</th>\n",
              "      <th>confirmed</th>\n",
              "      <th>deaths</th>\n",
              "      <th>iso_alpha3</th>\n",
              "      <th>delta_conf</th>\n",
              "      <th>delta_deaths</th>\n",
              "      <th>region</th>\n",
              "      <th>pred_conf</th>\n",
              "      <th>pred_deaths</th>\n",
              "      <th>delta_conf1</th>\n",
              "      <th>delta_deaths1</th>\n",
              "      <th>delta_conf_count</th>\n",
              "      <th>delta_deaths2</th>\n",
              "      <th>delta_deaths3</th>\n",
              "      <th>delta_deaths4</th>\n",
              "      <th>delta_deaths5</th>\n",
              "      <th>delta_deaths6</th>\n",
              "      <th>urban_pop_rate</th>\n",
              "      <th>pred_conf-population_log</th>\n",
              "      <th>population_log</th>\n",
              "      <th>land_area</th>\n",
              "      <th>days</th>\n",
              "      <th>days_10</th>\n",
              "      <th>days_100</th>\n",
              "      <th>days_1000</th>\n",
              "      <th>days_1_100</th>\n",
              "      <th>days_1_1000</th>\n",
              "      <th>delta_conf2</th>\n",
              "      <th>delta_conf3</th>\n",
              "      <th>delta_conf4</th>\n",
              "      <th>delta_conf5</th>\n",
              "      <th>delta_conf6</th>\n",
              "      <th>delta_conf_sum</th>\n",
              "      <th>delta_conf_sum3</th>\n",
              "      <th>days_mart</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Afghanistan</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>AFG</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.25</td>\n",
              "      <td>-4.289236</td>\n",
              "      <td>7.590266</td>\n",
              "      <td>0.806021</td>\n",
              "      <td>0.00</td>\n",
              "      <td>-0.12</td>\n",
              "      <td>-0.25</td>\n",
              "      <td>-0.49</td>\n",
              "      <td>-0.60</td>\n",
              "      <td>-0.30</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.01</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Afghanistan</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>AFG</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.25</td>\n",
              "      <td>-4.289236</td>\n",
              "      <td>7.590266</td>\n",
              "      <td>0.806021</td>\n",
              "      <td>0.01</td>\n",
              "      <td>-0.11</td>\n",
              "      <td>-0.24</td>\n",
              "      <td>-0.48</td>\n",
              "      <td>-0.59</td>\n",
              "      <td>-0.29</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.02</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Afghanistan</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>AFG</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.25</td>\n",
              "      <td>-4.289236</td>\n",
              "      <td>7.590266</td>\n",
              "      <td>0.806021</td>\n",
              "      <td>0.02</td>\n",
              "      <td>-0.10</td>\n",
              "      <td>-0.23</td>\n",
              "      <td>-0.47</td>\n",
              "      <td>-0.58</td>\n",
              "      <td>-0.28</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.03</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Afghanistan</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>AFG</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.25</td>\n",
              "      <td>-4.289236</td>\n",
              "      <td>7.590266</td>\n",
              "      <td>0.806021</td>\n",
              "      <td>0.03</td>\n",
              "      <td>-0.09</td>\n",
              "      <td>-0.22</td>\n",
              "      <td>-0.46</td>\n",
              "      <td>-0.57</td>\n",
              "      <td>-0.27</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.04</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Afghanistan</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>AFG</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.301030</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.25</td>\n",
              "      <td>-4.289236</td>\n",
              "      <td>7.590266</td>\n",
              "      <td>0.806021</td>\n",
              "      <td>0.04</td>\n",
              "      <td>-0.08</td>\n",
              "      <td>-0.21</td>\n",
              "      <td>-0.45</td>\n",
              "      <td>-0.56</td>\n",
              "      <td>-0.26</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.05</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23491</th>\n",
              "      <td>Zimbabwe</td>\n",
              "      <td>1.462398</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>ZWE</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.09691</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.462398</td>\n",
              "      <td>0.60206</td>\n",
              "      <td>0.047425</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.6</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.38</td>\n",
              "      <td>-2.709706</td>\n",
              "      <td>7.172104</td>\n",
              "      <td>0.774517</td>\n",
              "      <td>0.33</td>\n",
              "      <td>0.16</td>\n",
              "      <td>-0.09</td>\n",
              "      <td>-0.09</td>\n",
              "      <td>-0.09</td>\n",
              "      <td>0.21</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.017033</td>\n",
              "      <td>0.017729</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.013698</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.52</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23492</th>\n",
              "      <td>Zimbabwe</td>\n",
              "      <td>1.462398</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>ZWE</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.462398</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.09691</td>\n",
              "      <td>0.6</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.38</td>\n",
              "      <td>-2.709706</td>\n",
              "      <td>7.172104</td>\n",
              "      <td>0.774517</td>\n",
              "      <td>0.34</td>\n",
              "      <td>0.17</td>\n",
              "      <td>-0.08</td>\n",
              "      <td>-0.08</td>\n",
              "      <td>-0.08</td>\n",
              "      <td>0.22</td>\n",
              "      <td>0.047425</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.017033</td>\n",
              "      <td>0.017729</td>\n",
              "      <td>0.013698</td>\n",
              "      <td>0.032303</td>\n",
              "      <td>0.53</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23493</th>\n",
              "      <td>Zimbabwe</td>\n",
              "      <td>1.477121</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>ZWE</td>\n",
              "      <td>0.014723</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.462398</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.09691</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.38</td>\n",
              "      <td>-2.709706</td>\n",
              "      <td>7.172104</td>\n",
              "      <td>0.774517</td>\n",
              "      <td>0.35</td>\n",
              "      <td>0.18</td>\n",
              "      <td>-0.07</td>\n",
              "      <td>-0.07</td>\n",
              "      <td>-0.07</td>\n",
              "      <td>0.23</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.047425</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.017033</td>\n",
              "      <td>0.010743</td>\n",
              "      <td>0.032303</td>\n",
              "      <td>0.54</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23494</th>\n",
              "      <td>Zimbabwe</td>\n",
              "      <td>1.505150</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>ZWE</td>\n",
              "      <td>0.028029</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.477121</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>0.014723</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.09691</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.38</td>\n",
              "      <td>-2.694983</td>\n",
              "      <td>7.172104</td>\n",
              "      <td>0.774517</td>\n",
              "      <td>0.36</td>\n",
              "      <td>0.19</td>\n",
              "      <td>-0.06</td>\n",
              "      <td>-0.06</td>\n",
              "      <td>-0.06</td>\n",
              "      <td>0.24</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.047425</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.010358</td>\n",
              "      <td>0.032303</td>\n",
              "      <td>0.55</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23495</th>\n",
              "      <td>Zimbabwe</td>\n",
              "      <td>1.505150</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>ZWE</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1.505150</td>\n",
              "      <td>0.69897</td>\n",
              "      <td>0.028029</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.4</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.00000</td>\n",
              "      <td>0.09691</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.38</td>\n",
              "      <td>-2.666954</td>\n",
              "      <td>7.172104</td>\n",
              "      <td>0.774517</td>\n",
              "      <td>0.37</td>\n",
              "      <td>0.20</td>\n",
              "      <td>-0.05</td>\n",
              "      <td>-0.05</td>\n",
              "      <td>-0.05</td>\n",
              "      <td>0.25</td>\n",
              "      <td>0.014723</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.047425</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.015029</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.56</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>14835 rows × 35 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "         countries  confirmed  ...  delta_conf_sum3 days_mart\n",
              "0      Afghanistan   0.301030  ...         0.000000      0.01\n",
              "1      Afghanistan   0.301030  ...         0.000000      0.02\n",
              "2      Afghanistan   0.301030  ...         0.000000      0.03\n",
              "3      Afghanistan   0.301030  ...         0.000000      0.04\n",
              "4      Afghanistan   0.301030  ...         0.000000      0.05\n",
              "...            ...        ...  ...              ...       ...\n",
              "23491     Zimbabwe   1.462398  ...         0.000000      0.52\n",
              "23492     Zimbabwe   1.462398  ...         0.032303      0.53\n",
              "23493     Zimbabwe   1.477121  ...         0.032303      0.54\n",
              "23494     Zimbabwe   1.505150  ...         0.032303      0.55\n",
              "23495     Zimbabwe   1.505150  ...         0.000000      0.56\n",
              "\n",
              "[14835 rows x 35 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ye2UK5_7Yags",
        "colab_type": "text"
      },
      "source": [
        "# Модели"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cJwzcZOrYZdB",
        "colab_type": "text"
      },
      "source": [
        "## Обучаем Линейная регрессия"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8dRnl1jKdE3t",
        "colab_type": "text"
      },
      "source": [
        "### Функция построения признаков"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iebPwlF3LHas",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Визуализируем важность признаков\n",
        "def visualize_coefficients(classifier, feature_names):\n",
        "  n_top_features = len(feature_names)//2\n",
        "  # get coefficients with large absolute values \n",
        "  coef = classifier.coef_.ravel()\n",
        "  positive_coefficients = np.argsort(coef)[-n_top_features:]\n",
        "  negative_coefficients = np.argsort(coef)[:n_top_features]\n",
        "  interesting_coefficients = np.hstack([negative_coefficients, positive_coefficients])\n",
        "  # plot them\n",
        "  plt.figure(figsize=(15, 5))\n",
        "  colors = [\"red\" if c < 0 else \"blue\" for c in coef[interesting_coefficients]]\n",
        "  plt.bar(np.arange(2 * n_top_features), coef[interesting_coefficients], color=colors)\n",
        "  feature_names = np.array(feature_names)\n",
        "  plt.xticks(np.arange(0, 1 + 2 * n_top_features), feature_names[interesting_coefficients], rotation=60, ha=\"right\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CYmMbkHEdO7r",
        "colab_type": "text"
      },
      "source": [
        "### Ridge"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2MRvwAS3dTVO",
        "colab_type": "code",
        "outputId": "7af9e920-4040-45d8-87e4-99a8596f2581",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 469
        }
      },
      "source": [
        "model_Confirmed[\"Ridge\"] = linear_model.Ridge().fit(train_data, train_labels)\n",
        "model_Death[\"Ridge\"] = linear_model.Ridge().fit(train_data, train_death)\n",
        "visualize_coefficients(model_Confirmed[\"Ridge\"], train_data.columns)\n",
        "del model_Confirmed[\"Ridge\"]"
      ],
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xyXZZH2kdU12",
        "colab_type": "text"
      },
      "source": [
        "### Lasso"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yPOkf1uAdYgl",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#model_Confirmed[\"Lasso\"] = linear_model.Lasso().fit(train_data, train_labels)\n",
        "#model_Death[\"Lasso\"] = linear_model.Lasso().fit(train_data, train_death)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pCslqzwMdaYJ",
        "colab_type": "text"
      },
      "source": [
        "### SGDRegressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nXfWJhS0V4T8",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#model_Confirmed[\"SGDRegressor\"] = linear_model.SGDRegressor().fit(train_data, train_labels)\n",
        "#model_Death[\"SGDRegressor\"] = linear_model.SGDRegressor().fit(train_data, train_death)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LjSqy9eb4p5g",
        "colab_type": "text"
      },
      "source": [
        "## Нейронная сеть"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Mq548kJ3VCD_",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем MLPRegressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "r_sl8LEtqP-j",
        "colab_type": "code",
        "outputId": "17fdbc1e-fa08-4633-d6b5-ebb26026a1b8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"\n",
        "model_Confirmed[\"MLPRegressor\"] = MLPRegressor(hidden_layer_sizes=(100, 100),max_iter=5000,\n",
        "                                 learning_rate_init=0.01, random_state=0,\n",
        "                                 early_stopping=True).fit(train_data, train_labels)\n",
        "                                                            \n",
        "model_Death[\"MLPRegressor\"] = MLPRegressor(hidden_layer_sizes=(100, 100),max_iter=5000,\n",
        "                                 learning_rate_init=0.01, random_state=0,\n",
        "                                 early_stopping=True).fit(train_data, train_death)\n",
        "\"\"\""
      ],
      "execution_count": 45,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\nmodel_Confirmed[\"MLPRegressor\"] = MLPRegressor(hidden_layer_sizes=(100, 100),max_iter=5000,\\n                                 learning_rate_init=0.01, random_state=0,\\n                                 early_stopping=True).fit(train_data, train_labels)\\n                                                            \\nmodel_Death[\"MLPRegressor\"] = MLPRegressor(hidden_layer_sizes=(100, 100),max_iter=5000,\\n                                 learning_rate_init=0.01, random_state=0,\\n                                 early_stopping=True).fit(train_data, train_death)\\n'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 45
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AH48ukTguTMm",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем Keras"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vP3B5_Qg63uz",
        "colab_type": "code",
        "outputId": "25984b53-7e84-439a-ec01-ac4c9c92724b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"def build_model():\n",
        "  model = keras.Sequential([\n",
        "    layers.Dense(128, activation='relu', input_shape=[len(train_data.keys())]),\n",
        "    keras.layers.Dropout(0.2),\n",
        "    layers.Dense(32, activation='relu', kernel_regularizer=keras.regularizers.l2(0.1)),\n",
        "    keras.layers.Dropout(0.2),    \n",
        "    layers.Dense(1)\n",
        "  ])\n",
        "  model.compile(loss='mse', optimizer='adam', metrics='mse')\n",
        "  return model\n",
        "\n",
        "EPOCHS = 100\n",
        "model_Confirmed[\"keras\"] = build_model()\n",
        "history = model_Confirmed[\"keras\"].fit(train_data, train_labels, epochs=EPOCHS, validation_split = 0.1, verbose=0)\n",
        "model_Death[\"keras\"] = build_model()\n",
        "history = model_Death[\"keras\"].fit(train_data, train_death, epochs=EPOCHS, validation_split = 0.1, verbose=0)\n",
        "plt.rcParams['figure.figsize'] = [20, 10]\n",
        "hist = pd.DataFrame(history.history)\n",
        "hist['epoch'] = history.epoch\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Mean Square Error [$MPG^2$]')\n",
        "plt.plot(hist['epoch'], hist['mse'], label='Train Error')\n",
        "plt.plot(hist['epoch'], hist['val_mse'], label = 'Val Error')\n",
        "plt.legend()\n",
        "\"\"\""
      ],
      "execution_count": 46,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'def build_model():\\n  model = keras.Sequential([\\n    layers.Dense(128, activation=\\'relu\\', input_shape=[len(train_data.keys())]),\\n    keras.layers.Dropout(0.2),\\n    layers.Dense(32, activation=\\'relu\\', kernel_regularizer=keras.regularizers.l2(0.1)),\\n    keras.layers.Dropout(0.2),    \\n    layers.Dense(1)\\n  ])\\n  model.compile(loss=\\'mse\\', optimizer=\\'adam\\', metrics=\\'mse\\')\\n  return model\\n\\nEPOCHS = 100\\nmodel_Confirmed[\"keras\"] = build_model()\\nhistory = model_Confirmed[\"keras\"].fit(train_data, train_labels, epochs=EPOCHS, validation_split = 0.1, verbose=0)\\nmodel_Death[\"keras\"] = build_model()\\nhistory = model_Death[\"keras\"].fit(train_data, train_death, epochs=EPOCHS, validation_split = 0.1, verbose=0)\\nplt.rcParams[\\'figure.figsize\\'] = [20, 10]\\nhist = pd.DataFrame(history.history)\\nhist[\\'epoch\\'] = history.epoch\\nplt.xlabel(\\'Epoch\\')\\nplt.ylabel(\\'Mean Square Error [$MPG^2$]\\')\\nplt.plot(hist[\\'epoch\\'], hist[\\'mse\\'], label=\\'Train Error\\')\\nplt.plot(hist[\\'epoch\\'], hist[\\'val_mse\\'], label = \\'Val Error\\')\\nplt.legend()\\n'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 46
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2yO7CvRAYQIp",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем Pytorch"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NOUCE0wQducc",
        "colab_type": "code",
        "outputId": "5efa98b3-5b93-44b5-bc43-72d6c75f5dd2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"\n",
        "class Net(nn.Module):\n",
        "  def __init__(self):\n",
        "    super(Net, self).__init__()\n",
        "    self.fc1 = nn.Linear(len(train_data.keys()), 100)\n",
        "    self.fc2 = nn.Linear(100, 100)\n",
        "    self.fc3 = nn.Linear(100, 1)\n",
        "  def forward(self, x):\n",
        "    x = F.relu(self.fc1(x))\n",
        "    x = F.dropout(x, p=0.2)\n",
        "    x = F.relu(self.fc2(x))\n",
        "    x = F.dropout(x, p=0.2)\n",
        "    x = self.fc3(x)\n",
        "    return x\n",
        "  def predict(self, data):\n",
        "    pred_y = self(data)\n",
        "    return pred_y.detach().numpy()[:,0]\n",
        "\n",
        "model_Confirmed[\"pytorch\"] = Net()\n",
        "\n",
        "X = Variable(torch.tensor(train_data.values)).float()\n",
        "target = Variable(torch.tensor(train_labels.to_numpy().reshape(-1,1))).float()\n",
        "\n",
        "criterion = torch.nn.MSELoss()\n",
        "optimizer = torch.optim.Adam(model_Confirmed[\"pytorch\"].parameters(), lr = 0.005)\n",
        "epochs = 200\n",
        "for epoch in range(epochs):\n",
        "  net_out = model_Confirmed[\"pytorch\"](X)\n",
        "  loss = criterion(net_out, target)\n",
        "  optimizer.zero_grad()\n",
        "  loss.backward()\n",
        "  optimizer.step()\n",
        "  if (epoch + 1) % (epochs / 10) == 0:\n",
        "      print('epoch {}, loss {}'.format(epoch, loss.item()))\n",
        "model_Confirmed[\"pytorch\"].eval()      \n",
        "\n",
        "model_Death[\"pytorch\"] =Net()\n",
        "target = Variable(torch.tensor(train_death.to_numpy().reshape(-1,1))).float()\n",
        "for epoch in range(epochs):\n",
        "  net_out = model_Death[\"pytorch\"](X)\n",
        "  loss = criterion(net_out, target)\n",
        "  optimizer.zero_grad()\n",
        "  loss.backward()\n",
        "  optimizer.step()\n",
        "model_Death[\"pytorch\"].eval()\n",
        "\"\"\""
      ],
      "execution_count": 47,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\nclass Net(nn.Module):\\n  def __init__(self):\\n    super(Net, self).__init__()\\n    self.fc1 = nn.Linear(len(train_data.keys()), 100)\\n    self.fc2 = nn.Linear(100, 100)\\n    self.fc3 = nn.Linear(100, 1)\\n  def forward(self, x):\\n    x = F.relu(self.fc1(x))\\n    x = F.dropout(x, p=0.2)\\n    x = F.relu(self.fc2(x))\\n    x = F.dropout(x, p=0.2)\\n    x = self.fc3(x)\\n    return x\\n  def predict(self, data):\\n    pred_y = self(data)\\n    return pred_y.detach().numpy()[:,0]\\n\\nmodel_Confirmed[\"pytorch\"] = Net()\\n\\nX = Variable(torch.tensor(train_data.values)).float()\\ntarget = Variable(torch.tensor(train_labels.to_numpy().reshape(-1,1))).float()\\n\\ncriterion = torch.nn.MSELoss()\\noptimizer = torch.optim.Adam(model_Confirmed[\"pytorch\"].parameters(), lr = 0.005)\\nepochs = 200\\nfor epoch in range(epochs):\\n  net_out = model_Confirmed[\"pytorch\"](X)\\n  loss = criterion(net_out, target)\\n  optimizer.zero_grad()\\n  loss.backward()\\n  optimizer.step()\\n  if (epoch + 1) % (epochs / 10) == 0:\\n      print(\\'epoch {}, loss {}\\'.format(epoch, loss.item()))\\nmodel_Confirmed[\"pytorch\"].eval()      \\n\\nmodel_Death[\"pytorch\"] =Net()\\ntarget = Variable(torch.tensor(train_death.to_numpy().reshape(-1,1))).float()\\nfor epoch in range(epochs):\\n  net_out = model_Death[\"pytorch\"](X)\\n  loss = criterion(net_out, target)\\n  optimizer.zero_grad()\\n  loss.backward()\\n  optimizer.step()\\nmodel_Death[\"pytorch\"].eval()\\n'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 47
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FcNKWsjH5D7E",
        "colab_type": "text"
      },
      "source": [
        "## Бустинг и Лес"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3FKBe0XbYmwz",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем случайный лес"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WzzmuIGvYk-f",
        "colab_type": "code",
        "outputId": "e20f0f83-717a-4b3d-dad1-75ba9e3d104f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"\n",
        "param_grid = {\n",
        "    'n_estimators': [200], #Больше - лучше\n",
        "    'max_depth':[5, 10, 15]\n",
        "                }\n",
        "RandomForest1 = ensemble.RandomForestRegressor(random_state=42)\n",
        "model_Confirmed[\"RandomForest\"] = model_selection.GridSearchCV(estimator = RandomForest1, param_grid = param_grid, \n",
        "                                         cv = 3, n_jobs = -1).fit(train_data, train_labels)\n",
        "model_Death[\"RandomForest\"] = model_selection.GridSearchCV(estimator = RandomForest1, param_grid = param_grid, \n",
        "                                         cv = 3, n_jobs = -1).fit(train_data, train_labels)\n",
        "model_Confirmed[\"RandomForest\"].best_params_\n",
        "\"\"\""
      ],
      "execution_count": 48,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\nparam_grid = {\\n    \\'n_estimators\\': [200], #Больше - лучше\\n    \\'max_depth\\':[5, 10, 15]\\n                }\\nRandomForest1 = ensemble.RandomForestRegressor(random_state=42)\\nmodel_Confirmed[\"RandomForest\"] = model_selection.GridSearchCV(estimator = RandomForest1, param_grid = param_grid, \\n                                         cv = 3, n_jobs = -1).fit(train_data, train_labels)\\nmodel_Death[\"RandomForest\"] = model_selection.GridSearchCV(estimator = RandomForest1, param_grid = param_grid, \\n                                         cv = 3, n_jobs = -1).fit(train_data, train_labels)\\nmodel_Confirmed[\"RandomForest\"].best_params_\\n'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 48
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xJT8FNQJYo-j",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем xgboost"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WgrVV-nvAKza",
        "colab_type": "code",
        "outputId": "405b9ffd-e191-4996-a8c7-d83557ae7803",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"model_Confirmed[\"Xgboost\"] = xgb.XGBRegressor(objective ='reg:squarederror',n_estimators=iterat, max_depth=13, \n",
        "                                              n_jobs = -1).fit(train_data, train_labels)\n",
        "model_Death[\"Xgboost\"] = xgb.XGBRegressor(objective ='reg:squarederror',n_estimators=iterat//10, max_depth=13, \n",
        "                                          n_jobs = -1).fit(train_data, train_death)\"\"\""
      ],
      "execution_count": 49,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'model_Confirmed[\"Xgboost\"] = xgb.XGBRegressor(objective =\\'reg:squarederror\\',n_estimators=iterat, max_depth=13, \\n                                              n_jobs = -1).fit(train_data, train_labels)\\nmodel_Death[\"Xgboost\"] = xgb.XGBRegressor(objective =\\'reg:squarederror\\',n_estimators=iterat//10, max_depth=13, \\n                                          n_jobs = -1).fit(train_data, train_death)'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 49
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jwah1COMYgad",
        "colab_type": "text"
      },
      "source": [
        "### Обучаем Adaboost Regressor"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mYyabu1A7JEd",
        "colab_type": "code",
        "outputId": "421d48d7-81a5-4525-c7a2-9d009bac0cf6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 54
        }
      },
      "source": [
        "\"\"\"treeDepth = 30\n",
        "mdl = tree.DecisionTreeRegressor(max_depth=treeDepth)\n",
        "param_grid = {\n",
        "    'n_estimators': [iterat//10],\n",
        "    'learning_rate': [0.0001],\n",
        "    'loss' : [\"exponential\"]\n",
        "                }\n",
        "regrMdl = ensemble.AdaBoostRegressor(base_estimator=mdl)\n",
        "model_Confirmed[\"Adaboost\"] = model_selection.GridSearchCV(estimator = regrMdl, param_grid = param_grid, \n",
        "                                         cv = 2, n_jobs = -1).fit(train_data, train_labels)\n",
        "model_Death[\"Adaboost\"] = model_selection.GridSearchCV(estimator = regrMdl, param_grid = param_grid, \n",
        "                                         cv = 2, n_jobs = -1).fit(train_data, train_death)\n",
        "model_Confirmed[\"Adaboost\"].best_params_\"\"\""
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'treeDepth = 30\\nmdl = tree.DecisionTreeRegressor(max_depth=treeDepth)\\nparam_grid = {\\n    \\'n_estimators\\': [iterat//10],\\n    \\'learning_rate\\': [0.0001],\\n    \\'loss\\' : [\"exponential\"]\\n                }\\nregrMdl = ensemble.AdaBoostRegressor(base_estimator=mdl)\\nmodel_Confirmed[\"Adaboost\"] = model_selection.GridSearchCV(estimator = regrMdl, param_grid = param_grid, \\n                                         cv = 2, n_jobs = -1).fit(train_data, train_labels)\\nmodel_Death[\"Adaboost\"] = model_selection.GridSearchCV(estimator = regrMdl, param_grid = param_grid, \\n                                         cv = 2, n_jobs = -1).fit(train_data, train_death)\\nmodel_Confirmed[\"Adaboost\"].best_params_'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ThUgPuBoyOfo",
        "colab_type": "text"
      },
      "source": [
        "### Catboost"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pPh3Q-dKpdBE",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model_Confirmed[\"catboost\"] = CatBoostRegressor(iterations=iterat*5, depth=10,  l2_leaf_reg = 1, verbose=False).fit(train_data, train_labels) \n",
        "model_Death[\"catboost\"] = CatBoostRegressor(iterations=iterat, depth=10, verbose=False).fit(train_data, train_death)\n",
        "#print(model_Confirmed[\"catboost\"].get_all_params())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nobqZjAPONnR",
        "colab_type": "text"
      },
      "source": [
        "### Lightgbm"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lWFCYX9qAco6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "params = {'min_data': 10}\n",
        "d_train = lgb.Dataset(train_data, label=train_labels)\n",
        "model_Confirmed[\"lightgbm\"]  = lgb.train(params, d_train, iterat)\n",
        "d_train = lgb.Dataset(train_data, label=train_death)\n",
        "model_Death[\"lightgbm\"]  = lgb.train(params, d_train, iterat)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FHt8X6mKdutc",
        "colab_type": "text"
      },
      "source": [
        "## Обучаем свою модель"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4wNgf_jHuLMJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#model_Confirmed[\"new\"] = \n",
        "#model_Death[\"new\"] = "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7mK52KxpHbSN",
        "colab_type": "text"
      },
      "source": [
        "#Предсказываем новые данные"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZBOvOvtke27Q",
        "colab_type": "text"
      },
      "source": [
        "## Функция отображения данных"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fz2ffe27OeEh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_new(list_countries, Matrix, Death = False,subplot_x = 2, days_before = 10, point = True):\n",
        "  dat2 = data2.copy()\n",
        "  plt.rcParams['figure.figsize'] = [24, len(list_countries)*20/subplot_x**2]\n",
        "  if log_labels:\n",
        "    dat2.deaths = 10**dat2.deaths - 1\n",
        "    dat2.confirmed = 10**dat2.confirmed - 1\n",
        "  else:\n",
        "    dat2.deaths = 10**3*dat2.deaths\n",
        "    dat2.confirmed = 10**4*dat2.confirmed  \n",
        "  for i, Country in enumerate(list_countries):\n",
        "    plt.subplot(len(list_countries)// subplot_x + 1, subplot_x, i+1)\n",
        "    yyy = dat2.loc[(dat2.iso_alpha3 == Country)&(dat2.confirmed>0)&(dat2.days_mart>=(days_x - days_before)/sc_days)]\n",
        "    yyy = yyy.loc[(yyy.days_mart<=(days_x + 7)/sc_days)]\n",
        "    if Death:\n",
        "      if point:\n",
        "        plt.plot(yyy.date, yyy.deaths, label = \"Истинное значение\", marker='o')\n",
        "        ax = plt.gca()\n",
        "        for x,y,i in zip(yyy.date,yyy.deaths,range(0,len(yyy))):\n",
        "          if not i % ((subplot_x * (days_before+days_prog))//20):\n",
        "            plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(-5,10), ha=\"right\", va = \"bottom\")\n",
        "      plt.plot(yyy.date, yyy.deaths, label = \"Истинное значение\")\n",
        "      plt.gca().set(xlabel='Дата', ylabel='Смерти')\n",
        "    else:\n",
        "      if point:\n",
        "        plt.plot(yyy.date, yyy.confirmed, label = \"Истинное значение\", marker='o')\n",
        "        ax = plt.gca()\n",
        "        for x,y,i in zip(yyy.date,yyy.confirmed,range(0,len(yyy))):\n",
        "          if not i % ((subplot_x * (days_before+days_prog))//20):\n",
        "            plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(-5,10), ha=\"right\", va = \"bottom\")\n",
        "      else:\n",
        "        plt.plot(yyy.date, yyy.confirmed, label = \"Истинное значение\")\n",
        "      plt.gca().set(xlabel='Дата', ylabel='Заражения')\n",
        "    for name in Matrix:\n",
        "      yyy2 = Matrix[name].loc[:,[Country]]\n",
        "      if point:\n",
        "        plt.plot(yyy2, label = name, marker='o')\n",
        "        ax = plt.gca()\n",
        "        for x,y,i in zip(yyy2.index,yyy2.values.T[0],range(0,len(yyy2))):\n",
        "          if not i % ((subplot_x * (days_before+days_prog))//20):\n",
        "            plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(5,-10), ha=\"left\", va = \"top\")\n",
        "      else:\n",
        "        plt.plot(yyy2, label = name)\n",
        "    plt.title(dic_countries[Country])\n",
        "    plt.xticks(np.arange(0, (days_before+days_prog), ((days_before+days_prog)//10)*subplot_x))\n",
        "    #plt.yticks(np.linspace(0,10+days_prog,30//subplot_x))\n",
        "    plt.grid(True)\n",
        "    plt.legend()\n",
        "\n",
        "#Считаем скор\n",
        "def pred_score(predictions, day):\n",
        "  yyy = new_data[(new_data.days_mart==(days_x + day - 1)/sc_days) & (new_data.confirmed > 0)]\n",
        "  region_list_x = new_data[(new_data.days_mart==(days_x)/sc_days) & (new_data.confirmed > 0)].countries.to_list()\n",
        "  yyy = yyy[yyy.countries.isin(region_list_x)]\n",
        "  #print(len(yyy),len(predictions))\n",
        "  \n",
        "  if len(yyy)==len(predictions):\n",
        "\n",
        "    labels = yyy.confirmed.values  \n",
        "    if log_labels:\n",
        "      labels = 10**labels - 1\n",
        "    else:\n",
        "      labels = 10**4*labels\n",
        "    return np.mean(np.abs(np.log10((predictions+1)/(labels+1))))\n",
        "  else:\n",
        "    return 0      "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PvblEMdS5rT-",
        "colab_type": "text"
      },
      "source": [
        "## Расчёт значений"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ShbS_nOqC0ju",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#del model_Confirmed[\"Adaboost\"]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3bQVsTWd3MWV",
        "colab_type": "code",
        "outputId": "12b245e2-d35b-44a6-ba07-1bf9509ff03f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        }
      },
      "source": [
        "# Сколько дней прогнозируем?\n",
        "days_prog = 8   #256\n",
        "podgon = True\n",
        "\n",
        "\n",
        "Matrix_confirmed,Matrix_Death ={},{}\n",
        "data_list = pd.date_range('2020-04-'+str(days_x-31), periods = days_prog+1, freq ='d')\n",
        "data_list = data_list.strftime('%#m-%#d') \n",
        "new_data_list = data_x.copy()\n",
        "labels = new_data_list.iso_alpha3\n",
        "Matrix_confirmed[\"Среднее\"] = pd.DataFrame(np.zeros((days_prog+1, len(labels))),columns = labels,index = data_list)\n",
        "Matrix_Death[\"Среднее\"] = pd.DataFrame(np.zeros((days_prog+1, len(labels))),columns = labels,index = data_list)\n",
        "\n",
        "for model_name in model_Confirmed:\n",
        "  scores = []\n",
        "  new_data_list = data_x.copy()\n",
        "  labels = new_data_list.iso_alpha3\n",
        "  if log_labels:\n",
        "    predictions_confirmed = [(10**new_data_list.pred_conf - 1).tolist()]\n",
        "    predictions_Death = [(10**new_data_list.pred_deaths - 1).tolist()]\n",
        "  else:\n",
        "    predictions_confirmed = [(new_data_list.pred_conf*10**4).tolist()]\n",
        "    predictions_Death = [(new_data_list.pred_deaths*10**3).tolist()]        \n",
        "  new_data_list = new_data_list.drop(['iso_alpha3','region','confirmed','countries','deaths',\"delta_conf\",\"delta_deaths\"], axis='columns')  \n",
        "  for i in range(days_prog):\n",
        "    #print(data.iloc[-10:-7,-5:])\n",
        "    if model_name == \"pytorch\":\n",
        "      prediction_confirmed = model_Confirmed[model_name].predict(torch.tensor(new_data_list.values).float())\n",
        "      prediction_Death = model_Death[model_name].predict(torch.tensor(new_data_list.values).float())\n",
        "    else:\n",
        "      prediction_confirmed = model_Confirmed[model_name].predict(new_data_list).flatten()\n",
        "      prediction_Death = model_Death[model_name].predict(new_data_list).flatten()\n",
        "    prediction_confirmed[prediction_confirmed<0] = 0\n",
        "    prediction_Death[prediction_Death<0] = 0\n",
        "    #prediction_confirmed = prediction_confirmed + new_data_list.delta_conf_sum\n",
        "    #prediction_confirmed = prediction_confirmed + new_data_list.delta_conf1   \n",
        "    if podgon:\n",
        "      prediction_confirmed = prediction_confirmed*(0.95**(new_data_list.days_10.values*5))\n",
        "      prediction_Death = prediction_Death*(0.95**(new_data_list.days_10.values*5))\n",
        "    \n",
        "    if type_ans_delta:\n",
        "      if log_labels:\n",
        "        prediction_confirmed = prediction_confirmed + new_data_list.pred_conf.values\n",
        "        prediction_Death = prediction_Death + new_data_list.pred_deaths.values\n",
        "      else:\n",
        "        #prediction_confirmed = prediction_confirmed + new_data_list.pred_conf.values\n",
        "        #prediction_Death = prediction_Death + new_data_list.pred_deaths.values        \n",
        "        prediction_Death = (1+prediction_Death)*(new_data_list.pred_deaths.values  + 0.001) \n",
        "        prediction_confirmed = (1+prediction_confirmed)*(new_data_list.pred_conf.values  + 0.0001)\n",
        "\n",
        "\n",
        "    if ('delta_conf_sum' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf_sum\"] = 0\n",
        "    if ('delta_conf6' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf6\"] = new_data_list[\"delta_conf5\"]\n",
        "      if ('delta_conf_sum' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf6\"]\n",
        "    if ('delta_conf5' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf5\"] = new_data_list[\"delta_conf4\"]\n",
        "      if ('delta_conf_sum' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf5\"]\n",
        "    if ('delta_conf4' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf4\"] = new_data_list[\"delta_conf3\"]    \n",
        "      if ('delta_conf_sum' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf4\"]\n",
        "    if ('delta_conf3' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf3\"] = new_data_list[\"delta_conf2\"]\n",
        "      if ('delta_conf_sum' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf3\"]\n",
        "      if ('delta_conf_sum3' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum3\"] = new_data_list[\"delta_conf_sum3\"] + new_data_list[\"delta_conf3\"]        \n",
        "    if ('delta_conf2' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf2\"] = new_data_list[\"delta_conf1\"]\n",
        "      if ('delta_conf_sum' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf2\"]\n",
        "      if ('delta_conf_sum3' in new_data_list.columns):\n",
        "        new_data_list[\"delta_conf_sum3\"] = new_data_list[\"delta_conf_sum3\"] + new_data_list[\"delta_conf2\"]        \n",
        "    new_data_list[\"delta_conf1\"] = (prediction_confirmed - new_data_list[\"pred_conf\"])\n",
        "        \n",
        "    if ('delta_conf_sum' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"] + new_data_list[\"delta_conf1\"]\n",
        "      new_data_list[\"delta_conf_sum\"] = new_data_list[\"delta_conf_sum\"]/6\n",
        "    if ('delta_conf_sum3' in new_data_list.columns):\n",
        "      new_data_list[\"delta_conf_sum3\"] = new_data_list[\"delta_conf_sum3\"] + new_data_list[\"delta_conf1\"]\n",
        "      new_data_list[\"delta_conf_sum3\"] = new_data_list[\"delta_conf_sum3\"]/3\n",
        "\n",
        "\n",
        "    if ('delta_conf_count' in new_data_list.columns) and ('delta_conf1' in new_data_list.columns) and ('delta_conf2' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list[\"delta_conf1\"] < new_data_list[\"delta_conf2\"]) | \n",
        "                        ((new_data_list[\"delta_conf1\"] <0.02) & (new_data_list[\"delta_conf1\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "    if ('delta_conf_count' in new_data_list.columns) and ('delta_conf2' in new_data_list.columns) and ('delta_conf3' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list[\"delta_conf2\"] < new_data_list[\"delta_conf3\"]) | \n",
        "                      ((new_data_list[\"delta_conf2\"] <0.02) & (new_data_list[\"delta_conf2\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "    if ('delta_conf_count' in new_data_list.columns) and ('delta_conf3' in new_data_list.columns) and ('delta_conf4' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list[\"delta_conf3\"] < new_data_list[\"delta_conf4\"]) | \n",
        "                      ((new_data_list[\"delta_conf3\"] <0.02) & (new_data_list[\"delta_conf3\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "    if ('delta_conf_count' in new_data_list.columns) and ('delta_conf4' in new_data_list.columns) and ('delta_conf5' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list[\"delta_conf4\"] < new_data_list[\"delta_conf5\"]) | \n",
        "                      ((new_data_list[\"delta_conf4\"] <0.02) & (new_data_list[\"delta_conf4\"]>0)),\"delta_conf_count\"] +=0.2\n",
        "    if ('delta_conf_count' in new_data_list.columns) and ('delta_conf5' in new_data_list.columns) and ('delta_conf6' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list[\"delta_conf5\"] < new_data_list[\"delta_conf6\"]) | \n",
        "                      ((new_data_list[\"delta_conf5\"] <0.02) & (new_data_list[\"delta_conf5\"]>0)),\"delta_conf_count\"] +=0.2    \n",
        "\n",
        "    if ('delta_deaths6' in new_data_list.columns):\n",
        "      new_data_list[\"delta_deaths6\"] = new_data_list[\"delta_deaths5\"]\n",
        "    if ('delta_deaths5' in new_data_list.columns):\n",
        "      new_data_list[\"delta_deaths5\"] = new_data_list[\"delta_deaths4\"]\n",
        "    if ('delta_deaths4' in new_data_list.columns):\n",
        "      new_data_list[\"delta_deaths4\"] = new_data_list[\"delta_deaths3\"]\n",
        "    if ('delta_deaths3' in new_data_list.columns):\n",
        "      new_data_list[\"delta_deaths3\"] = new_data_list[\"delta_deaths2\"]\n",
        "    if ('delta_deaths2' in new_data_list.columns):\n",
        "      new_data_list[\"delta_deaths2\"] = new_data_list[\"delta_deaths1\"]\n",
        "    new_data_list[\"delta_deaths1\"] = prediction_Death - new_data_list[\"pred_deaths\"] \n",
        "    \n",
        "    if ('days_mart' in new_data_list.columns):  \n",
        "      new_data_list[\"days_mart\"] += 1/sc_days\n",
        "    if ('days' in new_data_list.columns):\n",
        "      new_data_list[\"days\"] += 1/sc_days\n",
        "    if ('days_after_Quarantine' in new_data_list.columns):\n",
        "      new_data_list['days_after_Quarantine'] += 1/sc_days\n",
        "\n",
        "\n",
        "    if ('days_10' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_10']>-100,'days_10'] += 1/sc_days      \n",
        "    if ('days_100' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_100']>-100,'days_100'] += 1/sc_days\n",
        "    if ('days_1000' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_1000']>-100,'days_1000'] += 1/sc_days\n",
        "    if ('days_5' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_5']>-100,'days_5'] += 1/sc_days      \n",
        "    if ('days_50' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_50']>-100,'days_50'] += 1/sc_days\n",
        "    if ('days_500' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_500']>-100,'days_500'] += 1/sc_days\n",
        "    if ('days_1_100' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_1_100']>-100,'days_1_100'] += 1/sc_days\n",
        "    if ('days_1_1000' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_1_1000']>-100,'days_1_1000'] += 1/sc_days\n",
        "    if ('days_1_10000' in new_data_list.columns):\n",
        "      new_data_list.loc[new_data_list['days_1_10000']>-100,'days_1_10000'] += 1/sc_days\n",
        "    if ('recovered' in new_data_list.columns) and ('pred_recovered' in new_data_list.columns):\n",
        "      new_data_list[\"recovered\"] = new_data_list[\"recovered\"] - new_data_list[\"pred_recovered\"] + new_data_list[\"recovered\"]*0.95\n",
        "    new_data_list[\"pred_deaths\"] = prediction_Death\n",
        "    new_data_list[\"pred_conf\"] = prediction_confirmed    \n",
        "\n",
        "    if log_labels:\n",
        "      prediction_Death = 10**prediction_Death - 1\n",
        "      prediction_confirmed = 10**prediction_confirmed - 1    \n",
        "    else:\n",
        "      prediction_Death = 10**3*prediction_Death \n",
        "      prediction_confirmed = 10**4*prediction_confirmed\n",
        "    \n",
        "    if ('pred_conf-population_log' in new_data_list.columns and 'population_log' in new_data_list.columns):\n",
        "      new_data_list['pred_conf-population_log'] = new_data_list['pred_conf'] - new_data_list['population_log'] + 3      \n",
        "      if ('days_1_100' in new_data_list.columns):\n",
        "        new_data_list.loc[(new_data_list['pred_conf-population_log'] > -1)&(new_data_list['days_1_100']<1), 'days_1_100'] = 1/sc_days\n",
        "      if ('days_1_1000' in new_data_list.columns):\n",
        "        new_data_list.loc[(new_data_list['pred_conf-population_log'] > 0)&(new_data_list['days_1_1000']<1), 'days_1_1000'] = 1/sc_days\n",
        "      if ('days_1_10000' in new_data_list.columns):\n",
        "        new_data_list.loc[(new_data_list['pred_conf-population_log'] > -2)&(new_data_list['days_1_10000']<1), 'days_1_10000'] = 1/sc_days\n",
        "\n",
        "    prediction_confirmed[prediction_confirmed>10**7] = 10**7\n",
        "    prediction_Death[prediction_Death>10**6] = 10**6      \n",
        "    predictions_Death.append(np.round(prediction_Death))\n",
        "    predictions_confirmed.append(np.round(prediction_confirmed))\n",
        "\n",
        "    score = pred_score(prediction_confirmed,i+1)\n",
        "    if score:\n",
        "      scores.append(score)\n",
        "    if ('days_10' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_10<1) & (prediction_confirmed > 10),\"days_10\"] = 1\n",
        "    if ('days_100' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_100<1) & (prediction_confirmed > 100),\"days_100\"] = 1\n",
        "    if ('days_1000' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_1000<1) & (prediction_confirmed > 1000),\"days_1000\"] = 1\n",
        "    if ('days_5' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_5<1) & (prediction_confirmed > 5),\"days_5\"] = 1\n",
        "    if ('days_50' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_50<1) & (prediction_confirmed > 50),\"days_50\"] = 1\n",
        "    if ('days_500' in new_data_list.columns):\n",
        "      new_data_list.loc[(new_data_list.days_500<1) & (prediction_confirmed > 500),\"days_500\"] = 1\n",
        "\n",
        "  Matrix_confirmed[model_name] = pd.DataFrame(predictions_confirmed,columns = labels,index = data_list)\n",
        "  Matrix_Death[model_name] = pd.DataFrame(predictions_Death,columns = labels,index = data_list)\n",
        "  Matrix_confirmed[\"Среднее\"] += Matrix_confirmed[model_name]\n",
        "  Matrix_Death[\"Среднее\"] += Matrix_Death[model_name]\n",
        "  print(model_name, sum(scores), scores)\n",
        "Matrix_confirmed[\"Среднее\"] = Matrix_confirmed[\"Среднее\"]//len(model_Confirmed)\n",
        "Matrix_Death[\"Среднее\"] = Matrix_Death[\"Среднее\"]//len(model_Confirmed)\n",
        "scores = []\n",
        "for i,(index, row) in enumerate(Matrix_confirmed[\"Среднее\"].iterrows()):\n",
        "  score = pred_score(np.abs(row),i)\n",
        "  if score:\n",
        "    scores.append(score)\n",
        "print(\"Среднее\", sum(scores), scores)  "
      ],
      "execution_count": 57,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "catboost 0.09604661941144575 [0.017447546311616286, 0.03192465877101355, 0.046674414328815916]\n",
            "lightgbm 0.10507033241692396 [0.02024958242022532, 0.03490553877324977, 0.04991521122344887]\n",
            "Среднее 0.09855291547838049 [0.01829899027402645, 0.033043295047522986, 0.04721063015683105]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w-Nlr45Vb_LH",
        "colab_type": "text"
      },
      "source": [
        "## Строим график"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "E62VuPfOcDHe",
        "colab_type": "code",
        "outputId": "ebd3a806-447d-4ddd-ac59-53a6a69706f2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 584
        }
      },
      "source": [
        "# Создаём список стран\n",
        "#list_countries = countries.iso_alpha3.to_list()[-100:-90]\n",
        "list_countries = []\n",
        "list_countries.extend([\"RU-MOW\"])\n",
        "#list_countries.extend([\"RUS\",\"FRA\",\"ITA\",\"CHN\",\"KOR\",\"USA\"])\n",
        "# Количество графиков в строке\n",
        "subplot_x = 1\n",
        "# Отмечать точки\n",
        "point = True\n",
        "\n",
        "#За сколько дней до начала прогноза\n",
        "days_before = 12\n",
        "\n",
        "#Строим для заражений, или смертей\n",
        "death_bool = False \n",
        "if death_bool:\n",
        "  Matrix = Matrix_Death\n",
        "else:\n",
        "  Matrix = Matrix_confirmed\n",
        "Matrix = {\"Предсказанное\": Matrix[\"lightgbm\"]}\n",
        "plot_new(list_countries, Matrix, death_bool, subplot_x, days_before, point)"
      ],
      "execution_count": 58,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAABY4AAAI3CAYAAAA88M/eAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nOzdeXhN1/7H8fdOJCQymBJJRAw1JQgxtDXVTGqo3vaig6lKVZWaQlRbQnXArVtTDVVqatH2IhTVVtCaQ35oI6a2iKlBDJFIJPv3x4nTHKI1hEPyeT3PeeTstc5a37134vF8sqxtmKaJiIiIiIiIiIiIiMg1DvYuQEREREREREREREQeLAqORURERERERERERMSGgmMRERERERERERERsaHgWERERERERERERERsKDgWERERERERERERERsKjkVERERERERERETEhoJjEREREREREREREbGh4FhEREREJBuGYXQzDMPMfDW4SZ+Dme1R97k8EREREZF7SsGxiIiIiMjfSwFeuP6gYRiPA49ktouIiIiI5CoKjkVERERE/t63QHvDMJyuO/4CsA84dP9LEhERERG5txQci4iIiIj8vS+AIkDLawcMw3AEOgILr+9sGIarYRjjDMM4YhjGFcMwDhiGEW4YhsN1/ZoahrHBMIxzhmFcNgzjkGEYk6/r42wYxluGYezLHOuUYRjLDMOofDvzGYbxtWEYe64b+4vMbTZezHLskcxjL9355RIRERGR3EDBsYiIiIjI3zsGbMR2u4pmgDfXBceGYRjAUmAwsBYYAOwG3gemZukXBKwEXICRwBvAEqBelj4OQCQwOnOMgcB/AGeg5u3Ml1l/ZcMwimQ51gDIAJ647hjAhr+/JCIiIiKS2+WzdwEiIiIiIg+BhcBHhmEUNE0zCXgR2Gqa5iFLdmvVFmgOjDRNMyLz2FTDMGYDvQzDmGya5t7MPvmBJ03TTMjy+fAsX3cBWgBDTNMcl+X4WOOvSW91vo2AgSUYXmYYRlmgBLCYG4Pj46ZpavsNERERkTxOK45FRERERP7ZEsAJeNowDBfgaWBBNv1aY1nF+/F1x/+TpR3gfOafT1+/hUUW/wbOZTMWpmmatzlfDHCRv0LiJ4BEYApQyTAMr8zjDbCEzCIiIiKSxyk4FhERERH5B6ZpngXWYFlp/BTgCizKpmsp4JRpmonXHY/DEvCWzny/CPgJmAmcNgxjsWEYLxiGkfV/BD4C7DdNM/VvSrul+UzTTAc281dw3AD4GdgCpABPGIZRHCiPtqkQERERERQci4iIiIjcqoVYtoXoC3xvmubpOx3INM1koCHQGPgMqIhlBfOWzBXN98IGIMQwDDcsAfKGzFB6W+b7a/sba8WxiIiIiCg4FhERERG5RcuAK1geYLfwJn3+AIobhuF53fEKWP7t/fu1A6ZpZpimGWWa5hDTNKsBr2F56N0zmV0OARUMw3D+m5pueT4sgbAj8CxQjr9WFm/gr+D4LLD3b+YTERERkTxCwbGIiIiIyC0wTfMy0BuIAP53k24rsPwbu991xwdm/rkSwDCMotl8dmfmn4Uy//wKKAy8cX3HLA/Hu6X5Mm0DUoFhwGUgOvP4RiAYaAX8lGX/ZBERERHJw/L9cxcREREREQEwTXPeP3RZAawFIgzDKIUlDG6CZZXvdNM0r63mfdswjEZYgt3fsQTErwJJmWMAzAM6AWMNw6iFZWVwASzbWyzKbL/V+TBNM8UwjO1YVkz/YJpmWmbTJsDEsgp5+m1eEhERERHJpRQci4iIiIjkENM0TcMw/oVlVfJzQBfgCPAmMDZL12VAANAV8ALOYHl43SjTNP/IHCvdMIw2mZ99AfgXlq0ktpC5Wvg25rtmI5bg2PoAPNM0LxmGsQuohR6MJyIiIiKZDP1PNBERERERERERERHJSnsci4iIiIiIiIiIiIgNBcciIiIiIiIiIiIiYkPBsYiIiIiIiIiIiIjYUHAsIiIiIiIiIiIiIjYUHIuIiIiIiIiIiIiIjXz2LuBhU6xYMbN06dL2LuO+SkpKomDBgvYuQ+xE9z9v0/0XfQ/kbbr/eZvuf96m+5+36f7nbbr/eZvuf94THR2dYJqmV3ZtCo5vU+nSpdmxY4e9y7ivoqKiaNSokb3LEDvR/c/bdP9F3wN5m+5/3qb7n7fp/udtuv95m+5/3qb7n/cYhvHHzdq0VYWIiIiIiIiIiIiI2FBwLCIiIiIiIiIiIiI2FByLiIiIiIiIiIiIiA0FxyIiIiIiIiIiIiJiQ8GxiIiIiIiIiIiIiNjIZ+8CcpOMjAwSEhJITEwkPT3d3uXkGE9PT2JjY+1dhtjJ7dx/R0dHChUqRLFixXBw0O+lREREREREREQeVgqOc9CxY8cwDIPSpUvj5OSEYRj2LilHXLx4EXd3d3uXIXZyq/ffNE3S0tI4deoUx44dIyAg4D5UJyIiIiIiIiIi94KWBOagpKQkSpQogbOzc64JjUVulWEYODs7U6JECZKSkuxdjoiIiIiIiIiI3AUFxzlM/z1f8jr9DIiIiIiIiIiIPPyU8IiIiIiIiIiIiIiIDQXHIiIiIiIiIiIiImJDwbGIiIiIiIiIiIiI2FBwLHTr1o02bdrYHDt9+jQeHh4YhsGZM2fsVJmIiIiIiIiIiIjYQz57FyA3WrornnFr4jiemIxfIRfCWlbk6ZAS97WGUaNGkZaWdl/nFBERERERERERkQeDVhw/YJbuimfYN3uIT0zGBOITkxn2zR6W7oq/bzUcPHiQGTNm8MYbb9zQtm/fPp566ik8PT1xc3OjTp067Nmzh5EjR2IYRrav0qVLAzBy5EiqVKliHSs1NZVy5cphGAYJCQkAzJkzBzc3N5s5ExISMAyDqKgoAKKiomw+c42bmxtz5syxvt+zZw/NmjXDxcWFIkWK0K1bN86fP2/zmdmzZxMUFESBAgWoUKECEyZMICMj46bX5qOPPiIgIID8+fNTvHhxevToweXLl63thmHw1Vdf2XymTZs2dOvWzfp+/vz51K5dG3d3d7y9vWnfvj3x8X/d3+uvE0CVKlUYOXKk9f358+d55ZVX8Pb2xt3dnYYNG7Jjxw5r+51ex86dO99wDsePH+e5556jcOHCFC5cmNatW3PgwIGbXiMREREREREREXn4acXxPRYR+Qu/Hr9wy/13HUkkNd02uExOS2fIV7v5YtuRWxojyM+DEW0r31adWb355ps0aNCA0NBQPvzwQ+vx48ePU79+ferVq8fatWspVKgQ27ZtIz09ncGDB/Pqq68CMH78eDZt2sQ333wDgKOjY7bzTJ48mVOnTt1xnX8nKSmJli1b8uijj7Jt2zbOnj1Lz5496d69O19//TUAM2fO5J133mHSpEnUrFmTvXv30rNnT5ycnHj99dezHffxxx9nyZIl+Pr68scff9ClSxfGjx/PO++8c8u1paamEhERQaVKlUhISGDo0KE8//zzbNiw4ZY+b5omrVu3xtPTkxUrVlCkSBE+//xzmjRpQlxcHL6+vrdcS1bR0dEsX77c5tjly5dp3bo19evXZ/369Tg7OzN+/HiaNWtGbGwsrq6udzSXiIiIiIiIiIg82BQcP2CuD43/6XhO2759O19//TVbt27l0qVLNm1TpkyhYMGCLFmyBGdnZwAqVKhgbb+2wtXNzQ1nZ2d8fHxuOs/Zs2cZM2YMQ4cO5e23387x81i4cCFJSUnMmzcPd3d3AGbMmEHjxo05ePAg5cqVY/To0YwdO5Z///vfAJQpU4bw8HCmTp160+C4bt261q8LFCiAh4cH6enpt1Vb9+7drV+XLVuWTz75hMDAQI4dO4a/vz8uLi4kJyff9PPr1q0jJiaGP//8ExcXFwBGjx5NZGQk8+bNY8iQIbdVzzWDBg0iLCzM5n58+eWXmKbJ7NmzMQwDgOnTp+Pt7c2KFSvo0KHDHc0lIiIiIiIiIiIPNgXH99jtrvyt98GPxCfeGBqWKOTCol51cqqsmxoyZAjt27enVq1a1i0Nrtm1axf169e3hsZ3Y/To0TRq1Ij69evf0JaUlHTDNgvZubYFRtbPXRMbG0twcLA1NAZL6Ovg4MCvv/6Kp6cnR48epVevXvTu3dva5+rVq5im+bfzLliwgFdeeYXLly/zzDPPMHToUJv2zp0722xNkZKSQqdOnazvd+7cSUREBDExMZw9e9Y635EjR/D396dKlSr89ttvbN68mTp1brzn0dHRXL58GS8vL5vjKSkpHDp0yOZ63Mp1BFi2bBmHDh1i1apVNsFxdHQ0f/zxh811BMtK5KxziYiIiIiIiIg81HYvhh9Gwflj4OkPTd+B4Ly9YE7B8QMmrGVFhn2zh+S0v1axujg5Etay4j2f+9tvv+Xnn38mNjb2ns5z6NAhPv30U3bt2sWxY8duaHd1dSUmJsb6/ty5czz66KM39Fu3bh2FCxe2vq9WrdotzW8YhnUf42nTptmsIr4VTz31FI8++iiHDh1i0KBBzJo1i759+1rbx40bR2hoqPX9K6+8Yv362hYazZo1Y968eXh7e5OQkECDBg1ITU0FoFWrVrzwwgvUrVsXV1dXDMOwWYGckZFB8eLF2bhx4w21eXh4WL++1et49epVhg4dypgxY6wrmLPOFRwczOLFi2/4XJEiRf72OomIiIiIiIiIPBR2L4bIfpCWmb+cP2p5D3k6PFZw/IB5OqQEAOPWxHE8MRm/Qi6EtaxoPX6vZGRkEB4ezquvvsojjzySbZ+QkBDmz59PamrqXa06Dg8P5+WXX6ZcuXLZBseGYVCuXDnr++sfgndNmTJlKFasmM3nrgkMDOSzzz7j4sWL1tWymzZtIiMjg8DAQIoXL46fnx+HDh2iS5cut1W/u7s77u7ulC9fnri4OGbMmGETHPv4+NjUn3Uf4H379pGQkMB7771HmTJlAKx7QWc9j/nz5zNp0iTOnDkDWMLka2rUqMGpU6dwcHCgbNmyN63zVq/j9OnTcXV1pXPnzje01ahRgy+++IJixYpRqFChm84lIiIiIiIiIvLQ+mHUX6HxNWnJluMKjuVB8nRIiXseFF9v3bp1ODk5/e1+w6+99hrTpk2jQ4cODB8+nMKFC7N9+3YCAwOpXr36Lc3z22+/cfz4cQ4ePJhTpWfrxRdfZMSIEXTp0oVRo0Zx7tw5evXqxTPPPGMNUyMiIujbty+FChWiVatWpKWlsXPnTuLj4xk2bFi243766afUqlWLIkWKEBcXx9SpU3nsscduua6AgADy58/P5MmT6dOnD7GxsTe95oULF7auqM4a1Ddr1ox69erRrl07xo4dS6VKlTh58iSrV6+mWbNmNGjQ4JbrAcsK6cjISJvg/ZoXX3yRsWPH0q5dO0aNGkVAQABHjx5l2bJlvPrqq5QvX/625hIREREREREReaBcuWRZYZyd8zcueMxLHOxdgDwYUlJSCAsLu2Hf3KxKlCjBhg0bSE1NpXHjxoSEhDBp0iTy5bv13z9cvnyZ4cOH3/NtDlxdXVmzZg0XLlzg0UcfpV27dtSpU4fPPvvM2qdHjx589tlnzJs3j2rVqtGgQQNmzJhhXQmcnc2bN9OqVSvKly9P9+7dadSoEZMmTbrlury8vPj8889ZunQpQUFBRERE8NFHH93WuRmGwbfffkuTJk3o2bMnFStWpEOHDsTFxeHn53dbYwE0btyYxo0bZ9vm6urK6tWrKVu2LO3bt6dSpUp07dqVc+fO2WwTIiIiIiIiIiLy0IldAVNu3NbTytP//tXyADL+6UFgYqtWrVrmjh07sm2LjY0lMDDwPld072Xd7kHynju5/7n1ZyEvioqKolGjRvYuQ+xI3wN5m+5/3qb7n7fp/udtuv95m+5/3pZn7n/iEVg1FOK+Be/KENgGNk203a7CyQXaTsz1W1UYhhFtmmat7Nq0VYWIiIiIiIiIiIjkfulpsOUTiHrf8r75KHj8NXB0gqLlLHsanz9mWWnc9J1cHxr/EwXHIiIiIiIiIiIikrsd3QYrBsCpvVAhFFqNg0IBf7UHd8jzQfH1FByLiIiIiIiIiIhI7pR8Dr6PgOg54OEHHedDpTZgGPau7IGn4FhERERERERERERyF9OEPUtgzZtw+YxlS4rGwyC/nuN1qxQci4iIiIiIiIiISO5x5hCsHAiHo6BETej0NfhWs3dVDx0FxyIiIiIiIiIiIvLwu3oFfpoAGz+CfPmh1Xio1R0cHO1d2UNJwbGIiIiIiIiIiIg83A6vt6wyPnMQqjwLLd8Ddx97V/VQU3AsIiIiIiIiIiIiD6dLf8J3w2H3Iihc2rItRblm9q4qV1BwLCIiIiIiIiIiIg+XjAzY+Tl8PwJSL8MTYdBgEDi52LuyXMPB3gWIiIiIiIiIiIiI3LJTv8BnLWFFfyheFXpvgiZv3dPQuHTp0hiGccOrdevWAEyZMoXg4GA8PDzw8PCgTp06rFy50maMbt263fD5xx9/3KbPlStX6Nu3L8WKFaNgwYI89dRTHDt27IZ65s+fT/Xq1SlQoADFihWjS5cuOX7OCo4l22/arK8//vjD3iWKiIiIiIiIiEhel5oE370N0xrA2UPw9CfQbQV4VbjnU2/fvp0TJ05YXzt37sQwDDp06ACAv78/H374ITt37mTHjh00adKEp59+mt27d9uM06xZM5txvv32W5v2/v378/XXX/PFF1+wceNGLly4QJs2bUhPT7f2mThxImFhYQwePJi9e/eybt062rVrl+PnrOD4QbR7MUyoAiMLWf7cvfieT3n9N+2JEydYvXr1PZ9XRERERERERERyjxMnTtC1a1e8vLwoUKAAQUFBrF+/3tp+6dIl+vbti7+/Py4uLlSsWJEJEybcMM62bdto3rw5bm5uuLu7UzckiIQPa8GmiVD9Bc69+D2d/7MKz0KF8PT0pHPnziQmJlo///vvv2e7QPJO8y4vLy98fHysr2+//RYPDw9rcNyuXTuefPJJypUrR4UKFRgzZgzu7u5s3rzZZpz8+fPbjFOkSBFr2/nz55k1axbjxo2jefPm1KhRg3nz5rF7926+//57ABITExk2bBhz586lU6dOlCtXjqpVq/Lss8/e0Xn9HQXHD5rdiyGyH5w/CpiWPyP73fPw+PpvWh8fH4oWLWptj4qKwjAMVqxYYV0GX7NmTaKjo23G2bRpEw0bNsTV1ZUSJUrQu3dvLly4YNPn2ljXv7I6f/48vXv3xtfXlwIFChAYGMiiRYsAmDNnDm5ubta+8fHxlC1bln79+lmPzZ8/n9q1a+Pu7o63tzft27cnPj7e2r5u3ToCAwNxcXHB09OT0NBQDhw4YG1fvXo1DRo0oHDhwhQpUoSWLVsSGxtrbb/2l8+OHTts6q5SpQojR460vjcMg6+++sqmT5s2bejWrZv1/blz5+jatSuFCxfGxcWFZs2a8csvv9z2dRURERERERERsafExETq1auHaZqsXLmS2NhYJk2ahLe3t7XPwIEDWblyJfPmzSM2Npbhw4cTHh7OvHnzrH22bt1KixYtaNSoEVvWLiN6VCMGV/gdJxdXeGkVtJvMCz36sHPnTlavXs3q1avZuXMnnTt3vqGm1atX2yyUbNKkyV2fp2mazJo1i06dOuHicuP2GOnp6Xz55ZdcunSJunXr2rT99NNPeHt7U6FCBXr27Mnp06etbdHR0aSlpdGiRQvrsZIlSxIYGMimTZsA+O6770hPT+fUqVMEBQVRokQJ/vWvf3H48OG7Pq/r6eF499qqcDi559b7H9sO6Vdsj6Ulw7LXIfrzWxvDpyo8+cGtz3kbBg8ezMcff0yJEiWIiIigTZs2HDp0CFdXV/bs2UOLFi2IiIjg008/5ezZs/Tv35/u3bvfEJ4C/PLLLxQpUoQVK1bQs2dP63HTNGnVqhXnzp1j9uzZVKhQgbi4OFJSUm4YIyEhgebNm9OwYUM+/vhj6/HU1FQiIiKoVKkSCQkJDB06lOeff54NGzYAlh+6KVOmUKZMGS5dusSQIUPo1asXP/74IwBJSUn079+f4OBgkpOTeffdd2nbti2//vorzs7OOXpNu3XrRlxcHMuWLaNw4cIMHz6c0NBQ9u/fj4uLy21fVxERERERERERexg7diy+vr7MnTvXeqxMmTI2fTZt2kTnzp1p3LgxYNk7eNasWWzdutUa/A4YMIA+vXszvEkhWNcFMtKp8Pq7UOd1yOdMbGwsq1ev5qeffqJOnToATJ8+nQYNGhAXF0fFihWt8xUtWhQfH58cPc+1a9fy22+/2eRZAHv27KFOnTqkpKTg5ubG//73P6pWrWptDw0N5ZlnnqFMmTL8/vvvvPXWWzRp0oTo6Gjy58/PyZMncXR0pFixYjbjFi9enJMnTwJw+PBhMjIyePfdd/nvf/9LkSJFGDVqFI0bNyY2NhZXV9ccO08Fxw+a60Pjfzp+n7399tu0bNkSgNmzZ+Pv78/ChQvp0aMH48aNo2PHjgwaNMja/5NPPiEkJITTp09bf7t05YrlXEqUKIGnpyeFChWymeP7779n8+bN/PLLLwQGBgJQtmzZG2o5f/48LVq0oHLlynz66ac2q5a7d+9u/bps2bJ88sknBAYGcuzYMfz9/SlXrhzlypUDICUlBV9fXw4dOmT9zPXL+2fPno2Hhwfbtm2jfv36t3/hbuLAgQMsX76c9evX88QTTwAwb948AgICWLBgwW1dVxERERERERERe1q6dCmhoaF07NiRdevW4efnR48ePejTp481t6lfvz6RkZH06NGDkiVLsmnTJmJiYggLCwPg9OnTbN68mRcCTlG/4x/sP5+PikFVGVmnGk3zWRbzbd68GTc3N5vVvPXq1aNgwYJs2rTJJjh+5plnSElJoXz58gwYMIB///vfd32eM2fOpHbt2lSrVs3meMWKFYmJieH8+fN89dVXdO3alaioKKpUqQLAc889Z+1btWpVatasSalSpVi5ciXPPPPMLc2dkZFBWloaEydOtK5MXrBgAT4+PkRGRtKxY8e7Pr9rFBzfa7e78ndClcxtKq7jWRJeWnnj8fvs2m9xANzc3KhatSq//vorYFlOf/DgQeuWEmBZPQxw6NAha8B55swZHBwcKFiwYLZz7Nq1C19fX2tonJ309HRat27Nrl276NWrF46OjjbtO3fuJCIigpiYGM6ePWut48iRI/j7+1u/DgoKIjk5mUqVKhEZGWn9/KFDh3j77bfZunUrf/75JxkZGWRkZHDkyBGbeZ544gkcHP7a8SU5OfmGWjt37myzNUVKSgqdOnUCIDY2FgcHB5vr6unpeUfXVURERERERETEng4fPszUqVMZMGAA4eHhxMTE0LdvXwBef/11wPJgt169ehEQEEC+fJZoctKkSbRp0wZSznN43hsAjIj8nXHhvQlp3Z0lX31Fy5YtiY6Oplq1apw8eRIvLy+bRYSGYeDt7W1dmevm5sb48eOpV68e+fLlY/ny5XTs2JHPP//cmsvcidOnT7Ns2TKmTJlyQ5uzs7N1oWLNmjXZvn07EyZMYNasWdmO5efnh7+/v3X7VB8fH9LT00lISMDLy8va79SpUzRo0AAAX19fAIKCgqztnp6e+Pn53ZBb3S0Fxw+apu9Y9jROyxJAOrlYjj/gMjIy6NGjBwMGDLihrUSJEtavDx8+TMmSJa1/OdyJlJQUHB0dmTVrFv379+fJJ58kICAAsGwz0bJlS5o1a8a8efPw9vYmISGBBg0akJqaah3Dz8+PmJgYEhISeP/99xkyZIh164c2bdrg7+/P9OnTKVGiBPny5SMoKMjm8wALFy60/tYIoFWrVjfUOm7cOEJDQ63vX3nllVs6x2t/+d3qdRURERERERERsaeMjAxq1arF+++/D0BISAgHDhxgypQp1uB40qRJbNq0ieXLl1OqVCk2bNjA4MGDKW0eJTRpCRmxJwDo1ac/3d/+j2WcGjVYt24d06ZN45NPPrmlWooVK2bzv7dr1apFQkICY8eOvavgeM6cOeTPn5/nn3/+H/tmZGRY/+d9dhISEoiPj7eGwTVr1sTJyYm1a9fywgsvAHDs2DFiY2Otq6vr1asHQFxcnHVx5KVLlzhx4gSlSpW64/PKjoLjB02w5UmM/DAKzh8DT39LaHztuJ1t2bLFum1EUlISe/fupUuXLgDUqFGDX375xfqblZtZv3699bck2QkJCeHEiRPExsbedNVx/vz5iYyMxMPDg1WrVvHyyy/z3XffYRgG+/btIyEhgffee8+6j84333xzwxj58uWzblkxatQoqlevzpkzZwDYt28fU6dOte63s3PnTq5evXrDGNe2vbgmu/2PfXx8bPpk3WsmMDCQjIwMNm/ebN2q4sKFC+zZs4eXXnoJuPXrKiIiIiIiIiJiT76+vjYrYcGSfVx7LlVycjLDhg1jyZIltG3bFoBgfzdivhrP+HEfEjq0Hr49P4LP2hIUHGIzTlBQkHVFrY+PD3/++SemaVoX3pmmyenTp/92P+PHHnuM2bNn3/H5mabJp59+ynPPPYebm5tNW3h4OK1bt6ZkyZJcvHiRhQsXEhUVxcqVlh0ELl26xMiRI3n22Wfx9fXl999/Z9iwYXh7e/Ovf/0LsKwcfvnllxkyZAje3t4ULVqUgQMHEhwcTLNmzQCoUKEC7dq144033mD69OkULlyYESNG4O3tbVm1nYMUHD+Igjs8MEHx9d599128vLzw8/Nj1KhRODs7W38DMnToUB5//HFeffVVevXqhbu7O/v27SMyMpLp06eTmppKZGQkP/74I4sXL7b+14HExEQATp48iY+PD02bNuWxxx7j2WefZcKECVSoUIGDBw+SlJTE008/DVhCXw8PDwCmTp1K5cqVmTZtGr179yYgIID8+fMzefJk+vTpQ2xsLG+//bbNeSxbtoyiRYtSsmRJ/vzzTyIiIggICKBo0aJkZGRQrFgxZs6cScmSJYmPjycsLOyuVkjfTPny5WnXrh29evVixowZFCpUiOHDh+Ph4XHL11VERERERERE5EFQr1494uLibI7t37/fuhI2LS2NtLQ0y5ajV6/AzxNh43gcL18io0g56PkjpQ0H/Pz8sh3n2oPm6tSpw6VLl9i8ebN1Je7mzZtJSjaqKRoAACAASURBVEqy2ff4ejExMdbVvXciKiqKAwcOMH/+/BvaTp48SadOnTh58iSenp4EBwezatUq67PCHB0d2bNnD3PnziUxMRFfX18aN27M4sWLcXd3t47z3//+l3z58tGxY0eSk5Np2rQpc+fOtdmmdd68eQwcOJC2bdtimib169fnhx9+yNEH48F9DI4Nw/gdyG699LemabbO7PMaEAb4Ar8A/U3T3JhljPzAeOB5wAX4AXjNNM1jWfoEAFOAJkAysBAYbJpmapY+DYGPgMrAcWCsaZrTcuxkc7EPPviAQYMGERcXR+XKlVmxYoV1r+Lg4GA2bNjAW2+9RcOGDUlPT6ds2bLW35ps2rTJugF5dhuR+/r6YpomDg4OrFq1irCwMDp16sTFixcpW7YsI0eOzLYmLy8vpk2bRteuXQkNDaVMmTJ8/vnnvPnmm0yZMoXg4GA++ugjm+0ijh07xuDBgzl69Cju7u48/vjjfPvttwA4ODiwaNEi+vXrR5UqVShXrhz/+c9/bnhgXk6ZPXs2/fv356mnniIlJYV69eqxevVqXFxcgH++riIiIiIiIiIiD4IBAwZQt25dxowZQ8eOHdm1axcTJ07kvffeA8DDw4OGDRsSPqgfbmudKMVx1qcFM3fPTsaO7QMOjhhAWFgYI0aMIDg4mJCQEBYvXsyWLVuYPHkyYFnFHBoaal2IB9CrVy/atGljfTDe559/jpOTEyEhITg4OBAZGcmUKVP48MMP7/j8GjdubH3u1PXmzJnzt591cXFhzZo1/zhH/vz5mTRpEpMmTbppH3d3d2bOnMnMmTP/cby7YprmfXkBXoBPllcIkAF0zWzvCKQBPYFAYBJwCQjIMsYnWILe5kANIAqIARwz2x2BPZnHa2T2Ow5MyjJGGSApc/zAzPnSgGdv5Txq1qxp3syvv/5607aH2YULF8x169aZgPnnn3/e8Tjr1q0zGzZseNN2y7ejPGguXLhw25/JrT8LedG6devsXYLYmb4H8jbd/7xN9z9v0/3P23T/8zbd/7wtJ+7/ihUrzODgYDN//vxm+fLlzY8//tjMyMiwNF5KME/M6mJ2q+5k+nnmMwvkdzYrVqxojhs37q8+mT744AOzZMmSpqurq1m7dm1z7dq1Nu1nz541X3zxRdPd3d10d3c3X3zxRfPcuXPW9jlz5piBgYGmq6ur6e7ubtasWdOcN2/eXZ9fbgPsMG+Sg963Fcemaf6Z9b1hGC8DF4DFmYcGAnNM07wWlfc1DCMU6A0MMwzDE3gZeMk0zbWZY3QG/gCaAWuAFlhWEZcyTfNoZp8hwKeGYQw3TfMC8Cpw3DTNvpnzxBqG8RgwGPj6Hpy6ZHJ2dqZIkSI3bS9evPh9rEZERERERERERHJa69atad26te3BjAzYNR/Wvo3PlYvMHjccnggD55tvrTB06FCGDh160/bChQtnu2XENV27dqVr1663Xb/8xcEekxqWXatfBuabpplsGIYzUBP47rqu3wHXNiapCThl7ZMZDsdm6VMHiL0WGmdaA+TP/Py1PtfPswaoZRiG092cl/y9unXrZvuQumuu7XksIiIiIiIiIiK5xOlYmNMKlr8OXoHw6k/QbMTfhsbyYLDXw/GaY9ky4trq4mJYtpk4dV2/U1hWE4Nle4t0ICGbPj5Z+lw/RkLm57L2+T6bMfJl1nHi+mINw3gFeAUsq2KjoqKyPSlPT08uXryYbdvDLD09nZo1a3LhwgWAXHmOcnPp6em3fc9TUlJu+nMiD5dLly7pXuZx+h7I23T/8zbd/7xN9z9v0/3P23T/87acvP8O6Vco9cciSh5dSrqjK4cq9uWkTxP49ZTlJQ88ewXHPYHtpmn+n53mvy2mac4AZgDUqlXLbNSoUbb9YmNjbZ6CmFtcvHgxV56X3Jo7uf8FChQgJCTkHlUk91NUVBQ3+ztP8gZ9D+Rtuv95m+5/3qb7n7fp/udtuv95W47d/wNrYWUYJP4B1V/EofloKhUsSqW7H1nuo/seHBuG4Q20A/pkOXxtVfD1m9wWB67tX3ASy6rkYsCf1/XZmKVPvevGuLaaOes42c1zlRtXM9820zSx7MQhkjeZN3m6qIiIiIiIiIjkchdOwOqh8OsyKFYRuq2E0vXtXZXcIXvscdwNuAJ8ce2AaZqpQDSWLSyyag5syvw6GkjL2scwDH8gMEufzUBg5vGsY1zJ/Py1PtnNs8M0zbQ7OqNMTk5OJCcn380QIg+95ORknJy0XbiIiIiIiIhInpGRDlunw+TasH8NNHnLspexQuOH2n1dcZz5ULwewJemaV66rvkjYJ5hGNuAn4FXAT9gGoBpmucNw5gFjDUM4zRwJvMzu/lrz+LvgF+AuYZhDAKKAuOAmaZpXsjsMw143TCM/wLTsaxQ7gY8f7fn5+3tTXx8PCVKlMDFxUUrjyVPMU2T5ORk4uPjKV78+kX9IiIiIiIiIpIrHd8Fkf3hRAw80hRaj4ciZe1dleSA+71VRSOgPNDp+gbTNBcZhlEUeAvwBfYCrUzT/CNLt/5YtpRYBLgAPwBdTNNMzxwj3TCM1sBULOFzMrAACMsyz2+GYbQCJgC9geNAP9M0v77bk/Pw8ADg+PHjpKXd1eLlB0pKSgoFChSwdxliJ7dz/52cnChevLj1Z0FEREREREREcqmUC7BuDGybAQW94N+fQeVnQAspc437GhybprkOuOl3j2maU7GEvjdrvwL0zXzdrM8RoM0/1LEeqPFP9d4JDw+PXBeaRUVF6UFneZjuv4iIiIiIiIhYmaZlD+PV4XDxJNTuAU3fhgKe9q5Mcth9fzieiIiIiIiIiIiIPITO/Q4rB8PBteATDM8tgBI17V1Vjli6K55xa+I4npiMXyEXwlpW5OmQEvYuy64UHIuIiIiIiIiIiMjNXU2FzZNh/VhwcITQD6B2T3DMHdHi0l3xDPtmD8lp6QDEJyYz7Js9AHk6PM4dd1dERERERERERERy3h+bYcUA+DMWAttC6IfgmbvC1HFr4qyh8TXJaemMWxOn4FhERERERERERETE6vJZWPsO7JoHngHw/CKoGGrvqu6J44nJt3U8r1BwLCIiIiIiIiIikpftXgw/jKLh+WOwyx/KNYPY5ZByHuq9AQ2HgnNBe1eZ40zT5H+74m/a7lfI5T5W8+BRcCwiIiIiIiIiIpJX7V4Mkf0gLRkD4PxRiJ4NRR6BrpFQvLK9K7wnzial8uY3e1j9y0nKFHPleGIKV65mWNtdnBwJa1nRjhXan4JjERERERERERGRvOqHUZCWzZYM6VdybWj8Q+wphn69hwvJaYQ/WYmeDcoS+X/HGbcmjuOJyfgVciGsZcU8vb8xKDgWERERERERERHJmzLSLSuMs3P+5ls4PKwuXbnKuyt+5cvtR6nk4868lx8l0NcDgKdDSuT5oPh6Co5FRERERERERETymj82werwm7d7+t+/Wu6Dbb+dZdCSGOLPJdO70SP0b1ae/Pkc7V3WA03BsYiIiIiIiIiISF6ReBTWvgO/fAMeJaB2D4hZYLtdhZMLNH3HfjXmoCtX0/nou/3M2HiYkoVdWdyrDrVKF7F3WQ8FBcciIiIiIiIiIiK5XWoS/PRf2DQRMKDRMKjbD5xdoeRj8MMozPPHMDz9LaFxcAd7V3zXfj1+gQGLYog7dZEXHgtgeKtACuZXHHqrdKVERERERERERERyK9OEPUtg7Qi4eByq/BuaR9huRRHcAYI7sD4qikaNGtmt1JySnmEyfcMhJqzdTyFXZ2Z3q03jSt72Luuho+BYREREREREREQkN4qPhlXhcGwb+FaH9rMh4HF7V3VP/Z6QxKAl/0f0H+doXdWXd5+uQuGCzvYu66Gk4FhERERERERERCQ3uXACfhgF/7cQ3IpDuylQ7QVwcLB3ZfeMaZos3HaEMStjyedg8PFz1Xmqmh+GYdi7tIeWgmMREREREREREZHcIC0FNk+GjR9BRhrUHwANBkF+d3tXdk+dvpDCkK93ExX3J/XLFWNc+2B8PV3sXdZDT8GxiIiIiIiIiIjIw8w0IXY5fPcWJB6BSm2gxWgoUtbeld1zK3Yf562le0lJS2dUu8p0eqwUDg5aZZwTFByLiIiIiIiIiIg8rE7ugdXD4PeN4F0ZuiyHsg3tXdU9d/5yGu8s38uymONUK1mIjzpU4xEvN3uXlasoOBYREREREREREXnYJCXAj+/Czs+hQCFo/R+o0Q0cc3/ct/HAn4Qt2U3CpSsMbF6B1xo9Qj7H3Lt/s73oioqIiIiIiIiIiDwsrqbCpskwsQbsmgeP9oJ+O6F2j3saGp84cYKuXbvi5eVFgQIFCAoKYv369db2S5cu0bdvX/z9/XFxcaFixYpMmDDBZowZM2bQuHFjChUqhGEY/P777zfMU7p0aQzDsHmFh4cDcDn1Ku8s28u/357BvhlvcGJiR957sQHD3xzG1atX79m551W5/1cQIiIiIiIiIiIiucH+72DNMDhzEMo1g5bvg1eFez5tYmIi9erVo379+qxcuRIvLy8OHz6Mt7e3tc/AgQP5/vvvmTdvHmXKlGHDhg307NmTYsWK0blzZwAuX75MixYtaNeuHQMGDLjpfO+88w69e/e2vndzc2PXkXMMXPx/xP26hzNfRzB06FBe6taV+Ph4Xn31VdLT0xk/fvy9uwh5kIJjERERERERERGRB9mfcbDmTTj4PRQtDy8sgQot7tv0Y8eOxdfXl7lz51qPlSlTxqbPpk2b6Ny5M40bNwYsK4dnzZrF1q1brcFx//79AdixY8ffzufu7o6Pjw8AqVczmPTjAaasO4ivpwvN8h9iT5XKjHl3NADlypVj7NixdOjQgREjRuDu7p4zJy3aqkJEREREREREROSBlHwOVoXD1DpwdDu0fA96b7qvoTHA0qVLeeyxx+jYsSPe3t5Ur16dyZMnY5qmtU/9+vWJjIzk6NGjgCVIjomJITQ09LbnGz9+PEWLFiWwSjBVn+7FxLWx/CvEn1X9G+Dl6kiBAgVs+ru4uJCSkkJ0dPTdnajYUHAsIiIiIiIiIiLyIEm/CttmWvYx3jYdanSx7GNcpw/kc77v5Rw+fJipU6dStmxZ1qxZwxtvvEF4eDhTpkyx9pk4cSLVqlUjICAAJycnGjZsyIcffkibNm1ua65+/fqxYMFC+v5nHmdLNebQusUEHl7CfzpUw6OAEy1btmTr1q3Mnz+fq1evEh8fz6hRowDLPsyScxQci4iIiIiIiIiIPCgOR8H0BvDtYCheGXptgLb/hYLF7FZSRkYGNWrU4P333yckJISXXnqJfv362QTHkyZNYtOmTSxfvpzo6GgmTJjA4MGDWb169W3N1eGlV5l5yIU5+0zadujEtKlTWfPNQs6cOQNAixYtGD9+PH369KFAgQJUqFCBVq1aAeDgoKgzJ+lqioiIiIiIiIiI2NvZw/DlizC3HaQmQYd50DUSfKrauzJ8fX0JCgqyORYYGMiRI0cASE5OZtiwYYwdO5a2bdsSHBzM66+/znPPPXfLD6wzTZMlO44S+t+N7I2/wNh/BzOzSy2aN6oPwMGDB619Bw4cSGJiIkeOHCEhIYF27doBULZs2Zw4Xcmk4FhERERERERERCSHnDhxgq5du+Ll5UWBAgUICgpi/fr1AKSlpTF06FCCg4MpWLAgvr6+vNCxPUcW9Icpj8GhddD0HeizjW2XfGjeogVubm64u7tTt25dEhISrPOULl0awzBsXuHh4Ta1bN++nWbNmlGoUCEKFSpE06ZN2bZt222fU7169YiLi7M5tn//fkqVKmU9r7S0NBwdHW36ODo6kpGR8Y/jJ1y6Qq950YR9tZsgPw9WvdGADrVKYhgGMTExgCW8zsowDPz8/HBxceGLL76gZMmS1KhR47bPTW4un70LEBERERERERERyQ0SExOpV68e9evXZ+XKlXh5eXH48GG8vb0BuHz5Mjt37mT48OFUDw7m/PZFDBrxIaE/prL7kx7kazESPHzZunUrLVu2JCwsjAkTJuDs7MzevXtxcnKyme+dd96hd+/e1vdubm7Wry9dukRoaCitW7dmy5YtmKbJmDFjaNmyJUeOHMHd3f2Wz2vAgAHUrVuXMWPG0LFjR3bt2sXEiRN57733APDw8KBhw4aEh4fj5uZGqVKlWL9+PXPnzmXs2LHWcU6ePMnJkyfZv38/AL/++iurog8ybcd5kh1ceS4gGe+kzZw54sLV855s376dAQMG8NRTTxEQEGAdZ9y4cYSGhuLg4MA333zDBx98wOLFi28IruXuKDgWERERERERERHJAWPHjsXX15e5c+daj5UpU8b6taenJ2vXroUjW2DVq3AihukvVafyiC3EVnydqh6WVbUDBgygT58+DB8+3PrZChUq3DCfu7s7Pj4+2dayb98+zp49S0REhLWG0aNHs2DBAuLi4qhVq9Ytn1ft2rVZunQpb775JqNHjyYgIIDRo0fz2muvWft8+eWXDBs2jBdffJGzZ89SqlQpRo8ezeuvv27tM23aNCIiIqzvW7duDUDV54cR+dEwko4f4LXXxjNq1CiuXLlCqVKl6NmzJ0OGDLGpZ9WqVYwZM4YrV65QrVo1li1bxpNPPnnL5yO3RsGxiIiIiIiIiIhIDli6dCmhoaF07NiRdevW4efnR48ePejTpw+GYUDiUfh+BOz9GjxKwDOfciHJH0bUpXDhwgCcPn2azZs388ILL1C/fn32799PxYoVGTlyJE2bNrWZb/z48bz//vuULFmS9u3bExYWhrOzMwAVK1bEy8uLWbNm8fbbbwMwc+ZMAgICqFy58m2fW+vWra1Bb3Z8fHyYPXv2344xcuRIRo4cyZbDZxi0+P84cT6Z1xqVo1/T8jjncwCfGmzZsuUfa/nxxx9vu365fQqORUREREREREREcsDhw4eZOnUqAwYMIDw8nJiYGPr27QtXU3m92hX4+WPAhIZDod4bpOLEoMaNadu2Lf7+/tYxAEaMGMG4ceMICQlhyZIltGzZkujoaKpVqwZAv379CAkJoWjRomzbto3w8HB+++03Pv30U8CyGjkqKoqnn36a999/H7Dsi7x27VpcXFzu/8UBUtLSGb8mjlk//0apIq4sebUuNUsVtkst8s8UHIuIiIiIiIiIiOSAjIwMatWqZQ1qQ6pX58DPkUx5fxiv9y4AlZ+B5hFQKICrV6/S6YUXSExMZPny5TZjAPTq1Yvu3btbxgkJYd26dUybNo1PPvkEgIEDB1o/ExwcjIeHBx07duTDDz+kaNGiJCcn0717dx5//HEWLFhAeno648ePp127duzYsYOCBQver8sCwN748wxYFMOB05fo9HgAb7YKxNVZ0eSDTHdHREREREREREQkB/j6+hIUFGR5E78TVocTeGYjHydehZdWQam6AFy9epXnn3+ePXv2EBUVRdGiRW3GAP4aJ1NQUBBHjhy56dyPPfYYAAcPHqRo0aIsXLiQQ4cO8fPPP1sfGrdw4UIKFy7M//73Pzp16pRj5/13rqZn8EnUIT7+4QBFCjrzefdHaVjB677MLXdHwbGIiIiIiIiIiEgOqFevHnG/7oGlr0HMAijoxf6iLSj1yAFraJyWlsZzzz3H3r17iYqKuuHhdqVLl8bPz4+4uDib4/v376dq1ao3nTsmJgb4K3i+fPkyhmHg4OBg7ePg4IBhGNZVzffabwlJDFwcw64jibSt5sfodpUp5Op8X+aWu6fgWERERERERERE5G6lpTCgkTd1e3/BmEl76Ni5O7ucajPx3X689957gGWlcfv27dm+fTuRkZEYhsHJkycB8PT0xMXFBcMwCAsLY8SIEQQHBxMSEsLixYvZsmULkydPBmDz5s1s2bKFxo0b4+npyfbt2xkwYABPPfUUAQEBADRv3pywsDBee+01+vXrR0ZGBh988AGOjo40adLknl4K0zSZv+UP3vt2H875HJj4fAhPVfO7p3NKzlNwLCIiIiIiIiIicqdME/atgDXDqZ34B0sHPcGb355m9KufEhDwHaNHj+a1114D4NixYyxbtgyAmjVr2gwze/ZsunXrBkD//v25cuUKgwYN4syZM1SuXJlVq1ZZH4yXP39+Fi1aREREBFeuXKFUqVL07NmTIUOGWMerVKkSkZGRREREUKdOHQzDoHr16qxatcr6IL574eT5FMK++j82HkjgiQpejH02GB/PAvdsPrl3FByLiIiIiIiIiIjciZN7YXU4/L4RvAKh81JaP9KY1mOz7166dGlM07yloYcOHcrQoUOzbatRowZbtmz5xzGaN29O8+bNb2m+nLAsJp63l+4lLd1k9NNV6PRYAIZh3Lf5JWcpOBYREREREREREbkdSQmwbgxEz4ECntBqPNR8CRzzZtSWeDmVt5buZcXuE4QEFOKjDtUpU6ygvcuSu5Q3v5tFRERERERERERuV3oabJsJUR9A6iV49BVoOBRci9i7MruJijvNkK92czYplbCWFen1RFnyOTr88wflgafgWERERERERERE5J8cWAurh8GZA/BIE2j5PnhXsndVdnM59SpjVsayYOsRKhR347NutalSwtPeZUkOUnAsIiIiIiIiIiJyM3/uhzVvwsG1UOQReH4RVGgJeXjv3ug/zjFwcQxHzl7mlSfKMrB5BQo4Odq7LMlhCo5FRERERERERESul3wO1o+FbTPAyRVajLFsTZHP2d6V2U3q1Qw+/mE/n0QdwtfThS96Ps7jZYvauyy5RxQci4iIiIiIiIhI3rZ7MfwwCs4fA88SUKYhxK2yhMc1u0Ljt8DNy95V2lXcyYv0XxRD7IkLdKjlz9ttgnAv4GTvsuQeUnAsIiIiIiIiIiJ51+7FENkP0pIt788fg5gFULQCdFkGvsH2rc/O0jNMZv10mPFr9uPhko+ZXWrRPKi4vcuS+0DBsYiIiIiIiIiI5F0/jPorNM7qanKeD42Pnr3MoCX/x7bfztIiqDjvP1OVom757V2W3CcKjkVEREREREREJG9KPALnj2bfdv7Y/a3lAWKaJot3HGVU5K84GAbj21fj2RolMPLwAwHzIgXHIiIiIiIiIiKStySdgY3jYfunN+/j6X//6rGzpbviGbcmjvjEZHw2/UBRNyd+OX6ROmWLMq59MP6FXe1dotiBgmMREREREREREckbrlyCLVPh54mQlgTVX4TiVeGHEbbbVTi5QNN37FfnfbR0VzzDvtlDclo6ACcvpHDyQgpPV/fjow7VcXDQKuO8SsGxiIiIiIiIiIjkbldTYefnsH4sJJ2GSm0swbBXRUu7a2HLXsfnj1lWGjd9B4I72Lfm+2TcmjhraJzV9t/PKTTO4xQci4iIiIiIiIhI7pSRAb98Az+OhnO/Q6l68NwCKPmobb/gDnkmKM4qPcMkPjGbBwMCx29yXPIOBcciIiIiIv/P3n3HdVW+fxx/HaYIuAABURAcuHMmoiXkLv1qy5yZldXXxrdMTSsTU8vcOfq5d2ppaWlDTcMRYmmihotcCIoKDmSIjPP744OfILW0FFTez8fj85Bzn+vc57o5YI+u7q4jIiIi9xbThEPr4YdhkLAbPGtBt2VQpRXoBW8A/BZ/gbdX7Lnu+XKlnAowG7kTqXAsIiIiIiIiIiL3jrgd8MNQOLoZSvnCozOg9pNgY1PYmd0RUjKyGLf2APMjjlLG2ZGeTfxYvv046Zk51hgne1sGtAksxCzlTqDCsYiIiIiIiIiI3P0SYyx9ivd9DcXdoe1H0LA32DkWdmZ3BNM0WROdQNjXezl18RLdG/syoE01SjrZ08C3NGPWHCD+fDo+pZwY0CaQTvV8CjtlKWQqHIuIiIiIiIiIyN0r+QSEj4Kdi8DeCZoPguBXwNG1sDO7Y8SdS2PoV9Gs33+a6t4l+L8e9annW9p6vlM9HzrV8yE8PJyQkJDCS1TuKCoci4iIiIiIiIjI3Sf9HGyZCNumQU42NHoeHhwALh6FndkdIzM7hzlbjjDxhxgA3nm4Or2bVsTOVm075O+pcCwiIiIiIiIiInePzHTYNh22TIBLFyz9i0PfhjL+hZ3ZHWXHsXO8s2IP+xMu0rK6J8M61sRHL7yTm6D/vCAiIiIiIiIiIne+7CzYMR8m1be8/K58I3hpMzw+s0CKxidPnqRXr154eHhQrFgxatSowcaNG63nhwwZQrVq1XB2dqZ06dK0aNGCiIiIfHPMmDGD0NBQSpUqhWEYHD169Kr7HDx4kE6dOuHu7o6rqytBQUF8//33+WLWr19PcHAwrq6ueHl58dZbb5GVlQXAhbRM3l6xhyemRXAhPZPpPRswq1dDFY3lpmnHsYiIiIiIiIiI3LlME/atgg3DIfEg+DS0FIsrNiuwFM6fP0/Tpk1p1qwZ33zzDR4eHhw+fJiyZctaYwIDA5k6dSr+/v6kp6czYcIE2rZtS0xMDJ6engCkpaXRunVrOnbsyBtvvHHNe7Vv356AgADWr1+Ps7Mz06ZNo2PHjuzdu5dKlSqxa9cuHn74YQYNGsSCBQuIj4/npZdeIisriwd79mP46r2cTb3Ms039eaNVVVwcVf6Tf0Y/OSIiIiIiIiIicmc6shl+CIP47eBeFZ5aBNXag2EUaBqjR4/G29ubBQsWWMf8/fPvcu7Ro0e+4/HjxzN79myioqJo06YNAK+//joA27dvv+Z9EhMTiYmJYfr06dx3330AjBo1igkTJrBz504qVarEZ599Ro0aNRg2bBgAlStX5s133+el3j1ZlhNEvcrlmNf7fmr5lLw1i5ciS60qRERERERERETkznJyNyx6HOa3h+QT8J/J8N+tUL1DgReNAVauXEnjxo156qmnKFu2LHXr1mXKlCmYpnnN+MuXLzNjxgxKlChB3bp1b/g+bm5uVK9enYULF5KSkkJ2djYzZszA1dWVpk2bApCRkUGxYsUsX2dlM2l9DGHfxpCdmUGPytl8+d9guAiKqAAAIABJREFUFY3lltCOYxERERERERERuTOcPQI/joQ9y6BYKWj1Ptz/AtgXbn/ew4cP88knn/DGG28waNAgoqKiePXVVwF45ZVXrHGrV6+mS5cupKWl4e3tzbp166xtKm6EYRisW7eORx99lBIlSmBjY0OZMmX47rvv8Pb2BqBNmzZMmDCBoeP+j43ZVfn9WBw5O5YDULuMia1NwRfW5d6kwrGIiIiIiIiIiBSulDOwaTRsnws2dtDsDWj6OjiVKuzMAMjJyaFhw4Z8+OGHANSrV4+YmBimTp2ar3AcGhpKVFQUiYmJzJw5k86dO7N161Zr0ffvmKZJ3759cXNzY/PmzTg5OTFr1iwef/xxfvnlF3x8fGgQ3JwHerzB8Lf7Y2ZdopijI0Pfe4/Bg3diY6PmAnLr6KdJREREREREREQKR8ZF+PFD+Pg++GU21OsOr+2ElmF3TNEYwNvbmxo1auQbq169OrGxsfnGnJ2dqVy5MkFBQcyePRt7e3tmzZp1w/fZsGEDq1atYsmSJTRt2pT69evzySef4OzszJw5c/jsl1hajN/IiQotGPXVdg4dPsrZpCQ6duwIQEBAwL9frEgu7TgWEREREREREZGClZVh2V28aQykJUKNjvDQEHCvUtiZXVPTpk05cOBAvrGDBw/i5+f3l9fl5OSQkZFxw/dJS0sDuGrncJYJS38+RmrqHu6vWIaRj9aiiqer9fySJUuoUKEC9evXv+F7ifydAt1xbBiGt2EY8w3DOGMYxiXDMPYahtE8z3nDMIwwwzBOGIaRbhhGuGEYNf80R2nDMBYahnEh97PQMIxSf4qpbRjGxtw54g3DeM8w8ndONwzj8dz7Z+T++ejtXb2IiIiIiIiISBGXkw27lsKUhvD9W+BZA/psgM4L7tiiMcAbb7xBZGQkI0eO5Pfff2fZsmVMmjSJl19+GYDk5GTeffddtm3bRmxsLDt27ODZZ58lLi6Ozp07W+dJSEggKiqKgwcPArB3716ioqI4e/YsAE2aNKFMmTL07t2bXbt2sfu3fTR/4jmOHjlCpk89Rj9eh6UvBLFywTT27NlDdHQ0w4cPZ9SoUUyaNAlbW9uC/+bIPavACse5xd2fAAN4BKgOvAqczhM2EHgzd7xR7rl1hmG45olZDNQH2uZ+6gML89ynBLAOOJU7x/+AAUC/PDFNgM+AT4G6uX8uMwyj8S1bsIiIiIiIiIiIWJgmHFwL0x+EFS9aXnzX40t4+mvwaVDY2f2tRo0asXLlSj7//HNq1arFO++8w/Dhw+nbty8AdnZ2REdH8+ijj1KlShU6dOhAUlISmzZtok6dOtZ5pk2bRr169ejevTsAjzzyCPXq1ePrr78GwN3dne+//56UlBQebB5K/YYNiYzYQqcBE9g69nk6N6qAjY3Bd999xwMPPEDDhg355ptv+Oqrr+jUqVPBf2PknlaQrSoGAidN03w6z9iRK1/k7gh+HRhlmuYXuWO9sBSPuwHTDcOojqVY3Mw0za25MS8Cmw3DCDRN8wDQHSgO9DJNMx34zTCMakA/wzDGm6Zp5t7nR9M0R+befqRhGKG5411v1zdAREREREREROReFhYWxrBhw/KNeXqUIeGjxnDsJ4xhybmjW4CW1pi+ffsydepUAIYMGcKyZcs4fvw4Dg4O1K9fn+HDhxMcHGyNz8jIoH///ixZsoT09HRatGjBJ598Qvny5a0xsbGxvPzyy2zYsAEnJye6devG2LFjcXBw+Edre+SRR3jkkUeuea548eKsWLHib+cICwsjLCzsL2N8q9aicq8POVDvJA08nBnZqTZNKrnli9mwYcMN5y3yTxVkq4pOwDbDMD4zDOO0YRhRhmG8kqeFhD/gBay9ckFu4XcTcOVvhiZAChCRZ96fgNQ/xWzOvfaKNUA5oGKemLXktybPHCIiIiIiIiIi8g8EBgZy8uRJTu7ZxMlPOrGn12VIPAgPj+Xk8WOWc7mfVatWAeRr5xAYGMjUqVPZs2cPW7Zswd/fn7Zt23Lq1ClrzOuvv84XX3zBkiVL2Lx5M8nJybRv357s7GwAsrOzeeSRR7h48SKbN29myZIlLF++nDfffLNgvxk3ITvHZMHWo7QYt5F1e0/xZquqfPe/B64qGosUlILccRwA9AUmAKOwtIiYnHtuCpaiMVhaTOR1CvDJ/doLOJO7axgA0zRNwzBO57neC4i7xhxXzh3J/fNa9/HiGgzDeAF4AcDT05Pw8PDrrfGelJKSUuTWLH/Q8y/a9PxFPwNFm55/0abnX7Tp+Rdtev5F2799/kePHuXypTTMlS/jlbCBbFtHjtfsyf7yHchOc4LfD+eLnzZtGhUqVMA0Tet9r+wajo2NBeDRRx9l9uzZzJ8/n/vvv5+UlBRmzZrFwIEDsbe3Jzk5mb59+9KlSxfGjRvH/fffz7Zt24iOjmbp0qUkJydjb29P7969GTNmDG3btsXZ2fkfr/F2OJaczbzoyxy5kENNNxueruGIp208W7fEF2ge+v2XvAqycGwDbDdNc3Du8U7DMKoAL2MpHN+xTNOcAcwAaNiwoRkSElK4CRWw8PBwitqa5Q96/kWbnr/oZ6Bo0/Mv2vT8izY9/6JNz79o+1fPP+0s4bMP8/mJ4zQYGI9jcVcaN2vCB93780BAwFXhKSkpbNy4kaFDh173npcvX2bSpEmUKFGCXr164enpyYYNG8jKyuL111/Hw8PDGjts2DBSU1MJCQlhw4YNVK9ePd9O5po1a/LBBx9QvHjxO+ZnPCUjiwnrDjJ36xHKODvwcZca/Oe+cvzxP+gXLP3+S14FWTg+Cez909g+LC+vA0jI/dMTiM0T45nnXALgYRiGcWXXcW6ri7J/ivH8030885z7q5gERERERERERETkxl1Og23/B1s+pnFOEvNea0m1Tv05fcmWESNGEBwcTHR0NG5u+VsuLF68mMuXL9OrV6+rply9ejVdunQhLS0Nb29v1q1bh6enpZSTkJCAra0t7u7u+a7x9PQkISHBGnMl/gp3d3dsbW2tMYVtTXQCYV9Hk5B8iW73+zKwTTVKFrcv7LRErAqycPwTEPinsarAsdyvj2Ap3LYCfgEwDKMY8AAwIDdmK+CCpUfxlT7HTQDnPMdbgY8Mwyhmmual3LFWwAngaJ6YVsCYPLm0In/vZBERERERERERuZ7sTNi5EMI/gpQEqNqOds8OAc+a1pCgoCACAgKYP38+/fr1y3f5zJkz6dixY75dw1eEhoYSFRVFYmIiM2fOpHPnzmzduhVvb+/bvqzbLf58OkO/iuaHfaeo5uXK1O71qe9burDTErlKQb4cbwIQZBjGO4ZhVDYM40ngNWAqWHoVAxOBtwzDeMwwjFrAPCwvw1ucG7MP+B6YbhhGE8MwmgDTgdWmaR7Ivc9iIA2YZxhGLcMwHgMGAePz9Eb+GHjIMIxBhmFUMwxjMBCae38REREREREREbke04ToFTC1Max+A0r7Qe/vodvSfEVjABcXF2rWrElMTEy+8aioKLZv306fPn2ueQtnZ2cqV65MUFAQs2fPxt7enlmzZgHg5eVFdnY2iYmJ+a45deoUXl5e1pi8L9MDSExMJDs72xpT0DKzc5ix6RAtx23kp98Tefvhaqx6tZmKxnLHKrDCsWmavwCdgM7Ab8BIYAjwSZ6w0VgKzFOB7YA30No0zYt5YroBu4A1uZ9dQM8897mAZfdwudw5pgLjgPF5YiKALsAzwG7gaeAp0zS33ar1ioiIiIiIiIjccw6Hw8xQWPYM2DpAlyXw7Brwa3LN8EuXLrF///6rdgrPmDEDf39/WrZseUO3zcnJISMjA4AGDRpgb2/PunXrrOfj4uLYt28fwcHBADRp0oR9+/YRFxdnjVm3bh2Ojo40aNDgJhZ8a/wae44Ok7fwwbf7aVrZjXX9HuSFBythb1uQezpFbk5BtqrANM1vgG/+4rwJhOV+rhdzDujxN/fZAzz4NzHLgeV/FSMiIiIiIiIiIsCJKPghDA7/CCXKQ8dP4L4uYGObL6x///506NABX19fTp8+zfDhw0lNTc3XxzgtLY1PP/2UgQMHXvUSuOTkZEaPHk2HDh3w9vbmzJkzTJ06lbi4OOuL7kqWLMlzzz3HwIEDKVu2LG5ubvTr1486depYC9GtW7emZs2aPP3004wbN46kpCQGDBhAnz59KFGixO39XuVxIT2T0d/vZ/HPsXi6FmNajwa0qelZaC+/E7kZBVo4FhERERERERGRu0jSIdgwAqK/BKfS0HokNHoe7ItdMzwuLo6uXbuSmJiIh4cHQUFBREZG4ufnZ4357LPPSE1NpXfv3lddb2dnR3R0NHPmzCEpKQk3NzcaNWrEpk2bqFOnjjVu4sSJ2NnZ8dRTT5Genk6LFi1YsGABtraWQratrS3ffPMNffv2pWnTpjg5OdG9e3fGjBlz1T1vB9M0+XrXCYav3sfZ1Ax6B/vTr3VVXBxVipO7h35aRURERERERESKut2fw/r3aX4hDnaWh6b/g9P74Nf5lpYUD/SHpq9BsZJ/Oc3SpUv/9la9e/e+ZtEYoHjx4qxYseJv53B0dGTy5MlMnjz5ujG+vr6sXr36b+e61Y4mpjLkq9/YHJNInfIlmde7EbV8/vr7JnInUuFYRERERERERKQo2/05rHoNMtMxAC4ch2/7AwY0fBaaDwTXwnmh3N0kIyubGRsPM/nH33GwteH9jjXp3tgPWxu1pZC7kwrHIiIiIiIiIiJF2fr3ITP96nFXL2g/vuDzuQtFHk7inRV7OHQmlUdqe/Nehxp4lrh2Ow+Ru4UKxyIiIiIiIiIiRVXGRcsO42u5mFCwudyFzqZe5oNv97F8RxwVyjgxt3cjQgPLFnZaIreECsciIiIiIiIiIkVNxkX4eQZETLl+TMnyBZfPXcY0TZbtiOPDb/dx8VIWfUMq8epDVXBysC3s1ERuGRWORURERERERESKikvJloLx1imQfg4qt4LyDeGnifnbVdg7QYv3Ci/PO9jvpy/y9orf+PnIWRpVLM3IR2tT1dO1sNMSueVUOBYRERERERERudddSoZt0y0F40vnoUpraD4IyjewnC8TAOvfx7wQh1GyvKVoXKdz4eZ8h7mUmc2UDb8zfdMhnB3t+Ojx2jzZoAI2evmd3KNUOBYRERERERERuVddupBbMJ5qKRhXbQvNB4JPg/xxdTpDnc5sDA8nJCSkUFK9k208eIYhK38j9mwaj9X34Z2Hq+Pm4ljYaYncVioci4iIiIiIiIjca9LPWwrGkVMtxeOq7XILxvULO7O7yumLlxi+eh+rdp0gwN2ZxX0aE1zJvbDTEikQKhyLiIiIiIiIiNwr0s/Dtmmw9RPIuACBD1sKxuXqFXZmd7SVO+MZs+YAJ86nU66UE2+2qkrq5SxGrzlARlYOb7SsykshATja6eV3UnSocCwiIiIiIiIicrdLPweR/weR0ywF42rtLQVj7/sKO7M73sqd8Qz+cg/pmdkAxJ9P583luzBNaFrZjRGdauPv7lzIWYoUPBWORURERERERETuVunnLLuLt02DjOTcgvFb4F2nsDO7a4xZc8BaNL7CNKF0cXsWPdcYw9DL76RoUuFYRERERERERORuk3YWIj+x9DHOSIbqHSwFY6/ahZ3ZXcU0TU6cT7/mufNpmSoaS5GmwrGIiIiIiIiIyN0i7SxsnQLbZsDli1CjIzw4ELxqFXZmd5X0y9l8vSuehZHHMK8TU66UU4HmJHKnUeFYREREREREROROl5pkKRj/PAMup0CNTpYexp41Czuzu8qhMyksijzG8h1xXLyURaCnK080KM/q3Se4lJljjXOyt2VAm8BCzFSk8NkUdgIiIiIiIiIiInIdqUnwQxh8XAe2TIAqreC/W6Hz/AItGoeFhWEYRr6Pl5eX9XxKSgqvvvoq5cuXx8nJicDAQCZMmJBvjpCQkKvm6NKli/V8eHj4VeevfJYtW2aNO3fuHD179qRkyZKULFmSnj17cv78+evmnpmdw3d7TtJtZiQtxm1kUeQxQgPLsuylJnz/+gOMffI+Rj1WB59SThiATyknPnysNp3q+dy6b6DIXUg7jkVEREREREREbpGwsDCGDRuWb8zT05OEhAQAhgwZwrJlyzh+/DgODg7Ur1+f4cOHExwcDMDRo0fx9/e/5tyj33uTAU+OBaBPnz5s2LCBEydO4OLiQnBwMKNGjaJ69erW+F9//ZW33nqLX375BVtbWx5//HHGjx+Pi4vLP1pbYGAg4eHh1mNbW1vr1/369eOHH35g4cKF+Pv7s2nTJvr06YO7uzs9e/a0xvXu3ZsPPvjAeuzk9Ec7iODgYE6ePJnvnpMmTWLy5Mm0a9fOOtatWzdiY2P5/vvvAXj++efp2bMnq1atyndtwoVLLPk5liU/x3L6YgY+pZwY0CaQzg0r4OHqmC+2Uz0fFYpF/kSFYxERERERERGRW+ivCqyBgYFMnToVf39/0tPTmTBhAm3btiUmJgZPT08qlC7GySWvQ9RiyEqHwEdYca4qLw8azhPPvGydp2HDhjz99NNUqFCBs2fPEhYWRsuWLTl69Cj29vacOHGCli1b8uSTTzJlyhSSk5N5/fXXeeaZZ1i+fPk/WpednV2+XcZ5RURE0LNnT0JDQwGoWLEis2fPZtu2bfkKx8WLF7/uHA4ODledW758OV27drUWu/ft28f333/Pli1baNKkCQDTp0/ngQce4MCBA1StWpWIQ0ksijzG2r2nyDFNmlf14IPGfoRWK4utjV52J3KjVDgWEREREREREbmF/qrA2qNHj3zH48ePZ/bs2URtDadN8T3Y/jIbr6xL0OhxeHAAeATyZatWtGzZMt9O5BdffNH6dcWKFRkxYgT33Xcfhw8fJjAwkNWrV2NjY8Mnn3xiLVxPmzaNOnXq8Pvvv1O5cuWbXtfhw4cpV64cjo6ONG7cmA8++ICAgAAAmjVrxqpVq3j++eepUKECERERREVFMWDAgHxzLF26lKVLl+Lp6Um7du0YOnQorq6u17xfeHg4MTExLF682Dq2detW6w7rK5o2bYqzszOj5n3FMff7OXwmldLF7Xm+mT/dGvvi5+Z802sVERWORURERERERERuqb8qsOZ1+fJlZkweT4nijtTd9ioUy4TaT1oKxu5VrHOtX7+ezz///Lr3S01NZe7cufj6+lKxYkUAMjIysLe3z7fb+UpbiC1bttx04bhx48bMmzePatWqcfr0aUaMGEFwcDDR0dG4ubkxadIkXnzxRXx9fbGzs5SbJk+eTPv27a1zdOvWDT8/P8qVK0d0dDSDBw9m9+7drF279pr3nDFjBnXr1qVhw4bWsYSEBDw8PDAMy87hPXEXWBR5jMv2rqyMiCb0qaaM73wfD9f2ppi97TXnFZEbo8KxiIiIiIiIiMgt8ncFVoDVq1fTpctTpKWl4e1iw7puxfFs9GhuwTh/QXfWrFl4eHjQsWPHq+71ySefMHDgQFJTUwkMDGT9+vU4Olp69z700EP069ePUaNG0a9fP1JTUxk0aBDAVX2Eb0TeHsMAQUFBBAQEMH/+fPr168fkyZOJiIjg66+/xs/Pj02bNtG/f38qVqxI27ZtAXjhhRes19euXZuAgAAaN27Mr7/+Sv369fPNn5SUxJdffsn48eOvysUElm0/zqJtsew6fh4ne1uKO9jSK7giE/s2vem1ici12RR2AiIiIiIiIiIi94p27drRuXNn6tSpQ8uWLVm9ejU5OTnMnz/fEnAxgdCsH4nqU4yIZ11oe38VOq8pyckmYVcVjbOyspg7dy69evXC3t7+qnt1796dnTt3snHjRqpWrcqTTz5JWloaADVr1mT+/PlMnDgRJycnvLy88Pf3x9PTExubf18OcnFxoWbNmsTExJCens7gwYMZPXo0HTp0oE6dOrzyyit06dKFsWPHXneOhg0bYmtrS0xMzFXnFixYgK2tLd27d7eOHU1MZfupHGLjE+i/bBcplzIJ61CDyLcfIiv1PPdVrfiv1yUif9COYxERERERERGR28RaYN27C74bBDvm4pydSeUHn6Lyg/0JcqtElSpVmDVrFkOGDMl37apVq0hISOD555+/5twlS5akZMmSVKlShaCgIEqXLs0XX3xhfRldt27d6NatG6dOncLZ2RnDMBg/fvw122bcrEuXLrF//35CQ0PJzMwkMzMzX1sMsLwUMCcn57pz7Nmzh+zsbLy9va86N2vWLJ588kmcXVxZG53AwshjbI5JJCfdnZzL6bzT0JY+TzTHMAwiIiJITU3N1/dYRP49FY5FRERERERERG6TS6ePsH/XL4QaP8PPjnBfV3igH7hVssbk5OSQkZFx1bUzZ86kefPmVK1a9W/vY5ompmlecx5PT08A5syZQ7FixWjVqtVNr6N///506NABX19fTp8+zfDhw0lNTaVXr16UKFGC5s2bM2jQIFxcXPDz82Pjxo0sWLCA0aNHA3Do0CE+/fRTHn74Ydzd3dm7dy9vvvkm9erVo2nT/O0ltmzZwt69e2n9whAeGP0jJy9cwqtEMfq1qkqXRi3odfwrJr8/kNrlSwGWFwW2b9+ewMDAm16XiFyfCsciIiIiIiIiIreItcBa2pHTP0xh+PRlpKZeplfXniS3epPRs5bRwTsR71R7zpw5w9SpU4mLi6Nz58755omNjWXNmjUsWLDgqnv8/vvvfPHFF7Rs2RIPDw/i4uIYNWoUjo6O+V5GN2XKFJo0aYKrqyvr1q1jwIABjBo1ilKlSt30uuLi4ujatSuJiYl4eHgQFBREZGQkfn5+ACxdupTBgwfTvXt3zp49i5+fH8OHD+eVV14BwMHBgfXr1/Pxxx+TkpJChQoVeOSRRxg6dKh1p7Jpmmw7cpbn3x6FvVsFVpx05YEqLoT9pyYtqpXFztbSYmPx4sW8+uqrtGnTBoD//Oc/TJky5abXJCJ/TYVjEREREREREZFbJO5IDF0fe4TEC6l4FDcIqulP5Nz/wy+4NWlpaURHRzNnzhySkpJwc3OjUaNGbNq0iTp16uSbZ/bs2ZQsWZLHH3/8qns4OjoSHh7OuHHjOH/+PJ6enjz44INs3boVLy8va9zPP//M0KFDSUlJoVq1akyfPt3axuJmLV269C/Pe3l5MXfu3Ouer1ChAhs3brzmueRLmaz4NZ5FkceIOZ1CyRav8d7A8nQP8sPf3fmq+NKlS7No0aKbW4CI3DQVjkVERERERERE/q0LcbBlAkvrRsB9jlD3WUtLitIVrSHFixdnxYoVNzTdsGHDGDZs2DXPVahQge++++5v57jWbuU7yd4TySyMPMZXUfGkXc7mvvIlGf1EHTrUKYeTg+3fTyAit5UKxyIiIiIiIiIi/9T547BlAuxcCKYJ9bpDs35Q2q+wM7sjXcrM5rvfTrIoMpYdx87haGdDx7rl6BHkR53yN99CQ0RuHxWORURERERERERu1vlY2Dwedua2TKjXw7LDuJRv4eZ1hzp+No1Pt8Xy+fbjnE29jL+7M+8+Up0nGpSnVHGHwk5PRK5BhWMRERERERERkRt1PhY2j4Odn1qO6z8Nzd6AUhUKN687UHaOycaDp1m49RjhB89gYxi0qu5JjyA/giu5YWNjFHaKIvIXVDgWEREREREREfk7545ZCsZRn4JhAw16WQrGJcsXdmZ3nMSUDD7ffpxPI2OJP59OWVdHXn2oCl3vr4B3SafCTk9EbpAKxyIiIiIiIiIi13PuaG7BeHFuwbh3bsHYp7Azu6OYpsmOY+dYGHmMb/ecJDPbpEmAG+88Up1WNTyxt7Up7BRF5CapcCwiIiIiIiIisvtzWP8+XIiz7CIO+i+c3gu7loJhCw2fhaavq2D8JykZWazcGc+iyGPsT7iIazE7ujf2o0eQL5XLuhZ2eiLyL6hwLCIiIiIiIiJF2+7PYdVrkJluOb5wHNa8bSkY39/HUjAu4V24Od5hDiRcZFHkMVbsjCclI4ua5Uow6rHa/KduOYo7qNwkci/Qb7KIiIiIiIiIFG3r3/+jaJyXiye0+6jg87lDXc7K4fvoBBZFHuPnI2dxsLOhfR1vegb5UbdCKQxDL7sTuZeocCwiIiIiIiIiRZNpwtEtlh3G13LxZMHmc4eKP5/O4m3H+OyX4ySmXMa3THHefrgaTzaoQGlnh8JOT0RuExWORURERERERKRoycmG/athy0Q48avlpXdmztVxJcsXfG6FZOXOeMasOUD8+XR8IjfQv1VVSrs4sCgylg37TwHwUDVPejbx44HK7tjYaHexyL1OhWMRERERERERKRoyL8GuxRAxGc4ehjIB0H4C2DrAt/3zt6uwd4IW7xVergVo5c54Bn+5h/TMbMCyw7jfsl2YgLuLA31DKtO1sS8+pZwKN1ERKVAqHIuIiIiIiIjIvS39HPwyG7ZNg9QzUK4ePDkfqncAG1tLjK2DpdfxhTjLTuMW70GdzoWbdwEZ/f1+a9H4ChMoXdyeiEEtcLCzKZzERKRQ6TdfRERERERERG6ZsLAwDMPI9/Hy8rKeN02TsLAwypUrh5OTEyEhIURHR181z5o1a2jSpAnFixenVKlSPPTQQ9e8X2JiIj4+PhiGQWJionX86NGjf+RQvAxG8wEYAw/xfcUh0OdHqNnpj6IxWIrEb/wGYectf97jRWPTNNlx7BxDVv7GiQuXrhlzPi1TRWORIkw7jkVERERERETklgoMDCQ8PNx6bGv7R4F29OjRjBs3jnnz5hEYGMj7779Pq1atOHDgAK6urgCsXLmS3r17M3LkSObNm0dOTg6//vrrNe/Vu3dv6taty4kTJ/4YPLUX1owE4PseLtwX0hEaPQ9lq1GmTBkwim5/3kNnUvhqZzwro04QezaNYvY2ONnbXrXjGKCcWlOIFGkqHIuIiIiIiIjILWVnZ5dvl/EVpmkyceJEBg0axOOPPw7A/PnzKVu2LIvT9f5fAAAgAElEQVQXL+bFF18kOzub1157jdGjR9OnTx/rtdWrV79qvo8//pi0tDTeeecdvv32W4jdBmvmQ8waSHEEwO3ZJXiFtr9NK707nLmYwapdJ/gqKp5dcRewMaBpZXf+16IKbWp58cPeU/l6HAM42dsyoE1gIWYtIoVN/7+BiIiIiIiIiNxShw8fply5cvj7+9OlSxcOHz4MwJEjR0hISKB169bWWCcnJx588EEiIiIA2LFjB8ePH8fBwYH69evj5eVF69at2blzZ7577Ny5k48++ogF8+ZiE7vVMri0C8TvgNB34bm1ADzWqy9ly5aladOmLF++vABWf2dIu5zFyp3x9JrzM0Efruf91XvJNk3efaQ6Wwe3YOFzjXm8QXlcHO3oVM+HDx+rbX35nU8pJz58rDad6vkU8ipEpDBpx7GIiIiIiIiI3DKNGzdm3rx5VKtWjdOnTzNixAiCg4OJjo4mISEBAE9Pz3zXeHp6Eh8fD2AtMr/33nuMGzcOf39/pk6dSkhICPv378fb25vU1FS6PPUUk19/DJ+VjxKzY79lolbvQ8hLYO+ES2IiY8eOpWnTptjZ2fH111/z1FNPMX/+fHr06FFw35AClJWdw5bfE1m5M5410adIz8zGp5QTLzUPoFNdH6p4ul732k71fOhUz4fw8HBCQkIKLmkRuWOpcCwiIiIiIiIit0y7du3yHQcFBREQEMD8+fMJCgr62+tzcnIAeOedd3jiiScAmDFjBj/88AMLFizgrdde5LXObWhW6hSPpy2EknWh+SCY/x7U6wn2ll2z7u7uvPnmm9Z5GzZsSGJiIqNHj76nCsemabI77gIrdsazevcJElMuU9LJnkfr+/BoPR8a+JbGxqbo9nQWkX9OhWMRERERERERuW1cXFyoWbMmMTExdOrUCYBTp07h6+trjTl16pS1J7K3tzcANWrUsJ63s7Ojir8vsVs+h+xJrP/pJMeTTeb/agP8hGluAcDLy4u33nqLkSNHXjOXxo0bM3fu3NuxzAIXm5TGyqh4Vu6M53BiKg52NrSsXpaOdX0ICfTA0c727ycREfkL6nEsIiIiIiIiIrfNpUuXrC0m/P398fLyYt26dfnOb968meDgYAAaNGiAo6MjBw4csASc3kfOFy9xaPdW/C7tg8B2rP3mK3bt3kNU1C6ioqKYNWsWAOHh4bz22mvXzSUqKspamL5ZYWFhGIaR73OtFwACvPjiixiGwdixY61jZ8+e5dVXX6VatWo4OTlRoUIF/vvf/5KUlGSNCQ8Pv+oeVz7Lli3jbOpl3p66FMMw8HN35n8tq/LjgFCOfdSemJEPE+pwmDY1vVQ0FpFbQjuORUREREREROSW6d+/Px06dMDX15fTp08zfPhwUlNT6dWrF4Zh8Prrr/PBBx9QrVo1qlatyogRI3BxcaFbt24AlChRgpdefJGh7w6mfMwCKqbsYMoOk3OX7eg5aSNUb0TVP90zMTERgGrVquHu7g7A/Pnzsbe3p169etjY2LBq1SqmTp3KRx999I/XFhgYSHh4uPXY1vbqAu3y5cv5+eefKVeuXL7xEydOEB8fz+jRo6lRowbx8fH07duXrl27snat5UV+wcHBnDx5Mt914ydMZMqUKaw+W5bBI38gM7MYzcO+pF1tb9rU9MSrpBOTJk1i8uTJV7UJERH5N1Q4FhEREREREZFbJi4ujq5du5KYmIiHhwdBQUFERkbi5+cHwMCBA0lPT+fll1/m3LlzNG7cmLVr1+Lq6go5OXDgW8ZU/w2HgAv0mvoTaTl21G/QkB83T8K7ev2bymXEiBEcO3YMW1tbqlatypw5c/5Vf2M7O7vr7jIGOHbsGP/73//44Ycfriri1qpViy+//NJ6XLlyZcaMGUP79u1JTk6mRIkSODg44OXlRXaOSeThJFbsjGfyrEU4VGnGgaQsnmvmT6d6PlT3LpFv7uXLl9O1a1dcXFz+8dpERP5MhWMRERERERERuWWWLl36l+cNwyAsLIywsLA/BrMyYMd8iJgMSTHYl/Jj9IQpjK7bHRyK/+09Q0JCME0z31ivXr3o1avXP1nCdR0+fJhy5crh6OhI48aN+eCDDwgICLAsISuLrl278u6771K9evUbmi85ORlHR0eKFy+OaZrsPZnMyp3xfL3rBKeSMzBORnP5bDyfzJ7HM/95CNtrvOQuPDycmJgYFi9efEvXKiKiwrGIiIiIiIiIFI5LF2D7HIj8P0g5Bd73wRNzoHpHsL2zShaNGzdm3rx5VKtWjdOnTzNixAiCg4OJjo7Gzc2NoUOH4u7uzn//+98bmu/8+fMMGTKErk/3Zvrmo6zcGU/M6RTsbQ1CAsvyaD0f5o9cRKm6dXmuU8vrzjNjxgzq1q1Lw4YNb9VSRUQAFY5FREREREREpKAln7AUi7fPhcsXodJD8NgM8G8OxtW7au8Ef249ERQUREBAAPPnz6d+/frMmzePqKioG5or/vQ5WrRuwwXDlfWurdiw5gCNKpZmRKdaPFLbm9LODiQlJfHYyhWMHz/+uvMkJSXx5Zdf/mWMiMg/pcKxiIiIiIiIiBSMMwfgp0mw+zMws6HmY9D0NctO47uMi4sLNWvWJCYmhuTkZE6ePIm3t7f1fHZ2Nm+99RYTJ04kLi6OjKxsftx/ms+3xrBkeF9M06Txf8fyZFBlOtb1oUKZ/C05FixYgK2tLd27d79uDjcSIyLyT6lwLCIiIiIiIiK3V2wk/PQxHPgW7JygYW9o8jKUrljYmf1jly5dYv/+/YSGhvLSSy/xxBNP5Dvfpk0bunTpwv3tnmTQF7v5ds9Jzl9I5tyXw/BwsWfZilXcX9UH4zo7rGfNmsWTTz5JyZIlr5vDjcSIiPxTKhyLiIiIiIiIyK2XkwMHv4efJsLxbeBUBkIGQ6M+4OxW2NndtP79+9OhQwd8fX05ffo0w4cPJzU1lV69elG2bFnKli1rjT2QcJG0LPhiXwpf2CdR3OE8D1Vy5ceJ7+DqYrJy5ee4utpx6tQpAMqUKYODg4P1+i1btrB3715mzJhx3XxuJEZE5N9Q4VhEREREREREbp2sDNj9OURMgsSDUMoX2o2Bej3AofjfX3+HiouLo2vXriQmJuLh4UFQUBCRkZH4+fkBkHDhEl/vimfFzhPsO5lMSkYWlV0debdLXVrV8OTniC1MjdoBQNWqVfPN/eOPPxISEmI9njlzJtWrV6dp06bXzedGYkRE/g2bwk5ARERERERE5F714YcfYhgGr7zyinUsJSWFV199lfLly+Pk5ERgYCATJky45vWmadKuXTsMw2D58uXW8aNHj/Lcc88REBCAk5MTAQEBDB48mPT09KvmWLRoEXXr1qVYsWK4u7vz9NNP3/qFAly6YGlH8fF98PUrYOcIj8+GV3dC4xfu6qIxwNKlSzlx4gSXL18mPj6eL774ggoBVfh8+3G6z4qkyaj1fPDtfhztbBj2n5qcjIvll88m0bGuD8Ud7AgJCcE0zWt+8haNAebPn8/evXv/Mp8biRER+Te041hERERERETkNoiMjGTGjBnUqVMn33i/fv344YcfWLhwIf7+/mzatIk+ffrg7u5Oz54988WOGzcOG5ur93zt37+f7Oxs/u///o8qVaqwb98+XnjhBZKSkvK1Lpg0aRIffvghY8aMISgoiPT0dA4ePHhrF3oxASI/ge1zISMZAkKg0ycQEArX6d97N7uclcOmg2dYERXPD3tPkZGVQ0W34vyvRRU61vXB3925sFMUEbklVDgWERERERERucUuXLhA9+7dmTNnDsOGDct3LiIigp49exIaGgpAxYoVmT17Ntu2bctXOP7ll1/4+OOP2bFjB56envnmaNu2LW3btrUeBwQE8M477zBkyBBr4fj8+fMMHjyYlStX0qpVK2ts7dq1b80izxy0tKPY/RnkZEHNRyH4NShX99bMfwcxTZNfY8+xYmc8q3ef5HxaJm7ODnRpVIFO9XyoW6HUdV9yJyJyt1LhWEREREREROQWe+GFF3jiiScIDQ29qnDcrFkzVq1axfPPP0+FChWIiIggKiqKAQMGWGMuXrxIt27dmDFjRr6Xrv2V5ORkSpcubT1eu3Yt2dnZnDp1iho1anDhwgXuv/9+xo0bR0BAwD9f3PGfLS0p9n8DdsWgfi9o8jKU8f/nc94BVu6MZ8yaA5w4n065Uk4MaBNILZ+SfBUVz8qoeI6fTaeYvQ2ta3jxaD0fmlVxx95WHUBF5N6lwrGIiIiIiIjILTRz5kx+//13Fi1adM3zkyZN4sUXX8TX1xc7O8u/lk+ePJn27dtbY1566SXatm1Lu3btbuiex44dY+zYsbz99tvWscOHD5OTk8OIESOYOHEiZcqU4f333yc0NJR9+/ZRvPhN9BzOyYGYNZaCcexWcCoNzQfC/S+As/uNz3OHWrkznsFf7iE9MxuA+PPpvPF5FKYJNgY0rezOGy2r0rqmFy6OKqWISNGgv+1EREREREREbpEDBw7w9ttvs2XLFuzt7a8ZM3nyZCIiIvj666/x8/Nj06ZN9O/fn4oVK9K2bVsWLlzIrl272L59+w3d89SpU7Rt25ZWrVrxxhtvWMdzcnLIzMxk0qRJtG7dGoBPP/0ULy8vVq1axVNPPZV/ot2fw/r3aX4hDnaWhxbvQY1OsGeZpSXFmf1Q0hfajYZ6PcDh3ujleykzmxHf7LUWja8wTSjpZMe6N5pTtkSxQspORKTwqHAsIiIiIiIicots3bqVxMREatasaR3Lzs5m06ZNTJs2jaSkJAYPHsyyZcvo0KEDAHXq1CEqKoqxY8fStm1b1q9fz969e3Fxcck391NPPUWTJk3YsmWLdSwhIYGHHnqIWrVqsXDhwnx9dr29vQGoUaOGdaxkyZKUK1eO2NjY/Inv/hxWvQaZ6RgAF47Dyr7wzQDIOA+eteGxWZY+xrZ3dykhMzuH3XHnifg9ia2Hk9h+7ByXs3KuGZucnqWisYgUWQX2t71hGGHA0D8NnzJN0yv3vJF7/gWgNLANeNk0zeg8c5QGJgH/yR36GnjVNM3zeWJqA1OA+4GzwHRguGmaZp6Yx4HhQCXgEPCOaZorbtliRUREREREpEjq1KkTDRs2zDfWu3dvqlSpYm0jkZmZia2tbb4YW1tbcnIsxcuRI0fSv3//fOdr167N2LFj6dixo3Xs5MmThIaGUrNmTZYsWWJte3FF06ZNAcsu6PLlywOQkpLCyZMn8fPzy5/4+vchMz3/WE4mZKVDjy+h0kNwl778LTvHZN/JZCIOJRJxKImfj5wl7bJld3EN7xI8HeTHip3xJKVevuracqWcCjpdEZE7RkH/Z8IDQEie47z/H8hA4E3gmdy494B1hmEEmqZ5MTdmMeALXHl17CxgIdABwDCMEsA6YBPQCKgGzAVSgXG5MU2Az7AUqb8EHgOWGYbR1DTNbbduqSIiIiIiIlLUlCpVilKlSuUbc3Z2pkyZMtSqVQuA5s2bM2jQIFxcXPDz82Pjxo0sWLCA0aNHA+Dj44OPj89Vc1eoUMH6UrsTJ04QEhJCuXLlmDhxIomJidY4Dw8PbG1tqVq1Kh07duR///sf06dPp3Tp0gwdOpSyZcvm66cMwIW4ay8o+zJUbvFPvx2FwjRNYk6nEPG7pVC87chZLqRnAlC5rAtPNChPcCU3Gvu7UdrZAYBaPiXz9TgGcLK3ZUCbwEJZg4jInaCgC8dZpmkm/Hkwd7fx68Ao0zS/yB3rBZwGugHTDcOojqVg3Mw0za25MS8Cm3OLyweA7kBxoJdpmunAb4ZhVAP6GYYxPnfX8evAj6Zpjsy9/UjDMEJzx7vevqWLiIiIiIiIwNKlSxk8eDDdu3fn7Nmz+Pn5MXz4cF555ZUbnmPt2rXExMQQExODr69vvnNHjhyhYsWKACxcuJB+/frRoUMHTNOkWbNmrF+/3vJivKwMiF4J26YB5tU3AShZ/h+usuCYpsmxpDS2Hk4i4lASWw8lkphi2T3sW6Y4bWt6EVzZjSYBbtdtO9GpnqVQP2bNAU6cT6dcKScGtAm0jouIFEUFXTgOMAzjBJCBpRXF26ZpHgb8AS9g7ZVA0zTTDcPYBARjaTfRBEgBIvLM9xOW3cTBWHYpNwE25xaNr1iDpS1FReBIbszkP+W1Brjxf0KLiIiIiIiI3KDw8PB8x15eXsydO/em5sjTfRGAZ555hmeeeeZvr3N1dWXmzJnMnDnzj8GLCfDjh7B9DqSeBveqcF9XSxE5K8+/Tts7WV6Qdwc6eSGdiN//KBSfuHAJAM8SjjxQxYMmlSyF4gplit/wnJ3q+ahQLCKSh/Hnf/jcthsZRjvAFdgPlAXexdJKoiYQiKUI7GeaZmyea+YAPqZptjEM423gedM0A/4072FgpmmaHxqGsRaIM03z2TznfYFjQLBpmlsNw7icO8+CPDFP587heJ3cX8DSexlPT88GS5cu/bffjrtKSkrKVS9lkKJDz79o0/MX/QwUbXr+RZuef9Gm539vck0+QPm4b/A48xOGmU2SW0PifdpzrvR9YBiUPbWRgMMLccw4Q4ajB4cDenLas3lhpw1AcobJvrPZ7EvKZt/ZbE6lWWoZrvZQzc2W6mUsHy9nI98LAuXm6fe/aNPzL3pCQ0N3mKbZ8FrnCmzHsWma3+U9NgwjEjgM9AIiCyqPf8I0zRnADICGDRuaISEhhZtQAQsPD6eorVn+oOdftOn5i34GijY9/6JNz79o0/O/h2Rdhr257Sjid4BjCWj8AjR6Hne3SrjnCw4Bhlqffw2gRmHkDFxIz2SbtfVEEgdOWV595OpoR+MAD16o5E5wJTcCPV2xsVGh+FbS73/RpucveRV0qwor0zRTDMOIBqoAK3OHPYHYPGGewJWeyAmAh2EYRm6v4iu9kcv+KcbzT7fyzHPur2Ku6r0sIiIiIiIicle6eAp2zLW0o0g5BW5V4OGxcF8XcHQt7OyukpqRxS9Hz7L1kKVYHH3iAjkmFLO3oVHFMnSq50OTSm7UKlcCO1ubwk5XRKRIKLTCsWEYxbC0qvgRS+/hBKAV8Eue8w8AA3Iv2Qq4YOlRfKXPcRPAOc/xVuAjwzCKmaZ5KXesFXACOJonphUwJk86rcjfO1lERERERETk7hO/A7ZNh9++hJxMqNIGGr8IAaFgc+cUXC9lZrMz9jxbDyUScSiJqOPnycoxsbc1qOdbmtdaVCG4kjv3VSiJo51tYacrIlIkFVjh2DCMscAqLDuKywJDsBR955umaRqGMRF42zCM/cBBLD2QU4DFAKZp7jMM43tgem7PYbC8NG+1aZoHco8XA0OBeYZhjACqAoOAYeYfzZw/BjYZhjEIy07nR4FQoNntW72IiIiIiIjIbZJ1GfZ+lduOYjs4uEKj5+H+PuBWqbCzAyAzO4fdcResheIdx86RkZWDjQF1ypfihQcDaFLJjYZ+ZXByUKFYROROUJA7jssDSwB34AyWvsZBpmkeyz0/GnACpgKlgW1Aa9M0L+aZoxswGViTe/w18MqVk6ZpXjAMo1XuHNuBc8A4YHyemAjDMLoAI4D3gUPAU6ZpbrulqxURERERERG5nVJOw/a5sH12bjuKytBuDNTtWujtKLJzTPadTCbiUCJbDyXx85GzpF7OBqC6dwl6BPkRXMmNRv5lKFHMvlBzFRGRayvIl+N1+ZvzJhCW+/l/9u48rKpq/+P4e4OCCMogs+KMOIJjitNPzAGnpLqlaU5d0wavlabXtFIrh9SyucRKzSwzK03NKcdMLCeC64BzKgKKCgoiImf//jh4gkSFAnP4vJ7nPJ2919prffc+PQVf1vmua/U5Czx6g3FigdY36LMQWHi9PiIiIiIiIiK3pPjt8Esk7PoWsi9BYAe4ZzBUa/uPlaMwTZMDJ9PYfPA0mw8ms+XQGVIzsgCo5uXMAw0r0LxaOZpWLYeHs8M/EqOIiBTOP1bjWEREREREREQK6PIl2PO9tRzF8a3WchSNH4Mmj4Nn9ZsejmmaHD1zISdRfJqog6dJTssEoIK7Ex3r+NC8mieh1crhU7bUTY9PRET+PiWORURERERERG5VaSdh+2zY+gmkJYJHNeg0BUIegVJli3SqRTvjmboyjviUDMpvWcuIjkFENChva09IzSAqV6I4PiUDAO8yjrSsXs6WKA7wKF2kcYmIyD9DiWMRERERERGRW82JnfDLDPjfN9ZyFNXbQdP3oNq9xVKOYtHOeF74NpaMLGsd4viUDEZ9G8POo2fJsphEHTzNzkWRpP78ZZ7rPL28+SUpEcMwME2T8ePHExkZydmzZ2natCnvv/8+derUAeDIkSO8+uqrrFu3joSEBPz8/OjRowcvv/wyTk5OtjHXrFnDSy+9RGxsLM7OzvTr148JEyZQosQfKYzY2FiGDBnCr7/+ioeHB4MHD+all17CMIwifzYiIncrJY5FREREREREbgXZWbB7MfwaCcd+AQcXaDQA7nkcPAOLdeqpK+NsSeMrLmZZmBP1Oy6OJWhaxYMygV78LzGQTRs3YGdnTdDa29vbkrVTpkzhjTfeYPbs2QQFBfHKK6/Qvn174uLiKFOmDHv37iU7O5sPP/yQwMBA9uzZw6BBgzh9+jSRkZEA/Pbbb3Tu3JlRo0bx2WefER8fzxNPPEF2djbTpk0D4Ny5c7Rv357WrVuzdetW9u7dy4ABA3B2dmb48OHF+pxERO4mShyLiIiIiIiI/JPSTlnLUWz7BM4ngEdVCH8d6vcq8nIUf2aaJrtOnLOVnfgzA4h+uT0l7O0Yd2QZh7Y64O/vl+84b731FqNGjeLBBx8EYM6cOXh7e/PFF18wePBgwsPDCQ8Pt11TtWpVxowZw0svvWRLHH/11VfUrl2b8ePHA1C9enWmTJnCww8/zNixYylTpgzz5s3jwoULzJkzBycnJ+rWrcvevXt58803GTZsmFYdi4gUkQJ/v8UwjLXXexVnkCIiIiIiInJ7mjRpEoZhMGTIENu5/v37YxhGnlezZs3yXPf4449TrVo1nJyc8PLyonv37uzZsydPn8qVK181zqhRo/L0eeaZZ2jcuDGlSpWicuXKxXaff8mJnfDdkzC9Nqx7DbxrQ6+vYch2aPZEsSaNj56+wLtr9tPuzQ10fXfTNfv5uzlRwv6P1MGhQ4fw9/enSpUq9OzZk0OHDgFw+PBhEhMT6dChg62vk5MTrVu3ZvPmzdcc/9y5c7i7u9uOMzMzKVUq72Z6Tk5OXLx4ke3btwMQFRVFq1at8pS36NixIydOnODIkSMFewAiInJDhVlx3AaYBVzIOR4EfAWkFnFMIiIiIiIicgfYsmULkZGRBAcHX9XWrl075s6dazt2cHDI0964cWP69u1LQEAAZ86cYdy4cbRr144jR45QsmRJW7+XX36ZJ5980nbs4uKSZxyLxUK/fv2IjY1l1apVRXVrf112FuxZYq1ffGwLlHSGhv3gnkHgVaNYpz6TfollMSdYFH2C7b+fBeCeKh78u2VVwOTVpXvylKtwKmnPiI5BtuOmTZsye/ZsatasycmTJ3nttddo3rw5u3btIjExEQAfH588c/r4+BAfH59vPL///jvTpk1j9OjRtnMdO3Zk+vTpfP755/Ts2ZOkpCReeeUVABISEgBITEykQoUKV81zpa1KlSp/5fGIiMifFLZUxQumaZ4EMAyjPzDONM1DRR6ViIiIiIiI3NZSU1Pp3bs3n376qa3sQG6Ojo74+vpe8/rBgwfb3leuXJnXXnuNkJAQDh06RFDQH8nMMmXKXHecd999F4Bp06b9s4nj9GTYPgu2fgrnT4B7Feg4CRr0hlKuxTZtxqVsVu1OZHH0CTbuO8Vli0mQTxlGhgdxX4g/FdxL2/qWdijB1JVxxKdkUN7NiREdg4hoUN7W3qlTpzxjN2vWjKpVqzJnzpyrVozfSFJSEuHh4bRv357nnnvOdr5Dhw5MmzaNp59+mv79++Po6MhLL73ETz/9hF0xbAooIiLXVpj/6l4ESgEY1oJBDsDzhmGoTrKIiIiIiIjkMWjQIP71r38RFhaWb/umTZvw9vamRo0aPP7445w8efKaY6WnpzNr1iwqVqx4VbmJadOmUa5cOerXr8+ECRO4dOlSUd7G35fwGyx6Ct6sDWtfA++a0GsB/GcHhD5VLEnjy9kWNuw7xbCvomn82mqemR/NnoRz/LtVFZY/04qVz7XmqTbV8ySNASIalOfnUW2ZHe7Mz6Pa5kka58fFxYU6deqwf/9+W/I+KSkpT5+kpKSrEvuJiYmEhYVRt25d5s6de1VN4mHDhpGSksLRo0dJTk6me/fugLUmMoCvr2++81xpExGRolGYpG8c1kTxFOAR4BxQC9hgGMaDpmkmFkeAIiIiIiIicnuZOXMmBw4c4PPPP8+3PTw8nAceeIAqVapw5MgRXnzxRdq2bcv27dtxdHS09fvggw8YOXIk6enpBAUFsWbNmjztQ4cOpUGDBpQrV45ff/2VUaNGcfjwYT7++ONiv8frulKO4tdIOBqVU46iT045iqAbX/8XmKZJzPFUFkXHs+S3BJLTMilTqgTdQvzpXr88Tat4YGdXtJvGXbx4kb179xIWFkaVKlXw9fVl9erVNGnSxNb+008/MXXqVNs1CQkJhIWFUadOHb788ktKlMg/LWEYBv7+/gB8+eWXBAQE0LBhQwBCQ0P573//y8WLF231kFevXo2/v/+tV8daROQ2VpjE8UvAfOBp4DLwBPAZMBXYAfgXeXQiIiIiIiJyW4mLi2P06NFs2rQpTy3i3Hr27Gl7X69ePRo1akSlSpVYtmwZDzzwgK2td+/etG/fnoSEBKZNm8ZDDz3Ezz//TOnS1pWyw4YNs/UNDg6mbNmy9OjRg9dff51y5coV0x1eR3oybJ8NWz/JKUdR2VqOon4vcHIrlimPJKezKDqexdEnOJycjoO9HW1rehPRwJ82Qd6UKnUsNBsAACAASURBVGlfZHM9//zzdOvWjYoVK3Ly5EleffVV0tPT6devH4Zh8OyzzzJx4kRq1qxJjRo1eO2113BxcaFXr14AnDhxgjZt2uDv789bb71FcnKybWwvLy/s7a2xTp06lfDwcOzs7Pj222+ZPHkyCxYssLX36tWL8ePH079/f1588UX27dvH5MmTGTt27FWrl0VE5K8rcOLYNM2lhmGUB2oAx3KtMB5mGMaWYolOREREREREbitRUVEkJydTp04d27ns7Gw2btzIRx99RHp6ep5VwwD+/v5UqFCB/fv35znv6uqKq6srgYGBNGvWDHd3d7755hv69OmT79xNmzYF4MCBAzc3cZzwG/wSCbFfQ3YmVA2DrtMhsD3YFV3i9orktEyW/mbd5C76WAqGAU2rePDE/1UlvK4frk75J+z/ruPHj/PII4+QnJyMl5cXzZo1Y8uWLVSqVAmAkSNHkpGRwdNPP83Zs2dp2rQpq1atokyZMgCsWrWK/fv3s3//fipWrJhn7MOHD9tWCy9fvpwJEyaQmZlJSEgIixcvzlNf2dXVldWrV/P000/TuHFj3N3dGT58eJ4/JIiIyN9XqPrEpmmmAlvzOb+gyCISERERERGR21ZERASNGzfOc27AgAEEBgYyevRoHBwcrromOTmZ+Ph4/Pz8rjmuaZqYpklmZuY1+0RHRwNcd5wik30Z9i6FX2bA0c1QsjQ0eNRajsK7ZpFPl555mdW7k/huZzybDiSTbTGp5VeWFzrV5L76/vi5OhX5nH82f/7867YbhsG4ceMYN25cvu39+/enf//+N5xn7dq1N+xTr149Nm7ceMN+IiLy1xU4cWwYRsPrtZumuePvhyMiIiIiIiK3Mzc3N9zc8pZlcHZ2xsPDg7p165KWlsa4ceN48MEH8fPz48iRI7zwwgt4e3tz//33A9YVw9988w3t2rXDy8uL48ePM3nyZBwdHenatStgXdm8ZcsWwsLCcHV1ZevWrTz33HPcd999eVazHjhwgLS0NE6cOMGlS5dsyeXatWvnm8S+ofTTsGO2tRzFuXhwqwQdJliTxkVcjiIr28Km/cksio5n1a4kMrKyKe/mxKDWVYmoX54g3zJFOp+IiEhuhVlxvA0wc97/uWiQCRT9929ERERERETkjmJvb09sbCyfffYZKSkp+Pn5ERYWxoIFC2wlDRwdHVm/fj1vvPEGKSkp+Pj40Lp1a6KiovD19bX1+eqrrxg/fjyZmZlUqlSJxx9/nJEjR+aZb+DAgWzYsMF23KBBAyBvaYQCSYiBX2dAzJVyFG2g8zSo0bFIy1GYpsnOYyks3hnP0pgETqdfwtWpJPc3LE9E/fI0ruRe5JvciYiI5KcwieNNQH1gMvAFfySRRURERERERK5p/fr1tvdOTk6sXLnyuv0DAgJYvnz5dfs0bNiQLVtuvN1O7rlvKGYBrHkFUo+DawUIexEcnKzlKH7/OaccRe+cchS1Cj5uARw8lcbinfEs/u0Ev5++gGMJO9rV8qF7fesmdw4l7Ip0PhERkRspzOZ4rQ3DeACYBEQAz5mm+XOxRSYiIiIiIiJys8QsgCVDISvDepx6DBY9AZjgVhE6vJZTjsK9yKY8ef4iS35LYHF0PDHHUzEMaF6tHEPCqhNe15cypYpnkzsREZGCKOzmeN8ahvE98DSwyDCMn4CRpmkeKJboRERERERERG6GNa/8kTS2MaG0JwyNLrJyFGmZl1n5v0QWRcfz84FkLCbULV+WF7vUoluIPz5lSxXJPCIiIn9XoRLHAKZpXgbeNgxjNjAG2G4YxizTNJ8t6uBEREREREREilVGinW1ceqx/NsvnP7bSeOsbAsb953iu53x/LgniYtZFiq4O/FUm+pENPCnurc2uRMRkVtPgRPHhmGcJ/+6xqWA/wBKHIuIiIiIiMitzzTh2K+wfTbs+g4uZ4BdSbBkXd3XtcJfnMJk++9nWRQdz7KYBM5eyMK9dEkeahRARAN/GlZ0xzC0yZ2IiNy6CrPi+D9oQzwRERERERG5XWWchd++siaMT+0BBxcI6QmN+kHy/rw1jgFKOsG9LxdqigMnz7No5wkWRcdz/GwGpUra0b62LxH1/Wldw4uS9trkTkREbg+F2RxvdjHGISIiIiIiIlL0TBOObrEmi3cvgssXwb8hdHsH6j4Iji7Wfv4NrP9c8wqkHreuNL73ZQh++IZTJJ27yPfR1mTxrhPnsDOgRXVPnmtXg451fXFxLHSVSBERkX9cYUpVVLxeu2maR/9+OCIiIiIiIiJF4MIZ+G2+NWGcHAcOZaB+b+vqYr+Q/K8JfrhAiWKAcxezWPG/RBZHx7P54GlME0IquPJy19p0DfHDu4w2uRMRkdtbYf7seYT8S1UYOeeLZotZERERERERkb/CNOH3zTmrixdDdiaUbwz3vQd1HwAH5781/KXLFtbHnWRRdDw/7jnJpcsWKpUrzX/aBhJR35+qXi5Fcx8iIiK3gMIkjpvk/NMANgC9gONFHpGIiIiIiIhIYaSfht++hB1zIHkfOJaFhn2tq4t96xV4mEU745m6Mo4TKRn4uzkxomMQ94X4s+33s3y3M54fYhNIzciinLMDjzQJIKJBeeoHuGmTOxERuSMVpsbx9ivvDcOwALGmaR4qlqhERERERERErsc04cgm6+riPd9D9iWocA90/wDqRBR6dfGinfG88G0sGVnZAMSnZPD8178x7vtdpGRk4VTSno51fOjeoDwtq3tqkzsREbnj6f90IiIiIiIixWjSpEkYhsGQIUMAyMrK4r///S/BwcE4Ozvj5+dHr169OHo0/21jTNOkU6dOGIbBwoUL87Tt27ePiIgIPD09KVOmDM2aNWPFihV5+hw9epRu3brh7OyMp6cnQ4cO5dKlS8VzszdDejL8/A681xjmdIX9q6HRAHhyMwxcDQ16/6WSFFNXxtmSxldctphkZGXzVo/6bHuxHW/1bEBYkLeSxiIiclf4O1u75lfvWERERERERHJs2bKFyMhIgoODbecuXLjAjh07GDNmDPXr1yc1NZXhw4cTHh5OTEwMJUrk/TXtjTfewM4u/0Rl165dqVq1KmvWrMHZ2ZmPPvqI7t27s3v3bqpVq0Z2djZdunShXLly/PTTT5w+fZp+/fphmibvvvtusd57kTJNOLwxZ3XxErBkQUAzaPU81O4ODqX/8tC/n05nWWwC8SkZ+bZfumwhokH5vzy+iIjI7arAiWPDMGL5I1nsBCw3DMP2Z2rTNIPzvVBEREREROQulJqaSu/evfn0008ZP3687byrqyurV6/O03fGjBnUqVOHPXv2UK/eHzV5t27dyttvv8327dvx8fHJc01ycjL79+9nxowZhISEADB58mSmT5/Ozp07qVatGqtWrWLXrl38/vvvBAQEADBlyhQGDhzIhAkTKFu2bHHdftFIOwW/fWFNGJ85BKVcoclAa+1i71p/edgryeJlMQnsOnEOgJL2BlnZV6+P8ndz+svziIiI3M4Ks+I493eivinqQERERERERO4kgwYN4l//+hdhYWF5Esf5OXfOmrx0d3e3nTt//jy9evUiMjISb2/vq64pV64ctWrVYu7cuTRp0gQnJyciIyMpU6YMLVq0ACAqKopatWrZksYAHTt2JDMzk+3btxMWFlYUt1q0LBY4cmV18VLr6uKKofB/o6D2fVDyryVyjyRbk8U/xP6RLK4f4MaYzrXoVM+XbUfO5qlxDOBU0p4RHYOK4q5ERERuO4XZHO/6P+mIiIiIiIgIADNnzuTAgQN8/vnnN+x76dIlhg8fTrdu3ahQoYLt/BNPPEF4eDidOnXK9zrDMFi9ejX3338/ZcuWxc7ODg8PD5YvX46fnx8AiYmJV61U9vT0xN7ensTExL9xh8Ug7SREz4Ptc+DsYXByh3sGQcO+4F3zLw15JVm8LCaB3QnWZHGDim682KUWner5UT7XauIK7tZyF1NXxnEiJQN/NydGdAy6bpmKSZMmMXr0aJ5++mnee+892/l9+/YxatQo1q5dy6VLl6hZsybz5s2jVi3rKuk2bdqwYcOGPGP16NGD+fPn5xlj5MiRbNq0iczMTOrUqcO4ceMIDw/Pc93nn3/OtGnT2Lt3Ly4uLnTu3JnPPvvsLz0vERGR3ApV49gwjFJAV6AaMMM0zRTDMKoBZ03TPFMcAYqIiIiIiNxO4uLiGD16NJs2baJkyZLX7Xv58mUeffRRUlJS+P77723n586dy2+//ca2bduuea1pmjz11FO2+sVOTk58/PHHPPjgg2zdupXy5W+DurwWCxxeb11dvHcZWC5DpRYQNgZqdYOSpQo95OHkdH4oYLL4zyIalC9wPeP86lcDHD58mBYtWtC3b1/Wrl2Lm5ubLamb24ABA5g4caLt2Mkpb1w3ql8N8M477zBp0iSmTp1Ks2bNyMjIYN++fQWKX0RE5EYKU+O4OvAj4AK4AV8DKcCTOccDiyNAERERERGR20lUVBTJycnUqVPHdi47O5uNGzfy0UcfkZ6ejqOjI5cvX+aRRx4hNjaW9evXU65cOVv/NWvWsHv37quSjT169CA0NJRNmzaxdu1alixZwpkzZ3BzcwPggw8+YPXq1cyaNYsXX3wRX19ffv755zxjJCcnk52dja+vbzE+hRs4nwTRn1tXF6f8Dk4e0PQJaNgPvGoUergryeKlMQnsyUkWN8xJFneu51fkdYqvVb8aYMyYMXTo0IE33njDdq5q1apXjVG6dOlrfgYFqV+dkpLCCy+8wKJFi2jfvr3t2tw1skVERP6O/Lfmzd9bwCrAB8i93ez3wC1YGEtEREREROTmi4iIIDY2lujoaNurcePG9OzZk+joaBwcHMjKyqJHjx7ExMSwbt26qxKIEyZMICYmJs8YANOmTbOVIbhw4QIAdnZ5f62zs7PDYrEAEBoayp49ezh+/LitffXq1Tg6OtKoUaNiewb5sljgwI/w1aMwvTaseQXcKsKDn8CwPdBxQqGSxodOpfHe2v10evsnwqatZ+rKOJxK2vFil1psHtWWb59qwcBWVYtlc7vc9atzs1gsLFmyhNq1axMeHo6XlxdNmjThq6++umqM+fPn4+npSZ06dXj++ec5f/68rS13/eq0tDSys7Ovql+9atUqsrOzSUpKonbt2pQvX57777+fQ4cOFfn9iojI3akwpSqaA81M08w2DCP3+aOAf5FGJSIiIiIicptyc3OzrQC+wtnZGQ8PD+rWrcvly5d56KGH2Lp1K0uWLMEwDFu9YVdXV5ycnChfvny+pSYCAgJsq1dDQ0Px8PBgwIABvPzyyzg5OTFz5kwOHTpE165dAejQoQN16tShb9++vPHGG5w+fZoRI0bw+OOPU7Zs2WJ+EjnOJVhXF+/4DFKOQuly0OxJ6+piz8BCDXXoVJptZfHeRGuitVEld17qWptOdX2LJUn8Z9erX33y5EnS0tKYOHEir776KpMnT2bt2rX07t0bFxcXunTpAkCvXr2oVKkS/v7+7Nq1ixdeeIGYmBhWrVoFFKx+9aFDh7BYLLz22mu89dZbeHh48MorrxAWFsaePXsoXbp0sT8LERG5sxWqxjGQX4GuikBqEcQiIiIiIiJyxzt+/DiLFy8GuGrV76xZs+jfv3+BxvH09GTFihWMGTOGtm3bkpWVRa1atVi0aBENGzYEwN7enmXLlvHUU0/RokULnJyc6N27N1OnTi3Se7qKJRsOrrXWLo5bDmY2VGkN7cZBza5QwrHAQx08lcYPMQksi/3nksVX3Kh+9ZWV3t27d2fYsGEA1K9fn23btvHee+/ZEseDBg2yXVOvXj2qVq1K06ZN2bFjBw0bNixQ/WqLxUJWVhbvvPMOHTp0AGDevHn4+vqyZMkSevToUdyPQ0RE7nCFSRyvAoYB/845Ng3DKAuMB5YVdWAiIiIiIiJ3ivXr19veV65cGdM0Cz1Gftc0btyYlStXXve6ihUrsnTp0kLP95ecOwE7c1YXpx6D0p7QfIh1dXG5agUeJr9kceNK7rzctTad6vni53rzksW5FaR+dYkSJahdu3ae62rVqsX8+fOvOW7jxo2xt7dn//79NGzYsED1q6+sPM49l6urK/7+/hw9erQob1tERO5ShUkcDwPWGYYRB5QCvgKqA0nAw8UQm4iIiIiIiNzqLNnW2sXbZ8O+FWBaoGob6PAqBHWBEg4FGubASWsZih9usWRxbhERETRu3DjPuQEDBhAYGMjo0aNxcHCgSZMmxMXF5emzb98+KlWqdM1xY2Njyc7OtiWDC1K/+kqt47i4OCpUqABAWloaCQkJ151LRESkoAqcODZN84RhGPWBR4CGWDfWiwTmmaaZcd2LRURERERE5M6SejxndfFcOHccnL2gxTPQsC94VC3QEFeSxctiEohLsiaLm1R2Z2y32nSq64eva6nivINCu1H9aoCRI0fy8MMP06pVK9q2bcu6deuYP38+ixYtAuDgwYPMmzePzp074+npye7duxk+fDgNGjSwJYMLUr+6Ro0adO/enWeeeYYZM2bg7u7O2LFj8fb2tvURERH5OwpV4zgnQfxpzsvGMAw70zQtRRmYiIiIiIiI3GKyL8OB1dbVxftXWVcXV2sL4ROhRqcCrS4+cPI8y2IS+SHWmiw2DOvK4ls1WVxYERERREZGMnHiRJ555hkCAwP57LPPbPWNHRwcWLNmDW+//TZpaWkEBATQpUsXxo4di729PVCw+tUAc+fOZdiwYXTr1g3TNGnZsiVr1qzRxngiIlIkCpw4NgzjAdM0v83nfCPgY6BBUQYmIiIiIiIit4iUY7BzrnV18fkT4OIDLZ+DBn3Ao8oNL7+SLF4We4J9SWkYBjSp5MG4brXpVM8Pn7K3b7I4d/3qK/r373/NTQ4DAgLYsGHDDcctSP3qMmXKMHPmTGbOnFmQUEVERAqlMCuOPzUMw8c0zQ8BDMNwAiYAg4Fi3pJXREREREREil3MAljzCv+Xehx2loda98HpA7B/tbW9+r3QeQrUCAf7ktcdan/SeZbl1Cy+05LFIiIid4PCJI7vBZYahlEe+An4EDgBNDFNc3dxBCciIiIiIiI3ScwCWDIUsjIwwFrDeMsH4FgWWg231i52v/6ma1eSxctiEth/MidZXNmD8ffVIbyur5LFIiIit5HCbI633TCMFsAK4AXgOeBd0zTN4gpOREREREREbgLThJWjISuffc9LlYV7X7rmpfuSzrMsxrqy+M/J4k51ffFWslhEROS2VNjN8Q4ZhhEKLAU6Ya1tfKE4AhMREREREZFidvEc/DYfts6E9FP590mNv+rUlWTxstgEDuRKFr/SvQ7hdZQsFhERuRMUZnO888CV1cUlAEcg2TCMywCmaZYt+vBERERERESkyJ2Kg19nwm9fwqU08G8ATh6Qcebqvq4VME2TfUlptprFV5LF91T2oK+SxSIiInekwqw4/g9/JI77AfcALwJnizooERERERERKWLZl2HfCvg1Eg5vAHsHqPMA3DMIKjRi6/czqLv9RZyMS7ZLMkwH5jk8ypdvbuDgqXRbsrhf9zp0rOuLdxkli0VERO5UhalxPBvAMIzxwP8BXUzTXFFMcYmIiIiIiEhRSE+GHZ/Btk8h9RiULQ9tX4KG/cDFy9bt2d2BNMoayMgSC/A3TnPCLMeUyw/z/bF6NKvqSP/mlZUsFhERuYvYFbSjYRh2hmHMBHoDy4HZhmH0LLbIRERERETktvX+++8THBxM2bJlKVu2LKGhoSxbtszWnpSURP/+/fH396d06dKEh4ezf//+PGO0adMGwzDyvHr2zPsryNmzZ+nTpw+urq64urrSp08fUlJS8vRZuXIloaGhlClTBk9PT7p3786+ffuK7+ZvFfE74Lsn4c3asGY8eFSBHp/DMzHQ+vk8SeMjyenEp2TwvaUlLS+9Q9XMebS89A7fW1piAPMHhdIntLKSxiIiIneRwpSq+B7wBpqbpnkyJ2n8lmEYfYAnTNM8ViwRioiIiIjIbadChQq8/vrrBAYGYrFYmDNnDhEREWzfvp169eoRERGBnZ0dixYtwtXVlTfffJN27dqxe/dunJ2dbeMMGDCAiRMn2o6dnJzyzNOrVy+OHj3KihXWL0MOHDiQPn36sGTJEgAOHz5M9+7dGTp0KHPnziUtLY2RI0fSuXNnDhw4cBOexE12ORN2LbKWo4jfBiWdoWEfaDIQvGvl6XoiJYNlMQksiTlBzPHUaw7p7+Z0zTYRERG5cxUmcWwPtDFN8wKAaZrzDcNYCbwJ7AK0OZ6IiIiIiADQvXv3PMcTJkzgww8/JCoqilKlSrFlyxaio6MJCQkB4MMPP8TX15cvv/ySgQMH2q4rXbo0vr6++c6xZ88eVqxYwaZNmwgNDQVgxowZtGrViri4OIKCgti+fTtZWVlMmjQJe3t7AF544QXatm1LcnIynp6exXH7N1/qcWspiu1z4EIylKsOnaZASE8o5WrrlpyWyQ+xCSz57QRbj1i3qwmu4MqYzrUoYW8wZUUcGVnZtv5OJe0Z0THopt+OiIiI/PMKkzjuZprm5dwnTNM8CwwwDGNu0YYlIiIiIiJ3iuzsbL7++mvS0tJo3rw5mZmZAJQq9UfZAzs7OxwdHdm0aVOexPH8+fOZP38+Pj4+dOrUibFjx1KmTBkAoqKicHFxoXnz5rb+LVq0wNnZmc2bNxMUFESTJk0oWbIkH3/8MQMHDuTChQvMmTOHJk2a3P5JY9OEIz9ZVxfv/QFMCwR1gnsehyptwM5amTD1QhYrdyWyJOYEPx9IxmJCoLcLw9vXoFuIP5U9/1jh7V7agakr44hPyaC8mxMjOgYR0aD8P3SDIiIi8k8qzOZ4l6/TtrZowhERERERkTtFbGwsoaGhXLx4ERcXF7777jvq1atHVlYWFStWZPTo0cycORMXFxemT5/O8ePHSUhIsF3fq1cvKlWqhL+/P7t27eKFF14gJiaGVatWAZCYmIiXlxeGYdiuMQwDb29vEhMTAahUqRKrV6/moYce4umnn8ZisdCgQQOWL19+cx9GUcpMg5j58OtMOLUXnNyh+RBo/G9wrwRAeuZlftxjXVm8Yd8psrJNKpUrzVNtqtMtxJ8g3zL5Dh3RoDwRDcqzfv162rRpcxNvSkRERG41hVlxjGEYYcAjQEXAIXebaZptizAuERERERG5zQUFBREdHU1qaioLFy6kX79+rF+/nrp16/Ltt9/y73//m3LlymFvb0+7du3o1KkTpmnarh80aJDtfb169ahatSpNmzZlx44dNGzYsEAxJCYm8u9//5u+ffvyyCOPcP78eV5++WUefvhh1q5di51dgfcL/+cl74etH0P0F5B5DvxCoPv7UPdBKOnExaxs1v/PurJ4zZ4kLmZZ8HMtRf/mlekW4k+98q55kuwiIiIi11PgxLFhGP2Bj4DvgDbAYqAGUAX4vBhiExERERGR25iDgwPVq1cHoFGjRmzdupXp06fzySef0KhRI1tS+dKlS3h5edG0aVMaN258zfEaN26Mvb09+/fvp2HDhvj6+nLq1ClM07QlRE3T5OTJk7a6yO+//z7Ozs5MmTLFNs7nn39OQEAAmzdvpmXLlsX4BIqAJRv2r7KWozi4FuxKQp374Z5BUKExWRaTTQeSWfJbHKt2JZGWeZlyzg481CiAbiH+NK7kjp2dksUiIiJSeIX58/rzwBDTNB8BsoAXTNNsgDVpnFYcwYmIiIiIyJ3DYrHY6htf4erqipeXF/v372fbtm1XbaqXW2xsLNnZ2fj5+QEQGhpKWloaUVFRtj5RUVGkp6fb6h5fuHDBtineFVeOLRZLkdxXsbhwBn5+G96pD1/2hJN7IexFGLab7PsjibpUlRe++x/3TPiRAbO2snp3Ep3r+TL33/fwy+h7eTWiLvdU8Sj2pPGkSZMwDIMhQ4bYzr300kvUrFkTZ2dn3N3duffee9m8eXOe6yIjIwkLC8PNzQ3DMDhy5Mg157h48SIhISEYhsG2bdts52fPno1hGPm+tm7dWuT3KiIicrcpTKmKqsCPOe8zAZec9+8B64FRRReWiIiIiIjczkaNGkWXLl0ICAjg/PnzfPHFF6xfv55ly5YB8PXXX+Pp6UmlSpWIjY3lmWeeISIigg4dOgBw8OBB5s2bR+fOnfH09GT37t0MHz6cBg0a0KJFCwBq1apFeHg4gwcPJjIyEoDBgwfTtWtXgoKCAOjSpQvTp0/nlVdesZWqGD16NAEBATRq1OgfeDI3cCLaWrv4fwvh8kWo1BLav4oZ1JmdJ9JZsu4Ey2JiOHk+k9IO9rSv7UO3YH9a1fDEsYT9jccvQlu2bCEyMpLg4OA854OCgnj//fepUqUKGRkZTJ8+nfDwcPbv34+Pjw9gTeh36NCB7t2789xzz113nueff54KFSoQExOT53yPHj0IDw/Pc27EiBFs3rz5uivXRUREpGAKkzg+DVzZQSEeqAvEAOUApyKOS0REREREbmOJiYk8+uijJCYm4urqSnBwMMuXL6djx44AJCQkMGzYMJKSkvDz86Nv37689NJLtusdHBxYs2YNb7/9NmlpaQQEBNClSxfGjh2bZwXxF198wX/+8x/buPfddx/vvfeerb1t27Z88cUXTJ06lSlTpuDk5ESzZs1YsWIFzs7ON+lp3MDlS7B7sbUcxfFfoWRpCHkEs8lA9lgq8v1vJ1i6dBPHz2bgUMKOsCAvuoX407amN6UdCrVtTZFJTU2ld+/efPrpp4wfPz5P26OPPprn+M033+STTz4hOjra9jk9++yzAHlWEOdn8eLFrFu3joULF/LDDz/kaXNycsLJ6Y9fRS9cuMCSJUsYOXKkajmLiIgUgcL8lPET0AGIBRYA7xiG0R64F1hdDLGJiIiIiMhtavbs2ddtHzp0KEOHDr1me0BAABs2bLjhPO7u7nz++fW3XOnZsyc9e/a8TVP0xAAAIABJREFU4Vg33bkTsG0WbJ8N6SfBoxp0nMThgO4s3pvOknknOHjqd+ztDFoFevJcuxq0r+ND2VIl/+nIGTRoEP/6178ICwu7KnGc26VLl4iMjKRs2bLUr1+/UHMcP36cJ598kuXLl+dJEF/LggULSE9P57HHHivUPCIiIpK/wiSOhwClct5PAi4DLbAmkV8r4rhERERERETuPKYJv2+2ri7eswRMC9ToyKlafVmYUoMlvyaye3E0hgFNq3jwWMsqdKrrh4ezwz8duc3MmTM5cODAdRP2S5cupWfPnly4cAE/Pz9Wr15tK1NRENnZ2fTu3Zvhw4cTEhJy3RrIV0RGRtK1a1fbxogiIiLy9xQ4cWya5plc7y3A68USkYiIiIiIyJ3mUjrELLDWLz65C0q5kd5wMD84duaLA3bs/CoF2EeDim683LU2XYL98Clb6obD3mxxcXGMHj2aTZs2UbLktVc+h4WFER0dTXJyMjNnzuThhx8mKirKtrHhjUycOBEHBweGDRtWoP67du0iKirKVkNbRERE/r5CFcQyDKMU0AuonXNqN/ClaZoZRR2YiIiIiIjIbe/0Qdj6Cez8HDJTuexVh211x/Nhcn02bk7HNM9R268s/w2vSddgPwI8Sv/TEV9XVFQUycnJ1KlTx3YuOzubjRs38tFHH5Geno6joyPOzs5Ur16d6tWr06xZMwIDA/n444/z1LG+njVr1vDTTz9dlZxu1qwZPXr0YN68eXnOR0ZGEhAQcNVmeSIiIvLXFThxbBhGQ2AJUBprnWOAx4AJhmF0MU1zRzHEJyIiIiIicnuxWODAams5igM/YtqV4Lhve+Zkd2D2MR8uH4OqnjC0bSDdQvyo7l3mxmPeIiIiImjcuHGecwMGDCAwMJDRo0fj4JB/SQ2LxUJmZmaB55k1axbp6em24xMnTtCxY0fmzZtHixYt8vS9ePEic+fOZejQodjZ2RXibkREROR6CrPiOBL4GRhgmmY6gGEYzsCnOW2Nr3PtVQzDeAGYCLxvmuaQnHMGMBYYBLgDvwBPm6a5K9d17sA7wH05p74H/mOaZkquPvWA94B7gDPADOBV0zTNXH0eBF4FqgEHgTGmaX5XmHsQERERERGxuXAGoufB1o/h7BEulvJilXt/Xj/VlPhDrpR3c2JgK3+6hfhR268s1l9/bi9ubm64ubnlOefs7IyHhwd169bl3LlzTJkyhW7duuHn58epU6d4//33OX78OA8//LDtmsTERBITE9m3bx8Au3fvJiUlhYoVK+Lh4UGVKlXyzOHi4gJAtWrVqFChQp62hQsXkpqaqk3xREREilhhEsd1gL5XksYApmmmG4bxCrCtMJMahtEMa3I45k9NI4HhQH8gDngZWG0YRpBpmudz+nwBVASufAfpY2Au0C1n7LLAamAj0ASoCcwC0oE3cvqEAl9hTVJ/CzwAfG0YRgvTNH8pzL2IiIiIiMhdLjEWfo3EjPka43IGB5yC+cDyLN+nNMQ925ku9/hxX31/GgS43ZbJ4sIoUaIEu3bt4tNPP+X06dOUK1eOJk2asHHjRoKDg239PvroI8aPH2877tKlC2Bdady/f/9CzTlz5kw6duxIxYoVi+QeRERExKowieO9gD/Wusa5+QH7CjqIYRiuwDysZS7G5jpvAM8Ck03T/CbnXD/gJNa6yjMMw6iFNWHc0jTNqJw+g4GfcpLLcUBvrOU0+uXUXv6fYRg1gWGGYbyZs+r4WWCdaZoTcqafYBhGWM75Rwr8RERERERE5O50+RLsXYL5SyTGsS1cMhz53mzJJ5ntSLCrTqf6fnwW4kfTKuWwt7uzk8Xr16+3vS9dujTffXfjL3KOGzeOcePGFXiOypUrk+sLpHls2LChwOOIiIhIwRUmcfwi8E7OCuMtOeea5ZwfZRiGx5WOpmmeuc44kcBC0zTXGYYxNtf5KoAvsCrXOBmGYWwEmmMtNxEKpAGbc133M9bVxM2xrlIOBX7604Z9K7GWpagMHM7p8+6f4loJDLlO3CIiIiIicrc7n4hl2ywu//opDhkniceHWVm9WV7iXprVqcbIEH9aVvekpL1q7YqIiMjtzbjWX22v6mgYllyHVy4y8jk2TdO0v8YYjwNPAM1M08wyDGM98D/TNIcYhtEcaxK4kmmaR3Nd8ylQ3jTNjoZhjAYGmqZZ9U/jHgJmmqY5yTCMVcBx0zQfy9VeEfgdaG6aZpRhGJdyxvksV5++OWM45hP3IKylNfDx8Wk0f/786z2qO05aWpqtppjcffT53930+Yv+Hbi76fP/67777juWLl1KYmIiYF0t+eijjxIaGmrrc+zYMWbOnMmOHTu4fPkyFStWZMyYMVSqVAmAM2fO8NFHH7Ft2zYuXLhA+fLl6dmzJ+3bt7eNsW/fPiIjI9m7dy/29va0atWKp59+GicnJwBWrFjB66+/nm+MH374ITVr1rzmPejzz8U0KZu6G/cjy6iYsgV7slmXHcI8SwfOeTbkHj8Hgr3scbC/c1YW6/O/u+nzv7vp87+76fO/+4SFhW03TTPfvesKs+I47O8EYRhGENbN8Fqappn1d8a62UzTjMS6UprGjRubbdq0+WcDusnWr1/P3XbP8gd9/nc3ff6ifwfubvr8/7rU1FTCw8MJDAzEYrEwZ84cXn75ZbZv305wcDCHDx/moYceom/fvrz11lu4ubmxd+9e6tWrR0BAAAAdOnTgzJkzLF++HC8vL7777jtGjBhBx44dad26NSdOnOCBBx7goYceYt68eZw7d45nn32WTz75hIULFwLQtGlTnn322TyxjRgxgs2bNzN48ODr1tu9Wz//rd/PIGDHVLzNU5w0PDnj2xLPc7vwvrCfc2Zp5lg6cqBSD5o2bsJbtXxwcSzMr1S3j7v18xcrff53N33+dzd9/pJbgX/KMU3z7xaOCgU8gV25fkC1B1obhvEE1s33AHyAo7mu8wESc94nAl6GYRg5tYqv1Eb2/lMfnz/N7ZOr7Xp9EhERERER+Zu6d++e53jChAl8+OGHREVFERwczJgxY+jQoQNvvPGGrU/Vqnm+VMfmzZt59913adq0KQDDhw/nnXfe4ddff6V169YsXboUOzs7PvjgA+ztrV/4++ijjwgODubAgQNUr14dJycn2+pjgAsXLrBkyRJGjhx5x2/S9lds/X4Gdbe/iJNxCQzwJRnfxEXEWzz4tNwzuN7TmwdDquJauuQ/HaqIiIhIsSt04S3DMPwNw2hmGEbr3K8CXLoIqAfUz/XaBszPeb8Pa+LW9t07wzBKAa34o6ZxFOCCNQl9RSjg/Kc+rXKuvaI9cAI4kqtPe/JqT97aySIiIiIif1t2djbz588nLS2N5s2bY7FYWLJkCbVr1yY8PBwvLy+aNGnCV199lee6li1bsmDBAk6fPo3FYmHx4sWcOnWKdu3aAZCZmUnJkiVtSWPAliTetGlTvrEsWLCA9PR0HnvssXzb71bZFpOtR85QfvsUa9L4T+zs7HjsmVd4MDRISWMRERG5axR4xbFhGP7AF0BrrDWNDf6obQzW1cPXZJpmCpDypzHTgTOmaf4v5/gtYLRhGHuxJpJfxLoZ3hc5Y+wxDGMFMCOn7jBYN81bappmXM7xF8BYYLZhGK8BNYBRwHjzj4LObwMbDcMYhTWhfT/WUhwtC/o8RERERESuJzY2ltDQUC5evIiLiwvfffcd9erVIzExkbS0NCZOnMirr77K5MmTWbt2Lb1798bFxYUuXboA1iRvz5498fT0pESJEjg6OvLll19Sv359ANq2bcuwYcOYPHkyw4YNIz09nVGjRgGQkJCQb0yRkZF07doVX1/fm/MQbmFpmZfZuO8UP+5JYt3ek5y9kMUhx+R8+/qYp29ydCIiIiL/vMIU5HoLyAZqA1uBcKzlHV4BniuieKYATsD7gDvwC9DBNM3zufr0At4FVuYcfw8MudJommaqYRjtc8bYBpwF3gDezNVns2EYPYHXcuI/CPQwTfOXIroPEREREbnLBQUFER0dTWpqKgsXLqRfv36sX78eDw8PwFrOYtiwYQDUr1+fbdu28d5779kSxy+++CLJycn8+OOPeHp6smjRIvr27cvGjRsJCQmhTp06zJkzh2HDhjFmzBhKlCjB0KFD8fHxwc7u6i8W7tq1i6ioKJYtW3bzHsItJj4lgzV7kli9O4lfDp3hUrYFt9IlCQvypl0tH5K+9cKPU1ddd9LwRKl2ERERudsUJnH8f0AX0zT3GoZhAqdM0/zZMIxM4FVgdWEnN02zzZ+OTWBczuta15wFHr3BuLFYV0Zfr89CYGHBIhURERERKRwHBweqV68OQKNGjdi6dSvTp0/nww8/pESJEtSuXTtP/1q1ajF//nwADh48yLvvvkt0dDQhISEAhISE8NNPP/Huu+/y8ccfA9CrVy969epFUlISzs7OGIbBm2++eVW9ZLCuNg4ICCA8PLw4b/uWYrGYxMan8uOeJH7cc5I9CecAqOrlTP8WlWlXy4eGFd0oYW9NtG89MgK3KzWOc2SYDhxrNEKJYxEREbnrFCZx7ARc+e7WGawb0u0DdgPBRRyXiIiIiMgdxWKxkJmZiYODA02aNCEuLi5P+759+6hUqRJg3cQOyFO/+MqxxWK5amwfH+u+z59++imlSpWiffu823lcvHiRuXPnMnTo0HxXI99JMi5l8/OBZNbstSaLT53PxM6AxpU9GNO5FvfW8qaql0u+1za5bzBbgYAdU/E2kzlpeHKs0Qia3Df45t6EiIiIyC2gMInjvUBNrBvMRQNPGIZxDHgaiC/60EREREREbk+jRo2iS5cuBAQEcP78eb744gvWr19vKxMxcuRIHn74YVq1akXbtm1Zt24d8+fPZ9GiRQDUrFmT6tWr89RTTzFt2jTKlSvHokWLWL16NYsXL7bN89577xEaGkqZMmVYvXo1I0aMYPLkybi5ueWJZ+HChaSmpt6xm+KdPH+RtXtO8uOeJDYdSOZilgUXxxL8X5AX7Wp506aGN+7ODgUaq8l9gyEnUeyb8xIRERG5GxUmcfw2f/zc9AqwAmu94UygbxHHJSIiIiJy20pMTOTRRx8lMTERV1dXgoODWb58OR07dgQgIiKCyMhIJk6cyDPPPENgYCCfffaZrb5xyZIl+eGHHxg1ahTdunUjLS2N6tWrM2vWLLp162ab59dff2Xs2LGkpaVRs2ZNZsyYQZ8+fa6KZ+bMmXTs2JGKFSvenAdQzEzTZG/ieX7cncSPe0/y2zHrHtwV3J3o2aQi7Wr5cE8VDxxK3Nmrq0VERESKU4ETx6Zpzsv1fodhGJWxrkA+appm/tsPi4iIiIjchWbPnn3DPv3796d///7XbA8MDOSbb7657hifffZZgeLZsGFDgfrdyjIvZ/PLoTP8uCeJNXtOEp+SgWFASAU3RnQM4t5a3gT5lMEwjH86VBEREZE7QoESx4ZhPIl1c7zlpmnOyTkeARjAB8DU4gtRRERERETuRmfTL7EuzlqCYkPcKdIvZVOqpB2tAr0Yem91wmp6412m1D8dpoiIiMgd6YaJY8MwemNNDK8EJhuGUQV4DpgC2AMvGYZx0DTNb4s1UhERERERueMdPJXGj7utq4q3/X4GiwneZRy5r3552tf2pnk1T0qVtL/xQCIiIiLytxRkxfGTwJOmac41DKMJEAU8bZrmDADDMBKAIYASxyIiIiIiUiiXsy1s+/2sNVm89ySHk9MBqO1XliFtA2lXy5u6/q7Y2akEhYiIiMjNVJDEcW1gM4BpmlsNw7AAW3K1bwAmFUNsIiIiIiJyBzp3MYsNcadYsyeJdXGnSM3IwsHejtBq5XisRWXa1vKhvJvTPx2miIiIyF2tINsMlwIu5DrOBM7nOr4AlCzKoERERERE5M5y7MwFZv18mN4fb6HhK6v5z5c72bg/mXa1fPjo0YbseLk9cx67hz6hlW/ppPH7779PcHAwZcuWpWzZsoSGhrJs2TJbe1JSEv3798ff35/SpUsTHh7O/v3784zRpk0bDMPI8+rZs2eePmfPnqVPnz64urri6upKnz59SElJydNn5cqVhIaGUqZMGTw9PenevTv79u0rvpsXERGRu0pBVhyfACoDCTnH/wYSc7UHAkeKNCoREREREbmtWSwm0cdTbPWK45Ksa0+qe7swsFVV2tf2pn6AO/a3WQmKChUq8PrrrxMYGIjFYmHOnDlERESwfft26tWrR0REBHZ2dixatAhXV1fefPNN2rVrx+7du3F2draNM2DAACZOnGg7dnLKmyzv1asXR48eZcWKFQAMHDiQPn3+n707j8u6zPc//rpvREDWm32RJVyASkWBETRzQ8EUxTMnyzRzfpNZek6mJwsrl5pcSsuZM5OV1bGGFrdMRx1TR2EmDMsNNbfcTRZREBVkUe7v7w/0Hhk1axJQeT8fDx8P7+91fa/vdX3vHg/p7eXnepTly5cDcPjwYQYMGMDTTz9Neno6paWlPPfcczzwwAMcOHCgHt6EiIiI3Ol+SnC8EfgPamobYxjGwn9p/w21S1eIiIiIiEgjdL7qIl/tP8W6PSdYv7eQU6VV2JlN/CrMk5f6RpEY5UeYt/ONB7qFDRgwoNbnqVOn8vbbb5OdnY2joyMbN24kJyeHdu3aAfD222/j7+/PZ599xuOPP267r1mzZvj7+1/zGXv27OHLL78kKyuLhIQEAN599126dOnCvn37iIiIYMuWLVy4cIHp06djZ1dzWOCECRPo0aMHp06dwtvbuy6WLyIiIo3ITwmOH71B+wtAyQ36iIiIiIjIHajgTAXr9p7gb7tPsOFgEVUXrbg6NqFbhC+JUb50a+2Le7M7s7JddXU1ixYtorS0lE6dOlFZWQmAo6OjrY/ZbMbBwYGsrKxawfH8+fOZP38+fn5+9OnTh8mTJ+Pq6gpAdnY2Li4udOrUyda/c+fOODs78/XXXxMREUFcXBz29va8//77PP7445w/f56PPvqIuLg4hcYiIiJyU9wwODYMw7hB+/GbNx0REREREbmVGYbBrryz/G1PTQmKnblnAAjxbMbQjqEk3u1LXJgn9nY/5TiV29POnTtJSEigoqICFxcXvvjiC9q0acOFCxcICQnhhRde4L333sPFxYXZs2dz/Phx8vPzbfc/8sgjhIaGEhgYyK5du5gwYQI7duxgzZo1ABQUFODj44PJ9M8yHiaTCV9fXwoKaqoGhoaGsnbtWh588EFGjx6N1Wqlffv2rFq1qn5fhoiIiNyxfsqOYxERERERaSSWbstl5up95JaUE7RxPeOTIki+15/sQ0WsuxQW55+pwGSCDiEWnk+OJDHKl5a+LrWCzjtZREQEOTk5nDlzhsWLF/PYY4+RmZnJvffey5IlS/jtb3+Ll5cXdnZ2JCYm0qdPH67cj/PEE0/Yft+mTRvCw8Pp2LEjW7dupUOHDj9pDgUFBfz2t79l2LBhDB48mHPnzjFp0iQGDRrE+vXrMZvv3OBeRERE6oeCYxERERERAWpC4wlLdlJ+oRqA3JJyxi3MYfxiExeqDZo1teP+Vj6M6+VL90hfvF0cGnjGDaNp06a0bNkSgJiYGDZt2sTs2bP54IMPiImJsYXKVVVV+Pj40LFjR2JjY687XmxsLHZ2duzfv58OHTrg7+/PyZMnMQzDFsYbhkFhYaGtLvJbb72Fs7Mzr7/+um2cjz/+mODgYL7++mvuu+++OnwDIiIi0hjor6FFRERE5KZ76623aNu2LW5ubri5uZGQkMDKlStt7RMnTiQyMhJnZ2csFgs9e/bk66+/rjXGwYMHGThwID4+Pri5uTFo0CBOnDhha7darfTv35+QkBAcHR0JCAhg6NCh5Obm1hpn3bp1dOrUCVdXV/z9/Xn++ee5ePFi3b6A29SMVXttofFlVgOa2pn58DdxbJ3Yi3cejeHB2OBGGxpfi9VqtdU3vszd3R0fHx/279/P5s2brzpU70o7d+6kurqagIAAABISEigtLSU7O9vWJzs7m7KyMlvd4/Pnz9sOxbvs8mer1XpT1iUiIiKNm4JjEREREbnpmjdvzmuvvcbWrVvZvHkzPXr0IDU1lR07dgA1/9T/rbfeYufOnWRlZXHXXXeRnJxsC4bLysro3bs3hmGwfv16NmzYQFVVFSkpKbVCsR49erBw4UL27dvH559/zqFDhxg4cKCtffv27TzwwAP06tWLbdu2sWDBAv7yl7+QlpZWvy/kFna6rIrPvj3GI+9tpOBsxTX7nK+qpluEL472dtdsb0zS0tL46quvOHLkCDt37mTChAlkZmYyZMgQABYtWkRGRgaHDh1i2bJl9OrVi9TUVHr37g3U/IXIK6+8wubNmzly5Ah//etfefjhh2nfvj2dO3cGICoqiuTkZEaOHEl2djbZ2dmMHDmSfv36ERERAUDfvn3ZunUrr7zyCvv372fr1q385je/ITg4mJiYmIZ5OSIiInJHUakKEREREbnp/nV35dSpU3n77bfJzs6mbdu2DB06tFb7m2++yQcffEBOTg5JSUls2LCBw4cPs3nzZiwWCwAfffQRFouF9evXk5iYiNls5plnnrGNERoaSlpaGgMGDKCiogJHR0cWLFjA3XffzcsvvwxAy5Ytef311xk0aBCTJ0/G1dW1jt/ErelsxQXW7jrB8h15ZO0/xUWrQbi3M66OTThXcfVu7EAPpwaY5a2poKCAoUOHUlBQgLu7O23btmXVqlUkJSUBkJ+fz7hx4zhx4gQBAQEMGzaMiRMn2u5v2rQp69at4w9/+AOlpaUEBwfTt29fJk+eXGsH8aeffsp///d/28bt378/f/rTn2ztPXr04NNPP2XmzJm8/vrrODk5ER8fz5dffomzs3M9vQ0RERG5kyk4FhEREZE6VV1dzaJFiygtLbX9M/srVVVVMXfuXNzc3IiOjgagsrISk8mEo6OjrZ+joyNms5msrCwSExOvGqe4uJhPPvmEjh072u6rrKysNQaAk5MTFRUVbNmyhW7dut3Eld7azlddZN2eQpZvzyPz+5NUXbQS5OHE413CSWkXwN0BbizLyatV4xjAyd6O8UkRDTjzW8uHH374o+1PP/00Tz/99HXbg4OD+fvf/37D51gsFj7++OMf7fPwww/z8MMP33AsERERkX+HgmMRERERqRM7d+4kISGBiooKXFxc+OKLL2jTpo2tfcWKFTz88MOcP3+egIAA1q5di5+fHwDx8fG4uLgwfvx4XnvtNaCmREB1dTX5+fm1nvP888/zpz/9ifPnzxMfH8+KFStsbUlJScyePZuPP/6Yhx9+mBMnTvDKK68AXDXOnajiQjV///4ky7fnsW5PIeUXqvF1dWBox1BS2gUQHexhO3wNILV9EAAzV+8jt6ScIA8nxidF2K6LiIiISOOhGsciIiIiUiciIiLIycnhm2++4amnnuKxxx7ju+++s7V3796dnJwcvv76a5KTkxk0aJAtzPXx8WHRokWsWrUKV1dX3N3dKSkpoUOHDpjNtX+EHT9+PNu2bWPNmjXY2dkxdOhQDMMAoHfv3syaNYvRo0fj6OhI69ateeCBBwCuGudOcaHaSsa+QsYtzCHu1b8xMn0LXx8s4tcxQcx/Ip7sCT2ZlHI37UMstULjy1LbB7EhrQcfJjuzIa2HQmMRERGRRko7jkVERESkTjRt2pSWLVsCEBMTw6ZNm5g9ezYffPABAM7OzrRs2ZKWLVsSHx9Pq1ateP/99231YHv37s3Bgwc5deoUTZo0wcPDA39/f8LDw2s9x9vbG29vb1q3bk1UVBTBwcFkZWXRpUsXAMaNG8fYsWPJz8/HYrFw5MgRJkyYcNU4t7Nqq8E3h4pYviOfVd/lU3L+Aq6OTUi+15+UdoF0auFFE7s7MygXERERkbqh4FhERERE6oXVaqWysvJnt3t7ewOwfv16CgsL6d+//4+OAVw1jslkIjAwEIDPPvuM4OBgOnTo8LPXcCuxWg22HjvNih35rNyZz8lzlTRrakevu/1IaRtIl9beODSxu/FAIiIiIiLXoOBYRERERG66tLQ0+vbtS3BwMOfOnePTTz8lMzOTlStXcvbsWV5//XVSUlIICAjg5MmTvPXWWxw/fpxBgwbZxpg3bx6RkZH4+vqSnZ3NmDFjGDt2LBERNQe1ZWdns3XrVu677z48PDw4ePAgEydOJCwsjPvuu882zsyZM0lOTsZsNrNkyRJmzJjBwoULsbO7/UJVwzDYmXuGFTvyWbE9j7wzFTg0MdMj0peUdoF0j/DFqentty4RERERufUoOBYRERGRm66goIChQ4dSUFCAu7s7bdu2ZdWqVSQlJXH+/Hl27drF//3f/1FUVISXlxdxcXH84x//oG3btrYx9u3bx4QJEyguLiYsLIwXX3yRsWPH2tqdnJxYvHgxkyZNoqysjICAAJKTk1mwYAGOjo62fqtWrWLq1KlUVlbSrl07li1bRp8+fer1ffxS+wrOsXx7Hst35HG06Dz2dibub+XDc8mRJN7th4uDfqwXERERkZtLP2GKiIiIyE334YcfXretWbNmfPHFFzccY8aMGcyYMeO67dHR0WRkZNxwnPXr19+wz63o0MlSVuzIZ/n2PPYXlmI2QeeW3ozq1oKke/zxaNa0oacoIiIiIncwBcciIiIiIreI46fP28LiXXlnMZkgLsyT3w24h+R7A/BxdWjoKYqIiIhII6HgWERERESkAZ04W8HKHfks35HHtmMlAEQHe/BS3yj6tg0gwN2pgWcoIiIiIo2RgmMRERERkXpWVFrJqu8KWL49j2+PFGMYEBXgxnPJEfRrE0iIV7OGnqKIiIiINHIKjkVERERE6sGZ8gus3lXAih35bDhwimqrQQsfZ8b0bEW/toG09HVp6CmKiIiIiNiYG3oCIiIiIneit956i7Zt2+Lm5oabmxsJCQmsXLnS1r5kyRKSkpLw8fHBZDIIp15QAAAgAElEQVSRmZl51RgHDx5k4MCB+Pj44ObmxqBBgzhx4kStPt9//z2pqal4e3vj6upKfHw8X375Za0+x44dIyUlBWdnZ7y9vXn66aepqqqqk3VLbWWVF1mWk8vjH20m7tW/8dziHRw+VcrI+8P569Nd+Nu4rjyT2FqhsYiIiIjccrTjWERERKQONG/enNdee41WrVphtVr56KOPSE1NZcuWLbRt25aysjI6derE0KFDGTZs2FX3l5WV0bt3b9q0acP69esBmDhxIikpKWzcuBGzuebv//v160d4eDjr1q3D2dmZd955hwEDBrB7925atGhBdXU1ffv2xcvLi6+++oqioiIee+wxDMPgj3/8Y72+k8ai4kI1mfsKWb49n3V7T1BxwYq/myOPJoSS0i6Qds3dMZlMDT1NEREREZEfpeBYREREpA4MGDCg1uepU6fy9ttvk52dTdu2bXn00UcBOHXq1DXv37BhA4cPH2bz5s1YLBYAPvroIywWC+vXrycxMZFTp06xf/9+3n33Xdq1awfAjBkzmD17Ntu2baNFixasWbOGXbt2cfToUYKDgwF4/fXXefzxx5k6dSpubm519QoalaqLVrIOnGT59nzW7CqgrKoab5emPBgTTEq7QGJDLZjNCotFRERE5Pah4FhERESkjlVXV7No0SJKS0vp1KnTT7qnsrISk8mEo6Oj7ZqjoyNms5msrCwSExPx8vIiKiqK9PR04uLicHJyYu7cubi6utK5c2cAsrOziYqKsoXGAElJSVRWVrJlyxa6d+9+cxfbiFystrLxUDHLt+fx5a4CzpRfwM2xCf3aBpLSLpD4cE+a2KkynIiIiIjcnhQci4iIiNSRnTt3kpCQQEVFBS4uLnzxxRe0adPmJ90bHx+Pi4sL48eP57XXXgMgLS2N6upq8vPzATCZTKxdu5aBAwfi5uaG2WzG09OTVatWERAQAEBBQQF+fn61xvb29sbOzo6CgoKbuNrGwWo12Hz0NMu357Hqu3xOlVbh3NSO3vf4k9IugPta+tC0icJiEREREbn9KTgWERERqSMRERHk5ORw5swZFi9ezGOPPUZmZib33nvvDe/18fFh0aJFPPXUU8yZMwez2czgwYPp0KGDrb6xYRiMGjXKVr/YycmJ999/n1//+tds2rSJoKCgul7iHWnptlxmrt5HXkk5gR5OPNu7NXf5uLBiex4rduRTcLYChyZmEqP86Nc2gO6Rvjja2zX0tEVEREREbioFxyIiIiJ1pGnTprRs2RKAmJgYNm3axOzZs/nggw9+0v29e/fm4MGDnDp1iiZNmuDh4YG/vz/h4eEArF+/nuXLl1NcXIyHhwcAc+bMYe3atcybN4+XXnoJf39/NmzYUGvcU6dOUV1djb+//01c7Z1h6bZcJizZSfmFagByS8oZt3A7BmBvZ6Jra18mPBBJzyg/XBz0o7SIiIiI3Ln0066IiIhIPbFarVRWVv7s+7y9vYGaoLiwsJD+/fsDcP78eQDbDuTLzGYzVqsVgISEBF599VWOHz9O8+bNAVi7di0ODg7ExMT822u501itBgdPljLlL7tsofFlBuDRzJ6/P9sd92b2DTNBEREREZF6puBYREREpA6kpaXRt29fgoODOXfuHJ9++imZmZmsXLkSgOLiYo4dO0ZJSQkABw4csO0ovrwTeN68eURGRuLr60t2djZjxoxh7NixREREADWhsKenJ7/5zW+YNGkSTk5OvPfeexw6dIh+/foBNbuW77nnHoYNG8Ybb7xBUVER48ePZ8SIEbi5uTXAm7k1lFVeZPsPJWw5epotx06z9ehpzlZcvG7/M+cvKDQWERERkUZFwbGIiIhIHSgoKGDo0KEUFBTg7u5O27ZtWbVqFUlJSQD85S9/4Te/+Y2t/4gRIwCYPHkyU6ZMAWDfvn1MmDCB4uJiwsLCePHFFxk7dqztHm9vb7788ktefPFFevTowYULF4iKimLp0qV06NABADs7O1auXMmoUaPo3LkzTk5ODBkyhJkzZ9bTm2h4hmGQW1LOlqM1AfGWY6fZk3+OaqsBQGs/F/q2DaBDiIWZq/dReO7qXeGBHk71PW0RERERkQal4FhERESkDnz44Yc/2j58+HCGDx/+o31mzJjBjBkzfrRPbGwsq1ev/tE+ISEhrFix4kf73EmqLlrZlXemJig+dpotR09z4mxNGNysqR3RwR6M7taCDqEW2gdbau0ktrcz16pxDOBkb8f4pIh6X4eIiIiISENScCwiIiIit7Wi0kq2Hiux7SjefryEyos1NZ6bW5yID/ciJtRChxALkf6uNLEzX3es1PZBAMxcvY+8knICPZwYnxRhuy4iIiIi0lgoOBYRERGR24bVarC/sLSmNvGlHcWHT5UBYG9n4p5Ad4bGhxIbaqFDqAU/N8ef/YzU9kEKikVERESk0VNwLCIiIiK3rNIrD7G7FBSfu3SInZdzUzqEWngoLpiYUAttgtxxtLdr4BmLiIiIiNwZFByLiIiIyC3BMAyOny63hcRbjp5mb8FZrAaYTNDa15V+bQOJCbUQG2oh1KsZJpOpoactIiIiInJHUnAsIiIiIg2i8mI1u/LOsvWKoLjwXM0hds5N7WgfYuG/erQiJtRCdLAH7k72NxhRRERERERuFgXHIiIiIlIvTpVW2g6w23L0NDtyz1B16RC7YE8nOrWoOcQuJtSTCH9X7MzaTSwiIiIi0lAUHIuIiIjITVdtNdhfeO6ftYmPnuZI0XkAmtqZuTfIjccSQokJtdAhxILvv3GInYiIiIiI1B0FxyIiIiLyi52ruMD2H86w+WgxW46eJudYCecqaw6x83ZpSocQC4N/FUJMqIV7dYidiIiIiMgtz9zQExAREZE70/Tp04mLi8PNzQ0fHx9SUlL47rvvavU5ceIEw4cPJzAwkGbNmpGcnMz+/fuvOZ5hGPTp0weTycTixYtt161WK/379yckJARHR0cCAgIYOnQoubm5tj4nT54kKSmJwMBAHBwcCA4OZvTo0Zw5c6ZuFn+bW7otl84z1jP8yzI6z1jP0m25tdoNw+BY0Xm+2Hacl5bupM8fvqLdy2sY+sE3/GHdfk6eq6R/dCBvDmrH38d3Y9OLicwdFsvIri2IDfNUaCwiIiIichvQjmMRERGpE5mZmYwaNYq4uDgMw2DSpEkkJiaye/duPD09MQyD1NRUzGYzS5cuxd3dnTfffNPWx9nZudZ4b7zxBmbztf/Ou0ePHrzwwgsEBASQm5vLs88+y8CBA/n2228BMJvNDBw4kGnTpuHt7c2BAwcYPXo0I0aMYOHChXX+Lm4nS7flMmHJTsovVAOQW1JO2pIdHD5VhrOD3aXSEyWcKq05xM7FoQntQzzoffkQuxAP3Bx1iJ2IiIiIyO1OwbGIiIjUidWrV9f6nJ6ejru7Oxs2bCAlJYX9+/ezceNGcnJyaNeuHQBvv/02/v7+fPbZZzz++OO2ezdt2sQf/vAHtmzZgp+fX61xzWYzzzzzjO1zaGgoaWlpDBgwgIqKChwdHfHy8uLJJ5+s1WfUqFFMnz69LpZ+W5u5ep8tNL6s4oKVP6yr2Qke4tmM+1t50yHUQkyohdZ+OsROREREROROpOBYRERE6sW5c+ewWq1YLBYAKitrdqw6Ov7zUDSz2YyDgwNZWVm24PjcuXM88sgjzJ07F19f3xs+p7i4mE8++YSOHTvWGvtKeXl5LFmyhK5du/7SZd0xjhaVsX5vIbkl5dft8+2LPfF11SF2IiIiIiKNgWoci4iISL0YM2YM0dHRJCQkABAZGUlISAgvvPACxcXFVFVV8dprr3H8+HHy8/Nt9z355JMkJyfTp0+fHx3/+eefx9nZGS8vL44dO8aKFSuu6jN48GCaNWtGUFAQrq6uzJs37+Yu8jZyodpK9sEipv11Dz3fyKTrzExeXr6bJtfZPRzk4aTQWERERESkEVFwLCIiInVu3LhxZGVl8fnnn2NnV3Mwmr29PUuWLOHgwYN4eXnRrFkzMjIy6NOnj62WcXp6Otu3b2fmzJk3fMb48ePZtm0ba9aswc7OjqFDh2IYRq0+s2fPZuvWrSxbtoxDhw7VKnHRGBSXVbFk63FGf7qVDr9by+D3NvLhhiMEejgxOeVu/j6+G7MebIfTvxxe52Rvx/ikiAaatYiIiIiINASVqhAREZE6NXbsWObPn09GRgbh4eG12mJiYsjJyeHMmTNUVVXh4+NDx44diY2NBWDdunXs3r0bFxeXWvc99NBDJCQkkJWVZbvm7e2Nt7c3rVu3JioqiuDgYLKysujSpYutj7+/P/7+/kRGRuLp6UmXLl146aWXCA4OrsM30HAMw2BP/jky9hWybs8Jtv1QgmGAj6sDD9wbQPdIX+5r5Y2Lwz9/JAz1qjmUcObqfeSWlBPk4cT4pAhS2wc11DJERERERKQBKDgWERGROjNmzBgWLFhARkYGkZGR1+3n7u4OwP79+9m8eTO/+93vAJg6dSrPPvtsrb5t2rRh1qxZDBgw4LrjWa1W4J91lP/dPrej8qpqvj54inV7C8nYW0j+mQoA2jZ3Z0zPVvSI9OXeQHfMP3KgXWr7IFLbB5GZmUm3bt3qaeYiIiIiInIrUXAsIiIidWL06NGkp6ezdOlSLBYLBQUFALi4uNh2EC9atAhvb29CQ0PZuXMnY8aMITU1ld69ewMQFBREUNDVO12Dg4Ntu5ezs7PZunUr9913Hx4eHhw8eJCJEycSFhbGfffdB8CKFSsoKioiJiYGFxcXdu3axfjx44mPj6dly5b18TrqVG5JOev3FrJ+zwm+PlhE5UUrzk3tuK+VN2MTW9MtwgdfN9UnFhERERGRn07BsYiIiNSJOXPmANCzZ89a1ydPnsyUKVMAyM/PZ9y4cZw4cYKAgACGDRvGxIkTf9ZznJycWLx4MZMmTaKsrIyAgACSk5NZsGABjo41YamjoyPvvPMOe/bsobKykuDgYAYOHEhaWtovX2gDqLYabDt2uiYs3lvI3oJzAIR4NuORjiH0iPTlV3d54tDE7gYjiYiIiIiIXJuCYxEREakT/3ow3bU8/fTTPP30079o3OjoaDIyMn70nsTERBITE3/Wc241Z85f4O/7T7J+zwn+/v1JTp+/QBOzidgwCy8+EEX3SF9a+DhjMl2/BIWIiIiIiMhPpeBYRERE5BZkGAYHCktZv7eQdXsL2XL0NNVWA0/npnSP8KVHlC9dWvng7mTf0FMVEREREZE7kIJjERERkVtExYVqvjlczPo9J1i/r5AfissBiApw46muLege6Ut0sAd2P3KwnYiIiIiIyM2g4FhERESkAZ04W0HGpV3FGw6c4nxVNY72Zjq38ObJri3oHuFLoIdTQ09TREREREQamXoLjk0m02hgJBB26dIu4FXDMFZeajcBk4EnAAvwDTDaMIxdV4xhAf4X6H/p0l+A/zYMo+SKPm2APwG/AoqBd4HfGVcURDSZTL8Gfge0AA4CLxqG8cVNXrKIiIjIVaxWgx25Zy4dbHeC73LPAhDk4cSvOzSnR6QvCS28cLTXwXYiIiIiItJw6nPH8XHgeWA/YAYeA5aaTKYYwzB2AM8B/wMMB/YBk4C1JpMpwjCMc5fG+BQIAZIvfX4fSAdSAEwmkxuwFvgHEAdEAvOAMuCNS30SgAXUhNRLgP8AFplMps6GYXxTV4sXERGRxutcxQWy9p9i3d5CMvcVcqq0CrMJOoRYeC45gh6RvkT4uepgOxERERERuWXUW3BsGMayf7n0oslkegpIMJlMO4FngBmGYXwOYDKZHgMKgUeAd00mUxQ1gfF9hmFkX+ozEvjqUri8DxgCNAMeMwyjHPjOZDJFAuNMJtObl3YdPwNkGIYx9dI8pppMpu6Xrg+uuzcgIiIijcnhU2W2XcXfHi7mQrWBm2MTukb40jPSl66tfbA4N23oaYqIiIiIiFxTg9Q4NplMdsCDgAvwNXAX4A+sudzHMIxyk8n0D6ATNeUmEoDSS/0v20DNbuJO1OxSTgC+uhQaX7aamrIUYcDhS33++C9TWg38181ZnYiIiDRGVRetbD5SzLq9hWTsLeTQqTIAWvm68P/uu4seEb7EhFpoYmdu4JmKiIiIiIjcmOmK0r91/7Ca+sPZgCM1IfAQwzBWmkymTtSEwKGGYRy7ov//AUGGYSSZTKYXgMcNwwj/lzEPAe8ZhjHdZDKtAY4bhvH/rmgPAY4CnQzDyDaZTFWXxvnzFX2GXRrD4TrzfoKa2sv4+fnFzJ8//ya8jdtHaWkpLi4uDT0NaSD6/hs3ff+i/wZ+3NlKgx2nLpJTWM2uomrKL0ITE0R52dHOp+aXT7PbNyjW99+46ftv3PT9N276/hs3ff+Nm77/xqd79+5bDMOIvVZbfe843gdEA+7AfwIfmUymbvU8h5/NMIy5wFyA2NhYo1u3bg07oXqWmZlJY1uz/JO+/8ZN37801v8Glm7LZebqfeSVlBPo4cT4pAhS2wdhGAa78s6SsbeQdXsL2X68BMMAPzcHUjsE0j3Cl84tvXF2aJB/1HXTNdbvX2ro+2/c9P03bvr+Gzd9/42bvn+5Ur3+X41hGFXAgUsft5hMpjhgLHC53rAfcOyKW/yAgku/LwB8TCaT6VKtYkw1J8j4/ksfv395rN8VbT/WpwARERFp9JZuy2XCkp2UX6gGILeknOcW72DBpmMcPnWegrMVmEzQrrkHYxNb0yPSl3sC3XSwnYiIiIiI3FEaejuMGXCgpvZwAdAL2ARgMpkcgS7A+Et9s6mpiZzAP+scJwDOV3zOBl4zmUyOhmFUXLrWC8gDjlzRpxcw84p59KJ27WQRERFppGau3mcLjS+rqray8VAxfdr40z3Cl24Rvvi4XrPClYiIiIiIyB2h3oJjk8k0A1gJ/AC4Ao8A3YC+hmEYJpPp98ALJpNpL/A98BI1dZA/BTAMY4/JZPoSePdSzWGoOTRvhWEY+y59/hSYDHxoMpleBVoDacDLxj+LOf8B+IfJZEoDlgIDge7AfXW2eBEREbnlna+6yJpdJ8gtKb9unzlDYupxRiIiIiIiIg2nPk9r8Qc+pqbO8TogDuhjGMaqS+2vA7OBt4DNQADQ2zCMc1eM8QiwHVh96dd24NHLjYZhnKFm93DgpTHeAt4A3ryiz9fAw8BwYAcwDHjIMIxvbupqRUSk0Zs+fTpxcXG4ubnh4+NDSkoK33333VX9vv/+e/7jP/4DDw8PmjVrRocOHdizZ4+tfcSIEbRo0QInJyd8fHwYMGBArfYrVVRU0K5dO0wmE5s3b67Vtm7dOjp16oSrqyv+/v48//zzXLx48eYu+jZzsdpKxr5Cnpm/jZjf/Y1nFuRgd52SE4EeTvU8OxERERERkYZTbzuODcMYfoN2A5hy6df1+pwGht5gnJ3A/TfosxhY/GN9REREfqnMzExGjRpFXFwchmEwadIkEhMT2b17N56engAcPnyYzp07M2zYMNavX4+Hhwd79+6tdZJxbGwsw4YNIzg4mOLiYqZMmUJiYiJHjhzB3t6+1jOfffZZmjdvzo4dO2pd3759Ow888ABpaWn8+c9/Jjc3lyeffJLq6mpmzZpV9y/jFmIYBtuPn2HptlxW7MjjVGkV7k72DOwQRGp0ELnF53lh6Xe1ylU42dsxPimiAWctIiIiIiJSvxq6xrGIiMgda/Xq1bU+p6en4+7uzoYNG0hJSQHgxRdfpHfv3rzxxhu2fuHh4bXuGzlypO33YWFhvPrqq7Rr145Dhw4REfHPMHPZsmVkZGSwePFi/vrXv9YaY8GCBdx99928/PLLALRs2ZLXX3+dQYMGMXnyZFxdXW/Oom9hR06VsTQnl6XbcjlSdJ6mTcwkRvmSGh1E1wgfHJrY1XS8yxOT2cTM1fvIKykn0MOJ8UkRpLYPatgFiIiIiIiI1CMFxyIiIvXk3LlzWK1WLBYLAFarleXLl5OWlkZycjJbtmwhLCyMZ599loceeuiaY5SVlTFv3jxCQkIICwuzXT9+/DhPPfUUq1atwsnp6pIKlZWVODo61rrm5ORERUUFW7ZsoVu3bjdtnbeSU6WVrNiex9KcPHJ+KMFkgoRwL0Z1a0lyG3/cHO2veV9q+yAFxSIiIiIi0qjVZ41jERGRRm3MmDFER0eTkJAAQGFhIaWlpUybNo3evXuzdu1aBg8ezJAhQ1i5cmWte+fMmYOLiwsuLi6sWrWKdevW4eDgAEB1dTVDhgzhf/7nf2jXrt01n52UlMQ333zDxx9/zMWLF8nNzeWVV14BID8/vw5XXf/OV11k6bZchs/7lo7T1jFl+W4qL1p54YFIvk7rwacj4hkUF3zd0FhERERERES041hERKRejBs3jqysLLKysrCzqymJYLVaARgwYADjxo0DIDo6ms2bN/OnP/2Jvn372u4fMmQIvXr1Ij8/n1mzZvHggw+yYcMGmjVrxrRp02jatKltjGvp3bs3s2bNYvTo0QwfPhwHBwcmTpzIV199hdl8+/898sVqK1kHTrF0Wy5rdp/gfFU1QR5OjLw/nNT2QbT2u/NLcYiIiIiIiNxMCo5FRETq2NixY5k/fz4ZGRm16hd7e3vTpEkT7r777lr9o6KimD9/fq1r7u7uuLu706pVK+Lj47FYLHz++ec8+uijrFu3jq+++uqqg/Li4+N56KGH+OSTT4Ca8Hrs2LHk5+djsVg4cuQIEyZMuKqm8u3ieofcDYgOYmD7IGJDLZjNpoaepoiIiIiIyG1JwbGIiEgdGjNmDAsWLCAjI4PIyMhabU2bNiUuLo59+/bVuv79998TGhp63TENw8AwDCorKwGYN28eZWVltva8vDySkpL45JNP6Ny5c617TSYTgYGBAHz22WcEBwfToUOHX7TG+nb5kLtlOXkcPlVmO+RuQHQQ3a485E5ERERERET+bQqORURE6sjo0aNJT09n6dKlWCwWCgoKAGy1igGee+45Bg0aRJcuXejRowcZGRnMnz+fpUuXAnDgwAE+//xzEhMT8fHx4fjx48yYMQMHBwf69esHwF133VXruZfHbtGiBc2bN7ddnzlzJsnJyZjNZpYsWcKMGTNYuHChrXTGrexah9zF3+XFU11b/OghdyIiIiIiIvLvUXAsIiJSR+bMmQNAz549a12fPHkyU6ZMASA1NZW5c+cybdo0xowZQ6tWrfjzn/9sq2/s4OBAZmYmb7zxBiUlJfj5+XH//feTnZ2Nv7//z5rPqlWrmDp1KpWVlbRr145ly5bRp0+fX77QOnK+6iJrd5/gi225fLX/FNVWg6gANyb0iaR/dCAB7k4NPUUREREREZE7loJjERGROmIYxk/qN3z4cIYPH37NtuDgYFatWvWznhsWFnbNZ69fv/5njdMQLh9ytywnj9W7CmyH3D1xfzip0UFE+OuQOxERERERkfqg4FhEREQa1LUOuXNzbMKA6CBSowOJC/PUIXciIiIiIiL1TMGxiIiINIhrHXLXM9KX1PY65E5ERERERKShKTgWERGRelNUWsmKHfl8sS33qkPuku71x91Jh9yJiIiIiIjcChQci4iISJ26fMjd0m25/EOH3ImIiIiIiNwWzA09ARERuXNNnz6duLg43Nzc8PHxISUlhe++++66/UeOHInJZGLWrFm1ro8YMYIWLVrg5OSEj48PAwYMYM+ePbb2I0eO8Nvf/pbw8HCcnJwIDw9nwoQJlJeX2/qcPHmSpKQkAgMDcXBwIDg4mNGjR3PmzJmbv3DhYrWVzH2FjF2QQ+yrf2PM/Bz2FZzjifvDWf3M/awa04WRXVsoNBYREREREblFacexiIjUmczMTEaNGkVcXByGYTBp0iQSExPZvXs3np6etfouXryYb7/9lsDAwKvGiY2NZdiwYQQHB1NcXMyUKVNITEzkyJEj2Nvbs3fvXqqrq3n77bdp1aoVe/bs4YknnqCoqIi5c+cCYDabGThwINOmTcPb25sDBw4wevRoRowYwcKFC+vlfdzpDMNgx/EzfKFD7kRERERERG57Co5FRKTOrF69utbn9PR03N3d2bBhAykpKbbrR48eZcyYMfztb3+jT58+V40zcuRI2+/DwsJ49dVXadeuHYcOHSIiIoLk5GSSk5NtfcLDw3nxxReZOHGiLTj28vLiySeftPUJDQ1l1KhRTJ8+/aatt7E6WlTG0m15LM3J1SF3IiIiIiIidwgFxyIiUm/OnTuH1WrFYrHYrl28eJHBgwfz0ksvERUVdcMxysrKmDdvHiEhIYSFhV2339mzZ2s951/l5eWxZMkSunbt+rPWIDWud8jdk13DSb43QIfciYiIiIiI3OYUHIuISL0ZM2YM0dHRJCQk2K5NnjwZb29vnnrqqR+9d86cOTz33HOUlZURERHBunXrcHBwuGbfo0ePMmvWLF544YWr2gYPHsyyZcsoLy+nX79+zJs375ct6g61dFsuM1fvI7eknKCN6xmfFEHve/yuOuQu0t9Vh9yJiIiIiIjcgRQci4hIvRg3bhxZWVlkZWVhZ1dTuiAzM5MPP/yQnJycG94/ZMgQevXqRX5+PrNmzeLBBx9kw4YNNGvWrFa/EydOkJycTK9evRg7duxV48yePZvJkyfz/fffM2HCBJ555hnefffdm7PIO8TSbblMWLKT8gvVAOSWlPM/C7djNsOFaoNAd0eeuD+c1OggIvxdG3i2IiIiIiIiUhcUHIuISJ0bO3Ys8+fPJyMjg/DwcNv1zMxM8vPzCQgIsF2rrq7m+eef5/e//z3Hjx+3XXd3d8fd3Z1WrVoRHx+PxWLh888/59FHH7X1KSgooEePHtx7772kp6djMl19EJu/vz/+/v5ERkbi6elJly5deOmllwgODq6j1d9+Xvtyry00vqzaMHCws+Pj38bpkDsREREREZFGQMGxiIjUqTFjxrBgwQIyMjKIjIys1TZq1Cj+8z//s9a1pKQkBg8ezGQnQpIAACAASURBVIgRI647pmEYGIZBZWWl7Vp+fj7du3fnnnvu4bPPPqNJkxv/EWe1WgFqjdPYGIbBD8XlbDpSzOajxWw6cpr8MxXX7FteVU3HcK96nqGIiIiIiIg0BAXHIiJSZ0aPHk16ejpLly7FYrFQUFAAgIuLCy4uLvj6+uLr61vrHnt7e/z9/YmIiADgwIEDfP755yQmJuLj48Px48eZMWMGDg4O9OvXD6g56K5bt24EBgby+9//nlOnTtnG8/Hxwc7OjhUrVlBUVERMTAwuLi7s2rWL8ePHEx8fT8uWLevpjTS8aqvBnvyzNUHxkdNsOlJM4bma4NzVsQmxoRYKz1ZwtuLiVfcGeqiGsYiIiIiISGOh4FhEROrMnDlzAOjZs2et65MnT2bKlCk/aQwHBwcyMzN54403KCkpwc/Pj/vvv5/s7Gz8/f0BWLNmDfv372f//v2EhITUuv/w4cOEhYXh6OjIO++8w549e6isrCQ4OJiBAweSlpb2yxd6CyuvqmbbD6dtIfG2YyWUVtaEwoHujiS08CI2zJO4MAutfV0xm01X1TgGcLK3Y3xSREMtQ0REREREROqZgmMREakzhmH87HuOHDlS63NwcDCrVq360XuGDx/O8OHDf7RPYmIiiYmJP3s+t5ui0ko2HTnN5iPFbDp6ml25Z7hoNTCZIMLPldT2gcSFeRIb5knQdXYQp7YPAmDm6n3klpQT5OHE+KQI23URERERERG58yk4FhERuU0ZhsHRovP/LDtxtJhDJ8sAaNrETHRzD564P5y4ME86hFhwb2b/k8dObR9EavsgMjMz6datWx2tQERERERERG5VCo5FRERuExerrezOP/vPHcVHTnOqtKY+sbuTPbGhFgbFBhMXZuHeIHccmtg18IxFRERERETkdqXgWERE5BZVVnmRnB9KbDuKtx47zfmqmrrDzS1OdGnlTWyYhbgwT1r6uGA2mxp4xiIiIiIiInKnUHAsIiJyizh5rtK2k3jz0WJ25Z2l+lJ94ih/Nx6MaU5smCexYRYC3K9dn1hERERERETkZlBwLCIi0gAMw+DwqbKa2sRHitl0pJgjRecBcGhiJjrYg6e6tiA2zEKHUAtujj+9PrGIiIiIiIjIL6XgWEREpB5cqLayK+/spR3FNaUnisqqALA0syc2zJNHOoYQG+bJvYHuNG1ibuAZi4iIiIiISGOm4FhERKQOlFZeZNux02w6XFN6IueHEsov1NQnDvFsRtcIH34V5klsmCctfJwxmVSfWERERERERG4dCo5FRERugsKzFWy6VHZi89FiduedxWqA2QR3B7rxUFwwcZfqE/u5OTb0dEVERERERER+lIJjERGRf7F0Wy4zV+8jr6ScQA8nxidFkNo+yNZuGAYHT5ax+Ugx314qO3GsuKY+sZO9He1DPPivHq2IC7PQPsSCi4P+uBUREREREZHbi/5PVkRE5ApLt+UyYclOW1mJ3JJy0pbs4EhRGc2a2rHpyGk2Hynm9PkLAHg5NyU2zMKwhFBiwzy5J9ANezvVJxYREREREZHbm4JjERGRK8xcvc8WGl9WccHK7/+2H4C7vJ1JjPKzlZ24y1v1iUVEREREROTOo+BYRETkkgvVVnJLyq/bvunFRHxcHepxRiIiIiIiIiINQ8GxiIg0egVnKvjs22PM33Tsun2CPJwUGouIiIiIiEijoeBYREQaJcMwyD5YRPrGo6zZfYJqq0HX1j6ktHXhk2+OUn7BauvrZG/H+KSIBpytiIiIiIiISP1ScCwiIo3KmfILLNl6nI83HuXgyTI8mtnz2/vuYkjHEEK9nAG4N8idmav3kVdSTqCHE+OTIkhtH9TAMxcRERERERGpPzr2XUSkDk2fPp24uDjc3Nzw8fEhJSWF7777rlYfwzCYMmUKgYGBODk50a1bN3bt2mVrz8zMxGQyXfPXokWLbP3CwsKuak9LS7O1nzx5kqSkJAIDA3FwcCA4OJjRo0dz5syZun8Rt4BdeWeYsGQH8dPW8fLy3bg62vPGg+3YOKEnLzwQZQuNAVLbB7EhrQeHZ/RlQ1oPhcYiIiIiIiLS6GjHsYhIHcrMzGTUqFHExcVhGAaTJk0iMTGR3bt34+npCcDrr7/OG2+8wYcffkhERASvvPIKvXr1Yt++fbi6utKpUyfy8/Nrjfu///u//PGPf6RPnz61rk+aNImnnnrK9tnFxcX2e7PZzMCBA5k2bRre3t4cOHCA0aNHM2LECBYuXFiHb6HhVFyoZtV3+aRnH2XrsRIc7c0MaBfE0PhQ2jR3b+jpiYiIiIiIiNyyFByLiNSh1atX1/qcnp6Ou7s7GzZsICUlBcMw+P3vf09aWhq//vWvAfjoo4/w9fXl008/ZeTIkTRt2hR/f/9a4yxevJjBgwfXCoYBXF1dr+p7mZeXF08++aTtc2hoKKNGjWL69Ok3Y6m3lB+Kz/PJN8dYuPkHisuquMvbmYn97uY/OzTHvZl9Q09PRERERERE5Jan4FhEpB6dO3cOq9WKxWIB4PDhwxQUFNC7d29bHycnJ+6//36+/vprRo4cedUYmZmZ7N+/n08//fSqtlmzZjF9+nSCg4N58MEHGT9+PE2bNr3mXPLy8liyZAldu3a9SatrWNVWg79/X8jHG4+Rsa8QE9Drbj8ejQ+jUwsvzGZTQ09RRERERERE5Lah4FhEpB6NGTOG6OhoEhISACgoKADAz8+vVj8/Pz9yc3OvOcbcuXOJjo4mNja21vWnn36a9u3b4+XlxbfffktaWhqHDx/m/fffr9Vv8ODBLFu2jPLycvr168e8efNu1vIaRFFpJQs3H+eTb45y/HQ5Pq4O/HePVgz+VTAB7k4NPT0RERERERGR25KCYxGRejJu3DiysrLIysrCzs7u3xqjqKiIJUuW8Oabb15z/Mvatm2Lm5sbDz30EK+99hpeXl62ttmzZzN58mS+//57JkyYwDPPPMO77777b82noRiGwdZjJXy88Sgrd+RTVW0lPtyTCX2i6H2PH/Z2OvtVRERERERE5JdQcCwiUg/Gjh3L/PnzycjIIDw83Hb9cj3iEydOEBISYrt+4sSJa9Yq/vOf/4ydnR1Dhgy54TM7duwIwIEDB2oFx/7+/vj7+xMZGYmnpyddunThpZdeIjg4+N9eX305X3WRZTl5pGcfZXf+WVwcmjD4V8EMjQ+llZ9rQ09PRERERERE5I6h4FhEpI6NGTOGBQsWkJGRQWRkZK22u+66C39/f9auXUtcXBwAFRUVfPXVV8ycOfOqsd5//30efPBB3N3db/jcnJwcAAICAq7bx2q1AlBZWfmT19MQDhSW8vHGo3y+5TjnKi8S6e/K1IH3khodhLOD/igTERERERERudn0f9siInVo9OjRpKens3TpUiwWi62msYuLCy4uLphMJp555hmmTZtGZGQkrVu35tVXX8XFxYVHHnmk1lhZWVns3r2buXPnXvWc7OxsNm7cSPfu3XF3d2fTpk2MHTuW/v3723Yyr1ixgqKiImJiYnBxcWHXrl2MHz+e+Ph4WrZsWfcv42e6UG1l7e4TpGcfJftQEU3tzDzQxp9HE0LpEGLBZNJhdyIiIiIiIiJ1RcGxiEgdmjNnDgA9e/asdX3y5MlMmTIFgOeee47y8nJGjx7N6dOn6dixI2vWrMHVtXbphffee4+oqCg6d+581XMcHBxYsGABL7/8MpWVlYSGhjJixAiee+45Wx9HR0feeecd9uzZQ2VlJcHBwQwcOJC0tLSbvOpfpuBMBZ99e4zPvj1G4blKgjyceC45gkGxwXi7ODT09EREREREREQaBQXHIiJ1yDCMG/YxmUxMmTLFFiRfz0cffXTdtg4dOrBx48YfvT8xMZHExMQbzqchGIZB9sEi0jceZc3uE1gNg66tfZgeH0q3CF/szNpdLCIiIiIiIlKfFByLiEiDOVN+gSVbj5O+8SiHTpbh0cyex++7i0c6hhDq5dzQ0xMRERERERFptBQci4hIvfsu9wyffHOUpdvyKL9QTfsQD94c1I4H2gTgaG/X0NMTERERERERafQUHIuISL2ouFDNX3fmk77xKNuOleBobyY1Ooih8aHcG+Te0NMTERERERERkSsoOBYRkTp1rOg8n3x7lIWbfuD0+QuEezszqd/d/DqmOe5O9g09PRERERERERG5BgXHIiJy01VbDf7+fSHp2UfJ/P4kZpOJXlF+PJoQSqcWXphMOuxORERERERE5Fam4FhERG6aotJKFm4+ziffHOX46XJ8XR347x6tGPyrYALcnRp6eiIiIiIiIiLyEyk4FhGRX8QwDLYeO0169lH+urOAqmorCeFevPBAFL3u9sPeztzQUxQRERERERGRn0nBsYiI/FvKKi+yLCeP9I1H2ZN/FleHJjzSMYQhHUNo5efa0NMTERERERERkV9AwbGIiPwsBwrP8fHGY3y+5TjnKi8SFeDGtIFtGBAdiLOD/lgRERERERERuRPo3w+LSJ37xz/+Qf/+/f9/e3ceH1V973/89SEEiAkkgWyQQACXACIYcSmClSgQiopc9SeKsty6tVqrgqhcF8CqV8WtLq2iXqi2Xq9aN6yCqCBaUi2pERSEokJIIAlLAgQCZPn+/pjJOBOSECDJhMz7+Xicx3DO+cz3nDOfSTj5zHe+X5KTkzEz5s2bF7C/tLSUG2+8kZSUFCIiIkhLS+Pxxx8PiJkzZw4ZGRnExMRgZqxfv/6A4xQXFzNhwgSio6OJjo5mwoQJlJSU+PavX78eMztgWbBgQVNc9lHt7a/yGfLgJ0xesJshD37CX7M38v7KzVw+5x8Mf2wpr3yRy/B+ifz112fy/m+HMv6MHioai4iIiIiIiLQi+itfRJpcaWkp/fv3Z+LEiUycOPGA/VOmTOGjjz7i5ZdfplevXixdupRrrrmGuLg4JkyYAMCePXsYOXIkF154Ibfcckutxxk/fjy5ubm+QvDVV1/NhAkTmD9/fkDcggULGDhwoG+9c+fOjXWprcLbX+Uz/c2VlJVXApBfUsbU11cAkBIbwe2j+nDpqSl0iWofzNMUERERERERkSakwrGINLnRo0czevRoACZPnnzA/mXLljFhwgQyMjIA6NmzJy+++CJffPGFr3B88803A7B8+fJaj7F69WoWLFjA559/zuDBgwF47rnnOOuss1izZg1paWm+2C5dupCUlNRo19fa/PcHq31FY39dItvx6bQMwtpYEM5KRERERERERJqThqoQkaAbOnQo8+fPZ+PGjYCnkJyTk8OoUaMa3EZWVhZRUVGceeaZvm1DhgwhMjKSZcuWBcRedNFFJCQkMGTIEN54443GuYijWFWV4+uNJTyycA2jnlhK4c59tcZt371fRWMRERERERGRENFshWMzm25m/zSznWa2xczmm1n/GjFmZjPNbJOZlZnZEjM7sUZMrJm9bGY7vMvLZhZTI+YkM/vU20a+md1jZlYj5mIzW2Vm+7yP/9F0Vy8i9XnyyScZOHAgPXr0IDw8nLPPPpuHHnqI888/v8FtFBQUEB8fj/+PupmRkJBAQUEBAFFRUTzyyCO89tprvP/++5x77rmMGzeOP//5z41+TS3d3vJKFq8p4r/eWsngBz/mwmf+zh+WrCM6IpzoiNq/jNItJqKZz1JEREREREREgqU5h6oYBvwB+CdgwL3AR2bWzzm33RtzGzAVmAysAe4BFplZmnNulzfmFaAHUN0V8QXgZeACADPrBCwClgKnAX2AucBu4FFvzGDg/4AZwJvARcDrZjbEOfdFE1y7iNTjqaeeYtmyZbz77rukpqaydOlSbr31Vnr27HlIvY4PJi4ujqlTp/rWTz31VLZu3crDDz/MlVde2WjHaamKd+9n8ZoiFq0qZOnaLezeX0lkuzDOTotneN9EMtISiI1sd8AYxwAR4WFMy0yrp3URERERERERaU2arXDsnMv0XzezCcAOYAgw39sj+GbgQefcX70xk4AiYDzwnJn1xVMwHuqcy/LGXAd85i0urwGuAI4BJjnnyoBvzKwPMMXMHnPOOe9xFjvn7veezv1mluHdfnkTvgwiUkNZWRnTp0/n9ddf54ILLgBgwIAB5OTk8MgjjzS4cJyUlMSWLVtwzvl6HTvnKCoqqnc84zPOOIO5c+ce+YW0UBu27WbRqkIWrSpk+YZiKqsciZ3aMzY9meH9EhncuwsdwsMCnjM2PRmA2QvXkF9SRnJMBNMy03zbRURERERERKT1C+bkeB3xDJVR7F3vBSQBH1YHOOfKzGwpcCbwHDAYKAX8Byz9O57exGfi6aU8GPjMWzSuthD4HdAT+NEb81SN81kI/KYRrktEDkF5eTnl5eWEhQUWL8PCwqiqqmpwO4MHD6a0tJSsrCzfOMdZWVns3r07YNzjmnJycujatevhnXwLVFXl+DqvhEWrCvlodSFrC0sB6JPUkeuHHcuIfon07xZNm4OMVTw2PZmx6cksWbKEYcOGNcOZi4iIiIiIiEhLEszC8e+BHCDLu17dJbCwRlwhkOwXs8XbaxgA55wzsyK/5ycBebW0Ub3vR+9jbceptVuimV0LXAuQmJjIkiVL6ruuVqe0tDTkrll+0hj5LysrIz8/H4CKigo+/fRTKioq6NixI4mJiQwcOJAbb7yRdevWkZiYyNdff828efO47rrrfMfevn0727dvZ/369QC8+uqrxMXFkZCQQKdOnQA4/fTTufLKK33DUTz66KMMHjyYzZs3s3nzZhYsWEDbtm05/vjjMTOysrJ44YUXuPbaa4/q9/j+SseqbZV8VVRJzpZKduxztDFIi23D+D7tSE8II/6YKmAz29dtZum6hretn3/ReyC0Kf+hTfkPbcp/aFP+Q5vyH9qUf/EXlMKxmT0GDMUz5ETlweKDzTk3B5gDcOqpp7pQ632nHoehrTHyv2TJEq655hrf+rx585g3bx6TJk1i3rx5LFiwgOnTpzN79my2b99Oamoq999/P1OnTvUNOzFz5kxmzZrla2P69OkAzJ07lzFjxgCwYMECbrzxRt++MWPG8PTTTxMT45k/c8OGDTz00ENs2LCBsLAwTjjhBObOnXtUjm+8ffd+Pl7t6VW8dO1WysoriWrflrPTkhjhHa84+pjwIz6Ofv5F74HQpvyHNuU/tCn/oU35D23Kf2hT/sVfsxeOzexx4DIgwzn3g9+uAu9jIpDrtz3Rb18BEG9mVt3r2Ds2ckKNmMQah03021dfTAEi0uiGDRuG3xcFDpCUlHTQcYZnzpzJzJkz642JjY3lz3/+c537J02axKRJk+ptoyX7cetuFq0q4KNVRSzfsJ0qB12jO3DJoBRG9EvkjN6dad827OANiYiIiIiIiIgcRLMWjs3s98A4PEXj72rs/hFP4XYE8E9vfAfgLGCaNyYLiMIzRnH1OMeDgUi/9SzgITPr4Jzb6902AtgErPeLGQHM9jv+CALHThYRCarKKkfOxmIWrSpi0aoCvt+yG4B+XTvxm3OOZ2S/RE7s1snXK1tEREREREREpLE0W+HYzJ4BJgBjgWIzqx5PuNQ5V+odq/gJ4L/M7DtgLXAXnsnwXgFwzq02swXAc95xh8Ezad57zrk13vVXgBnAPDO7DzgBuAOY5Tc28u+BpWZ2B/A28B9ABp7hM0REgqZsfyWfr9vKR6sK+fi7QraW7qdtG+NnvbswcXBPzu2bQErsMcE+TRERERERERFp5Zqzx/H13sePa2yfBcz0/vthIAJ4BogFvgBGOud2+cWPB54CFnrX3wV+U73TObfDzEZ421gOFAOPAo/5xSwzs8uA+4B7ge+Bcc65L47sEkVEDt3W0n18srqID1cV8vm6Lewtr6Jj+7YM65PAiH6JnH1CPNERRz5esYiIiIiIiIhIQzVb4dg5d9DvUnt7BM/kp0JybTHFQL0zWTnnVgI/P0jMG8AbBzsnEZGmsK6olI9WF7JoVSH/yi3GOUiOiWDcqd0Z0S+J03t1pl3bNsE+TREREREREREJUapKiDSxpUuXMmbMGJKTkzEz5s2bd0DM2rVrueiii4iJieGYY47hlFNOYfXq1b79c+bMISMjg5iYGMyM9evX13m8vXv3MnDgQMyM5cuXB+wzswOWZ599trEuVepRWeX45/rtPPD+as55ZAnDH/uUBz/4jn0Vldx07vH87bdD+fz2DGZd2J+hx8epaCwiIiIiIiIiQdWsk+OJhKLS0lL69+/PxIkTmThx4gH7f/zxR4YMGcLEiRP55JNPiImJ4bvvviMqKsoXs2fPHkaOHMmFF17ILbfcUu/xbr31VlJSUlixYkWt+59//nnOP/9833p0dPRhXpkczJ79FXz2760sWlXIJ98VsX33fsLDPOMV/+eQnpzbN5FuMRHBPk0RERERERERkQOocCzSxEaPHs3o0aMBmDx58gH777zzTkaOHMmjjz7q29a7d++AmJtvvhnggB7ENb3zzjssXryYN954g/fff7/WmJiYGJKSkmrdJ0euaNdePlldxKJVhXy+biv7Kqro1KEtGX7jFXfsoPGKRURERERERKRlU+FYJIiqqqqYP38+d9xxB6NGjSI7O5uePXty6623Mm7cuENqKy8vj1//+td88MEHRETU3Yv1pptu4le/+hW9evXiqquu4tprr6VNGw2LcLicc6wrKuXDVYV8tLqQnI0lOAcpsRGMP6MHI/omclqvzoSH6TUWERERERERkaOHCsciQVRUVERpaSkPPPAAv/vd73jwwQf55JNPuOKKK4iKiuK8885rUDuVlZVcccUVTJ06lYEDB9Y5BvK9995LRkYGUVFRfPzxx0ydOpWtW7dy1113NeJVtQ5vf5XP7IVr2FRSRreYCKZlpjE2PRmAisoqsjcUs8hbLF6/bQ8AA1KimTL8BIb3S6RPUkfMDjonqIiIiIiIiIhIi6TCsUgQVVVVAXDhhRcyZcoUAE4++WSWL1/O008/3eDC8QMPPEC7du18bdTl7rvv9v375JNPprKykvvvv1+F4xre/iqf6W+upKy8EoD8kjLueHMFX+cVs6OsgsXfFVG8p5x2YW0YfGwXrj6rN8P7JpIU3SHIZy4iIiIiIiIi0jhUOBYJori4ONq2bUu/fv0Ctvft25dXX321we18/PHHfPbZZ4SHB46d+7Of/Yxx48bxl7/8pdbnnXHGGezcuZPCwkISExMP/QJaqdkL1/iKxtX2llcx9+8biI4I59w+CQzvl8jPT4gnqr1+jYqIiIiIiIhI66OKh0gQtWvXjtNOO401a9YEbF+7di2pqakNbmfu3Lns3r3bt75p0yYyMzP5y1/+wpAhQ+p8Xk5ODh06dCAmJubQT76V2Vteycr8HWRvKCa/pKzWGAOy7xpOW41XLCIiIiIiIiKtnArHIk2stLSUdevWAZ6hKXJzc8nJyaFz58706NGD2267jUsvvZSzzjqLc845h8WLF/Pqq6/y9ttv+9ooKCigoKCAtWvXArBq1SpKSkro0aMHnTt3plevXgHHjIqKAuDYY48lJSUFgPnz51NQUMDgwYOJiIhg8eLF3HPPPVx77bW0b9++OV6KFqVo516yNxSTvaGY5RuK+XbTDsorHQBhbYzKKnfAc7rFRKhoLCIiIiIiIiIhQYVjkSa2fPlyMjIyfOszZsxgxowZTJo0iXnz5jF27FjmzJnDAw88wE033cTxxx/PSy+9FDC+8bPPPsusWbN869X75s6dy+TJkxt0HuHh4fzhD39gypQpVFVV0bt3b+69915uuOGGxrnQFqyisorvCnbxr9xiX7E4r9jTq7h92zYMTInhqqG9GZQayyk9Yvjs31sDxjgGiAgPY1pmWrAuQURERERERESkWalwLE1u6dKlPPLII2RnZ7Np06Z6i53XXXcdc+bMYfbs2dx6660ArF+//oAetdUefvhhpk2b1qCYal9++SV33nknWVlZmBknnXQS7777LnFxcUd2oXUYNmwYzh3Ye9Xf5MmT6y0Az5w5k5kzZzb4mD179jzgmKNGjWLUqFENbuNotqOsnK/8isQ5G0vYs99TBE7s1J5TUzvzn0N6MSg1ln5dO9GubWAv4rHpyYBnrONNJWV0i4lgWmaab7uIiIiIiIiISGunwrE0udLSUvr378/EiROZOHFinXFvvPEGX375Jd26dQvY3r17dzZv3hyw7a233uKGG27gkksuaXAMwBdffEFmZibTpk3j8ccfp127dnzzzTcHTConRw/nHD9u3U32hmJfj+K1haWAZ8iJvl078v8GpXBKaiyDUmNJjonAzA7a7tj0ZBWKRURERERERCRkqXAsTW706NGMHj0aoM5etRs2bOCmm27io48+4he/+EXAvrCwMJKSkgK2vfnmmwwfPtzXy7ghMQC33HILN9xwA3feeadv2wknnHDY1ybNb295JV9vLCE7t5h/eXsUF+8pB6BTh7ackhrLmIHdOCU1loEpMUS21685EREREREREZFDpYqKBF1FRQWXX345d911F3379j1o/A8//MDHH3/Ma6+9dkgxRUVFZGVlMX78eIYOHcratWtJS0tj5syZnHvuuY1yLdL4Cnb8NIlddm4x3+bvoMI7cV3v+EiG901kkLc38bHxUbRpc/DexCIiIiIiIiIiUj8VjiXoZsyYQVxcHL/+9a8bFP/CCy8QHx/PhRdeeEgxP/zwg+94s2fPJj09nddff53MzEyys7MZOHDgkV2IHLGKyipWb95F9obtZOeW8K8NxeSX+E1i1z2Ga37em1NTY0nvEUvnyHZBPmMRERERERERkdZJhWMJqiVLljBv3jxycnIaFF9RUcHcuXOZNGlSneMS1xVTVVUFeCbg++UvfwlAeno6ixcv5tlnn+WPf/zjEV6NHKqSPfv5KreE7A3FLN+wna837qCs3DOJXVKnDgzqGctVQz2T2PWtZRI7ERERERERERFpGiocS1AtWbKEzZs307VrrtuWkgAAHI5JREFUV9+2yspKbr/9dp544gny8vIC4ufPn09BQQFXX311nW3WFVN9jH79+gVs79evH7m5uUd6KXIQzjm+37LbNy5xdm4x64p+msSuX9dOjDutu2/YiW4xEUE+YxERERERERGR0KXCsQTV9ddfzyWXXBKwLTMzk8svv5xrrrnmgPjnn3+es88+u94J7eqK6dmzJ926dWPNmjUB29euXctJJ510BFchtSnbX8nXeSW+8Yn/lVtMiXcSu+iIcAalxvIf6cmc0iOWgd2jOaadfh2JiIiIiIiIiLQUqtRIkystLWXdunWAZ7iI3NxccnJy6Ny5Mz169CAhISEgPjw8nKSkJNLS0gK25+bmsnDhQl566aU6j1VfjJkxbdo0ZsyYwYABA0hPT+e1117jH//4B08//XQjXGlo21RSFlAkXrVpp28Su2PjIxnZr3oSu870jovUJHYiIiIiIiIiIi2YCsfS5JYvX05GRoZvfcaMGcyYMYNJkyYxb968Brfz4osvEh0dzcUXX3zYMTfffDP79u1j6tSpbNu2jRNPPJEPPvhAE+PV4u2v8pm9cA35JWUk/+MTpmWmMTY9GYDyyipWbdrpG3LiXxuK2bxjLwAdwttwcvcYrju7N4NSY0nvHkusJrETERERERERETmqqHAsTW7YsGE45xocv379+lq3z5o1i1mzZtX73IbE3H777dx+++0NPp9Q9PZX+Ux/c6Vvorr8kjJue2MF763YxK69FXydV8Lecs9kg92iO/jGJa6exC48TJPYiYiIiIiIiIgczVQ4FhGfveWV/Lh1N7Pmf+srGlfbX1nFR6uLGJgSzeWn92BQaiyn9NAkdiIiIiIiIiIirZEKxyIhqGTPfr7fUsq6Is/y/ZbdrCsqZWPxHurrHG7AO78Z2mznKSIiIiIiIiIiwaHvkzeTpUuXMmbMGJKTkzGzA8b2vfvuu+nTpw+RkZHExsZy7rnnsmzZsoCYffv2ceONNxIXF0dkZCRjxowhLy8vICY3N5cLLriAyMhI4uLi+O1vf8v+/ft9+zdv3sz48ePp06cPYWFhTJ48uakuWYLMOUd+SRmfrt3C/3z+I//11koufS6LU+9bxMn3LuLiP2Zx+19X8qesDWwqKeOklGhuPOd4nrw8nYSO7WttU72LRURERERERERCg3ocN5PS0lL69+/PxIkTmThx4gH709LSeOaZZ+jVqxdlZWU8/vjjjBo1in//+98kJiYCnond3nnnHf73f/+XLl26MGXKFM4//3yys7MJCwujsrKS8847jy5duvDZZ5+xbds2Jk2ahHOOp556CvAUn+Pi4rjjjjuYM2dOs74G0jT2V1SxYdtub89hby/iLaX8sGU3e/b/NNxEpw5tOS4hinP6JHBcQhTHxkdxXEIUKbHHENbGAtqsqnIBYxwDRISHMS0zrdmuS0REREREREREgkeF42YyevRoRo8eDVBrL98rr7wyYP2xxx7jxRdfJCcnh8zMTHbs2MGLL77I3LlzGTFiBAAvv/wyqampfPTRR2RmZvLhhx/y7bffsmHDBrp37w7Aww8/zNVXX839999Pp06d6NmzJ08++SQAb7zxRhNesTS2XXvLfUNKVBeIvy8qZcP2PVRW/TS+RLfoDhybEMWlp3bmuIQoX5E4LqodZlbPEX4yNj0ZgNkL15BfUkZyTATTMtN820VEREREREREpHVT4bgF2r9/P3PmzKFTp06cfPLJAGRnZ1NeXs7IkSN9cd27d6dv374sW7aMzMxMsrKy6Nu3r69oDJCZmcm+ffvIzs4mIyOj2a9FDo1zjqJd+/je22vYv0hcuHOfL65tG6NnXCQnJHbkFycleQrE8R3pHR9JZPvG+bEem57M2PRklixZwrBhwxqlTREREREREREROTqocNyCvPfee1x22WXs2bOHrl27smjRIt8wFQUFBYSFhREXFxfwnMTERAoKCnwx1fHV4uLiCAsL88VIy1BRWUXu9j2+HsTVBeLvt5Sya2+FLy6qfVuOjY9kyHFxAcNL9Oh8DOFhGqJcRERERERERESahgrHLUhGRgY5OTls3bqV559/nksvvZSsrCy6du0a7FOTw7RnfwU/1BheYl1RKeu37aa88qfhJRI6tue4hCjGnpwcUCBO7NS+wcNLiIiIiIiIiIiINBYVjluQyMhIjjvuOI477jh+9rOfcfzxx/PCCy9w9913k5SURGVlJVu3biU+Pt73nMLCQs466ywAkpKS+Pvf/x7Q5tatW6msrCQpKalZr6U1ePurfGYvXMOmkjK61TPGr3OO7bv3+yal+75ot/exlPySMl9cG4PULpEcGx/FOX0TOM5bHO4dH0V0RHhzXpqIiIiIiIiIiEi9VDhuwaqqqti3zzOu7aBBgwgPD2fRokWMHz8egLy8PFavXs2ZZ54JwODBg7nvvvvIy8sjJSUFgEWLFtG+fXsGDRoUnIs4Sr39VT7T31xJWXklAPklZUx/cwXbSvfROz4qYHiJdVtKKdlT7ntuh/A2HBsfxak9Y7ksvjvHeieoS+1yDO3bhgXrkkRERERERERERBpMheNmUlpayrp16wBPQTg3N5ecnBw6d+5MTEwMDz/8MBdccAFdu3Zly5YtPPPMM+Tl5XHppZcCEB0dzVVXXcVtt91GQkICXbp0YcqUKQwYMIDhw4cDMHLkSE488UQmTpzIo48+yrZt25g2bRrXXHMNnTp18p1LTk4OADt37qRNmzbk5OTQrl07+vXr18yvSsv10ILvfEXjamXlVfzub6t9610i23FsfBS/6N/VO7xEJMclRNEtOoI2bTS8hIiIiIiIiIiIHL1UOG4my5cvJyMjw7c+Y8YMZsyYwaRJk/jDH/7At99+y//8z/+wbds2unTpwmmnncbSpUsZMGCA7zlPPPEEbdu2Zdy4cZSVlXHuuefy0ksvERbm6cUaFhbG3/72N66//nqGDBlCREQEV1xxBbNnzw44l/T09ID1+fPnk5qayvr165vuBWjB9uyv4Jv8nazIK2FF3g5W5JWwecfeOuNf/9VgjouPIjayXTOepYiIiIiIiIiISPNR4biZDBs2DOdcnfvfeuutg7bRvn17nnrqKZ566qk6Y3r06MF7771Xbzv1nUdrt6+iktWbdwUUidcVlVLlfUm6RndgQEo023fvZ+feigOenxwTwWk9OzfzWYuIiIiIiIiIiDQvFY6l1aqorGJtYSkr80v42lskXlOwi/JKT5W4S2Q7BqRE84v+XRmQEs1JKdEkdOwAHDjGMUBEeBjTMtOCci0iIiIiIiIiIiLNSYVjaRWqqhw/btvNirwSvt64g5X5O/h20w72llcB0LFDWwakRHPV0N4MTIlmQPcYukV3wKz2sYjHpicDMHvhGjaVlNEtJoJpmWm+7SIiIiIiIiIiIq2ZCsdy1HHOkVdc5htqYkXeDr7J38GufZ6hJSLCw+if3Inxp6cysHs0A1JiSO18zCFPWDc2PVmFYhERERERERERCUkqHEuLV7RzL1/n7WBlnmfIiZX5O9i+ez8A7cLa0LdrRy5M78aAlBgGpERzXHwUbcPaBPmsRUREREREREREjl4qHEuLUrx7PyvyfyoSr8groXDnPgDaGJyQ2JHhfRN8ReK0pI60bxsW5LMWERERERERERFpXVQ4lqAp3VfByrwdvsnrVubtIHf7Ht/+3nGRDO7dxVckPrFbNBHtVCQWERERERERERFpaiocS7PYW17Jqs07WbHRMybxivwdfL+lFOc8+5NjIhjYPZrLT+/BwJRo+qdE06lDeHBPWkREREREREREJESpcCyNrryyijUFu1hR3Zt44w7WFu6iospTJY7v2J6BKdFcMKAbA7pHMyA5mi5R7YN81iIiIiIiIiIiIlJNhWOp09tf5TN74RryS8pI/scnTMtMY2x6ckBMZZXjhy2lAZPXrdq8k/0VVQBER4QzICWa6/r09g05kdSpA2YWjEsSERERERERERGRBlDhWGr19lf5TH9zJWXllQDkl5Qx/c0VbC3dR0KnDr4i8bf5O9i93xMT2S6ME5OjmTQ41Vck7tH5GBWJRUREREREREREjjIqHEutZi9c4ysaVysrr+K+v60GoF3bNvTr2olLBqX4isS946MIa6MisYiIiIiIiIiIyNFOhWOp1aaSsjr3vXfjUNKSOhIe1qYZz0hERERERERERESaiyp/UqtuMRG1bk+OiaB/crSKxiIiIiIiIiIiIq2Yqn9Sq2mZaUSEhwVsiwgPY1pmWpDOSERERERERERERJqLhqqQWo1NTwY8Yx3nl5SRHBPBtMw033YRERERERERERFpvVQ4ljqNTU9mbHoyS5YsYdiwYcE+HREREREREREREWkmGqpCRERERERERERERAKocCwiIiIiIiIiIiIiAVQ4FhEREREREREREZEAKhyLiIiIiIiIiIiISAAVjkVEREREREREREQkgArHIiIiIiIiIiIiIhJAhWMRERERERERERERCaDCsYiIiIiIiIiIiIgEUOFYRERERERERERERAKocCwiIiIiIiIiIiIiAVQ4FhEREREREREREZEAKhyLiIiIiIiIiIiISAAVjkVEREREREREREQkgArHIiIiIiIiIiIiIhKgWQvHZvZzM3vXzPLNzJnZ5Br7zcxmmtkmMyszsyVmdmKNmFgze9nMdniXl80spkbMSWb2qbeNfDO7x8ysRszFZrbKzPZ5H/+jyS5cRERERERERERE5CjS3D2Oo4BvgJuAslr23wZMBW4ETgOKgEVm1tEv5hXgFGCUdzkFeLl6p5l1AhYBhd42bgKmAVP8YgYD/wf8BTjZ+/i6mZ3RGBcpIiIiIiIiIiIicjRr25wHc869D7wPYGbz/Pd5ewTfDDzonPurd9skPMXj8cBzZtYXT7F4qHMuyxtzHfCZmaU559YAVwDHAJOcc2XAN2bWB5hiZo8555z3OIudc/d7D3+/mWV4t1/edK+AiIiIiIiIiIiISMvXksY47gUkAR9Wb/AWfpcCZ3o3DQZKgWV+z/s7sLtGzGfe51ZbCHQDevrFfEighX5tiIiIiIiIiIiIiISsZu1xfBBJ3sfCGtsLgWS/mC3eXsMAOOecmRX5PT8JyKuljep9P3ofaztOErUws2uBa72rpWa25qBX07rEAVuDfRISNMp/aFP+Re+B0Kb8hzblP7Qp/6FN+Q9tyn9oU/5DT2pdO1pS4bjFcs7NAeYE+zyCxcyWO+dODfZ5SHAo/6FN+Re9B0Kb8h/alP/QpvyHNuU/tCn/oU35F38taaiKAu9jYo3tiX77CoB473jIgG9s5IQaMbW1QQNiChAREREREREREREJcS2pcPwjnsLtiOoNZtYBOIufxjTOAqLwjFFcbTAQWSPmLO9zq40ANgHr/WJGEGgEgWMni4iIiIiIiIiIiISkZi0cm1mUmZ1sZid7j93Du97DO27xE8DtZnaRmfUH5uGZDO8VAOfcamAB8JyZDTazwcBzwHvOuepxh18B9gDzzKy/mV0E3AE85jc28u+Bc8zsDjPrY2bTgQzv8eVAITtMhwDKf6hT/kXvgdCm/Ic25T+0Kf+hTfkPbcp/aFP+xcf85plr+oOZDQMW17LrT865yd5hJ2YA1wGxwBfADc65b/zaiAWeAsZ4N70L/MY5V+IXcxLwDHA6UAw8C9zrP6memV0C3Af0Br4H7nTOvdlIlyoiIiIiIiIiIiJy1GrWwrGIiIiIiIiIiIiItHwtaYxjEREREREREREREWkBVDhu5czsejP70cz2mlm2mZ1VS4yZ2Qdm5rxDeByszd+b2XJvm+sPEnu8me0ys9IjuAw5TMHKv7fNm83sOzPbZ2abzezBRrgkOQRBzH+mmWV5f/a3mtk7ZnZCI1ySHILGzr+ZDTSz/zWzjWZWZmZrzOw2M2tTI+4kM/vUG5NvZvd4h6KSZhSM/JvZMO/P+2Yz22NmK8zsl01xfXJwwfod4Beve8AgCuL/AboHbAGCmH/dA7YATZD/eDNbaGabvD/XG83sGTOLrhGne8AWoL78m9kSb879l1cb0GYPM5tvZru9P9tPmlk7v/0XmdmHZrbF+/P/hZmNqa9NOXqocNyKmdk4PBMBPgCkA8uAD8ysR43QqUDVITTdBvgT8NJBjt8OeBVYeghtSyMJcv4fBa4Hbgf6AqPR+6BZBSv/ZtYLeAf4zHvc4UAE8P6hnL8cmSbK/yBgCzABOBHPnAR345mAtvq4nYBFQCFwGnATMA2YcrjXIocuWPkHzgRWApcA/YE/AnPMbPzhXYkcriC+B6qPr3vAIApy/nUPGGRBvAfQPWAL0ET5rwLeAi4ATgAmA+cCz/sdV/eALUAD8z8X6Oq3XHeQNsOAvwEdgbOAy/Hc6z3qF3Y28Alwnve47wNv1fahhRyFnHNaWumCZ3LB52ts+zfw337rpwEbgQTAAZccQvu3Auvr2f84nl9Kk4HSYL8eobYEK/9AGlAO9A32axDKSxDzfwlQCYT5bcvwth8X7NclVJamzr9fGw8D2X7rvwZ2AhF+2+4C8vHOq6Cl9ea/jpjXgL8G+zUJtSXY7wHdA4Zm/nUP2DKWIOZf94AtYGnG/P8W2Oy3rnvAFrAcLP/AEuDpQ2zzF3g+POjut+1KYC/QqZ7nfQk8GuzXRMuRL+px3Ep5e3oMAj6ssetDPD2CMLOOwCvAtc65okY+/nnA+cCNjdmuNEyQ838h8AMwysx+MLP1ZvYnM0toxGNIPYKc/3/i+aPxajML8x5nEvBP59zWRjyO1KGZ898JKPZbHwx85pwr89u2EOgG9DyC40gDBTn/hxsjjSjY7wHdAwZXkPOve8AgC3L+dQ8YZM2VfzPrBlwEfOq3WfeAQdaQ/Htd5h1u4lsze8T7nqjPYGC1c26j37aFQHvv8erSEd0DtgoqHLdecUAYnq+K+CsEkrz/fhZY4Jz7oDEP7P2P5HngSuecxrULjqDlH+gNpAKX4elpNAHoA8yvOQ6aNJmg5d85twEYAcwC9gE7gJPwFBGkeTRL/s3sFDw/43/025xUx3Gr90nTC2b+a8acj+errHMO9zhyWIL2HtA9YIsQzN8BugcMvqDlX/eALUKT5t8841zvwdOLeBfwn367dQ8YfA3J/yvAFXi+DfA74GLgrwdpt7bcbsXzDYNac2tmNwApwMsNPHdpwfQfeIgyswnAQDzjDtUV84GZlXqXbw+h+ZeBPzrnvjjS85Sm0cT5b4Pn08cJzrmlzrnP8PzhcDqer0VJkDVl/s0sCXgRzxjIpwHD8NxYvqY/GluGxsi/maXhGevsCefcwW42pQVprvyb2RA8f5z81jn3ZeOcvTSGJn4P6B6whWvi/OsesIVryvzrHrDla4T83wKcgufbBb2BJ5rsZKVJOOfmOOcWOudWOudeBcYBI7wfBh1JDcDHzC4GZgPjvR8oyVGubbBPQJpM9SdAiTW2JwIFeHoA9QNKLXCi0/8zsyzn3FDgajwTGoDna0cNdQ5wtpnN8K4b0MbMKoDrnXPqedT0gpn/zUCFc26t37Z/e8+nB55xl6RpBTP/NwC7nXO3VW8wsyvxjKN2JvD5IbQlh6dJ829mfYDFwKvOuZqTIhXUcdzqfdL0gpn/6piheCZFucc5V2ePZGkywXwP6B4w+IKZf90DBl8w8697wOBr0vw75wq87XxnZtuBz8zsPu8QBroHDL6D5b82y73POR74F7XnvwAYUuN51b2bA9o1s0vwfHg00Tk3/9AvQVoiFY5bKefcfjPLxvN1odf9do3A81WEp4FHajxtJZ4Jr97xtpF/mIc/qcb6hcCdeHobHG6bcgiCnP+/A23N7Fjn3Pfebb3x/MeiTxybQZDzfwyemw9/1evqbdIMmjL/ZtYPz4zJrznnbqklJAt4yMw6OOf2+h13E7D+sC5IDkmQ84+Z/RxPT7QZzjn1RAqCIL8HdA8YZEHOv+4BgyzI+dc9YJA1898A1Tlt733UPWCQNSD/tTkJz+/ozd42ast/FnCXmaU45/L82twHZFcHmdmlwJ+ASc65N47kWqSFaa5Z+LQ0/4Lnawf78Xxq1Bf4PVAKpNYR36AZVYHjgJOBx/D8R3Cyd2lXR/xkNKN2yOQfz01ENp7JEtK9y6fAP4A2wX5dQmUJYv7PwTPr7j14Prk+BVgA5AKRwX5dQmVpivwDJ+IZ3+xVPOOZ+Ra/mGg8PQ9eBfrjmThlJzA12K9JKC1BzP8wYDeeryf6x8QH+zUJtSVY74FanjMZ3QOGTP7RPWCLWIKYf90DtoClifJ/Pp6JDvvjmejuPGAVkOUXo3vAFrDUl3/gWO/P56nePI4GVuPpaRxWT5theD5g+MT7e304ng+Dn/KLuQxPD+WbavyO6Bzs10RLI7yvgn0CWpo4wXA9nk/4qj8N+nk9sQ0tHC3xxtZcetYRPxn90RBS+Qe64vmUcxdQBPwFSAz26xFqSxDzf5n3eKXAFmA+0C/Yr0eoLY2df2BmHbl3NeJOApYCe/H0XpgBWLBfj1BbgpF/YF4dMeuD/XqE4hKs3wE1njMZ3QOGVP7RPWCLWIKYf90DtoClCfI/HE+v0xKgDFgLPATE1ojTPWALWOrKP9Adz4d527z71uEpLB+0uItnuKH3gD3e5z8JtPfbv6SO3xFLgv16aDnyxbxJFhEREREREREREREBNNaQiIiIiIiIiIiIiNSgwrGIiIiIiIiIiIiIBFDhWEREREREREREREQCqHAsIiIiIiIiIiIiIgFUOBYRERERERERERGRACoci4iIiIiIiIiIiEgAFY5FREREREREREREJIAKxyIiIiIiIiIiIiISQIVjEREREZEmYmbrzczVsvQM9rmJiIiIiNRHhWMRERERkaZ1L9DVu4wK8rmIiIiIiDSICsciIiIiIk2nLbDTOVfgnCsAtvnvNLNRZvaZmRWb2XYzW2hmff3219ZbuXqZ7I2ZYmYrzGy3meWb2QtmFtOsVykiIiIirY4KxyIiIiIiTac9sK+e/ZHAE8DpwDBgBzDfzNp593f1WwAu9lv/P++2KuBm4ERgvLetpxrtCkREREQkJJlzLtjnICIiIiLS6piZAfuBq51zf/JuOxX4J9DLObe+ludEAjuBs51zn9fY54AM59ySgxx3FPAOEOGcq2qESxERERGREKQexyIiIiIiTSMJz1AVm+sKMLNjzewVM/vezHYChXju0Xs09CBmdo6ZLTKzPDPbBbwJtPMeX0RERETksKhwLCIiIiLSNE70Pn5XT8x7QDxwHXAGkA5U4Cn8HpSZpQJ/A1YD/w8YBPzSu7tBbYiIiIiI1KZtsE9ARERERKSVGgHkO+dya9tpZl2APsD1zrnF3m2ncGj36KfiKRDf4pyr9LZx/hGdtYiIiIgIKhyLiIiIiDQq78R2I4AbgJfNzH/IiC7ex3hgE7AVuMbMNgLJwGw8PY4b6t94vkV4s5m9CfwMz0R5IiIiIiJHRENViIiIiIg0rjPxDEERCfwKzxjH1csCb8yX3rhxwADgG+AZ4G5gX0MP5JxbAdwETAFWAVcDtzbGRYiIiIhIaDPnXLDPQURERESk1TCzYcA851zPemLWA5Odc0ua56xERERERA6NehyLiIiIiDSu/cCWg8Rs8caJiIiIiLRI6nEsIiIiIiIiIiIiIgHU41hEREREREREREREAqhwLCIiIiIiIiIiIiIBVDgWERERERERERERkQAqHIuIiIiIiIiIiIhIABWORURERERERERERCSACsciIiIiIiIiIiIiEuD/A0AGUjeuLuzGAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 1728x1440 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "D9dg3lMgJ8QG",
        "colab_type": "text"
      },
      "source": [
        "# Сохраняем прогноз"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uhEYxJJyEDZL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "ans = Matrix_confirmed[\"Среднее\"].stack().reset_index()\n",
        "ans_Death = Matrix_Death[\"Среднее\"].stack().reset_index()\n",
        "ans[\"prediction_deaths\"] = ans_Death[0]\n",
        "ans.columns = ['date','region','prediction_confirmed', 'prediction_deaths']\n",
        "ans['date'] = pd.to_datetime('2020-'+ans['date'])\n",
        "ans['prediction_confirmed'] = ans['prediction_confirmed'].astype('int')\n",
        "ans['prediction_deaths'] = ans['prediction_deaths'].astype('int')\n",
        "ans.sort_values(by=['region', 'date']).to_csv('submission.csv', index=False)\n",
        "ans.sort_values(by=['region', 'date']).to_csv('/content/sberbank-covid19-forecast-2020/submission/submission.csv', index=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PB_eo5GDtiGh",
        "colab_type": "text"
      },
      "source": [
        "# Считаю ошибку на новых данных для всех участников"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d6qtU9Ah3xZt",
        "colab_type": "text"
      },
      "source": [
        "## Загружаем правильные данные"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0UFcaU82l8Js",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data2 = data2[data2.days_mart<0.59]\n",
        "new = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/time.csv\").iloc[:,:4]\n",
        "new.columns =  [\"countries\",\"iso_alpha3\",\"date\",\"confirmed\"]\n",
        "new[\"date\"] = pd.to_datetime(new[\"date\"])\n",
        "new[\"days_mart\"] = (new.date - pd.to_datetime(\"2020-03-1\")).dt.days\n",
        "new[\"days_mart\"] = new[\"days_mart\"]/100\n",
        "new[\"confirmed\"] = pd.to_numeric(new[\"confirmed\"])+1\n",
        "new[\"confirmed\"] =  np.log10(new[\"confirmed\"])\n",
        "new[\"date\"] = new[\"date\"].apply(lambda x: pd.Series(x.strftime(\"%m-%d\")))\n",
        "data2 = pd.concat([data2, new],ignore_index=False)\n",
        "data2 = data2.sort_values(by=['iso_alpha3','date'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "na9bzNnM9oLk",
        "colab_type": "text"
      },
      "source": [
        "## Функции"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i9c7Dn2WChQs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_new(Country, Matrix, Death = False,subplot_x = 2, days_before = 10, point = True):\n",
        "  dat2 = data2.copy()\n",
        "  if log_labels:\n",
        "    dat2.deaths = 10**dat2.deaths - 1\n",
        "    dat2.confirmed = 10**dat2.confirmed - 1\n",
        "  else:\n",
        "    dat2.deaths = 10**3*dat2.deaths\n",
        "    dat2.confirmed = 10**4*dat2.confirmed  \n",
        "  yyy = dat2.loc[(dat2.iso_alpha3 == Country)&(dat2.confirmed>0)&(dat2.days_mart>=(days_x - days_before)/sc_days)]\n",
        "  if Death:\n",
        "    if point:\n",
        "      plt.plot(yyy.date, yyy.deaths, label = \"Истинное значение\", marker='o')\n",
        "      ax = plt.gca()\n",
        "      for x,y,i in zip(yyy.date,yyy.deaths,range(0,len(yyy))):\n",
        "        if not i % ((subplot_x * (days_before+days_prog))//10):\n",
        "          plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(-5,10), ha=\"right\", va = \"bottom\")\n",
        "    plt.plot(yyy.date, yyy.deaths, label = \"Истинное значение\")\n",
        "    plt.gca().set(xlabel='Дата', ylabel='Смерти')\n",
        "  else:\n",
        "    if point:\n",
        "      plt.plot(yyy.date, yyy.confirmed, label = \"Истинное значение\", marker='o', linewidth = 5)\n",
        "      ax = plt.gca()\n",
        "      for x,y,i in zip(yyy.date,yyy.confirmed,range(0,len(yyy))):\n",
        "        if not i % ((subplot_x * (days_before+days_prog))//10):\n",
        "          plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(-5,10), ha=\"right\", va = \"bottom\")\n",
        "    else:\n",
        "      plt.plot(yyy.date, yyy.confirmed, label = \"Истинное значение\", linewidth = 5)\n",
        "    plt.gca().set(xlabel='Дата', ylabel='Заражения')\n",
        "  Matrix2 = {}\n",
        "  for j, name in enumerate(Matrix):\n",
        "    yyy2 = Matrix[name].loc[:,[Country]]\n",
        "    Matrix2[pred_score3(yyy,yyy2)+\" \"+name] = Matrix[name]\n",
        "  Matrix = OrderedDict(sorted(Matrix2.items(), key=lambda t: t[0]))  \n",
        "  for j, name in enumerate(Matrix):\n",
        "    yyy2 = Matrix[name].loc[:,[Country]]\n",
        "    if point:\n",
        "      plt.plot(yyy2, label = str(j+1)+\") \" + name, marker='o')\n",
        "      ax = plt.gca()\n",
        "      for x,y,i in zip(yyy2.index,yyy2.values.T[0],range(0,len(yyy2))):\n",
        "        if not i % ((subplot_x * (days_before+days_prog))//10):\n",
        "          plt.annotate(int(y), (x,y), textcoords=\"offset points\", xytext=(5,-10), ha=\"left\", va = \"top\")\n",
        "    else:\n",
        "      plt.plot(yyy2, label = str(j+1)+\") \" + name)\n",
        "    \n",
        "    plt.title(dic_countries[Country])\n",
        "    plt.xticks(np.arange(0, (days_before+days_prog+1), ((days_before+days_prog)//10)*subplot_x))\n",
        "    #plt.yticks(np.linspace(0,10+days_prog,30//subplot_x))\n",
        "    plt.grid(True)\n",
        "    plt.legend()\n",
        "    \n",
        "def pred_score2(predictions, day):\n",
        "  yyy = data2[(data2.days_mart==(57 + day)/sc_days) & (data2.confirmed > 0)].sort_values(by=['iso_alpha3'])\n",
        "  region_list_x = data_x.iso_alpha3.to_list()\n",
        "  yyy = yyy[yyy.iso_alpha3.isin(region_list_x)]\n",
        "  #print(len(yyy),len(predictions))\n",
        "  if len(yyy)==len(predictions):\n",
        "    labels = 10**yyy.confirmed.values  - 1\n",
        "    return np.mean(np.abs(np.log10((predictions+1)/(labels+1))))\n",
        "    #return np.sum(np.abs(predictions - labels))\n",
        "  else:\n",
        "    return 0    \n",
        "\n",
        "def pred_score3(yy,yy2):\n",
        "  ers = 0\n",
        "  for i in ['04-27','04-28','04-29','04-30','05-01','05-02','05-03']:\n",
        "    yyy = yy[yy.date == i]\n",
        "    yyy2 = yy2[yy2.index == i]\n",
        "    er = np.abs(np.log10((yyy2.values[0,0]+1)/(yyy.confirmed.to_list()[0]+1)))\n",
        "    #er = np.abs(yyy2.values[0,0] - yyy.confirmed.to_list()[0])\n",
        "    ers += er\n",
        "  return str(round(ers/85,5))\n",
        "  #return str(round(ers))\n",
        "def open_sub(name):\n",
        "  ers = 0\n",
        "  data = pd.read_csv(\"/content/sberbank-covid19-forecast-2020/submission/\"+name+\".csv\").sort_values(by=['region','date'])\n",
        "  data222 = data2[['date','iso_alpha3','confirmed']]\n",
        "  data222.confirmed = 10**data222.confirmed - 1\n",
        "  data222.date = \"2020-\" + data222.date\n",
        "  submission = data.merge(data222, how = 'left', left_on=('date','region'), right_on=('date','iso_alpha3'))\n",
        "  submission = submission[submission.confirmed>0]\n",
        "  submission = submission.loc[submission.region.str.contains('-')]\n",
        "  #submission = submission.loc[submission.region != \"RU-NEN\"]\n",
        "  for i in ['2020-04-27','2020-04-28','2020-04-29','2020-04-30','2020-05-01','2020-05-02','2020-05-03']:  \n",
        "    sub = submission[submission.date == i]\n",
        "    er = np.mean(np.abs(np.log10((sub.confirmed.values+1)/(sub.prediction_confirmed.values+1))))\n",
        "    #er = np.sum(np.abs(sub.confirmed.values-sub.prediction_confirmed.values))\n",
        "    ers += er\n",
        "    #print(i, er)\n",
        "  #print(\"Общая\", ers)\n",
        "  data = data[data.region.str.contains('-')]\n",
        "  data = data[(data.date >=\"2020-04-27\") & (data.date <=\"2020-05-03\")]\n",
        "  data = pd.DataFrame(data[\"prediction_confirmed\"].values.reshape((85,-1)).T,columns = labels,index = data_list)\n",
        "  return data, round(ers,3)   "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YuqpF8Wa955m",
        "colab_type": "text"
      },
      "source": [
        "## Графики в img.zip"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XVagSZFLvVRp",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "days_prog = 7\n",
        "data_x = data_x[(data_x.iso_alpha3.str.contains('-'))]\n",
        "labels = data_x.sort_values(by=['iso_alpha3']).iso_alpha3\n",
        "data_list = pd.date_range('2020-04-27', periods = days_prog, freq ='d')\n",
        "data_list = data_list.strftime('%#m-%#d') \n",
        "#niw = pd.DataFrame(pd.read_csv(\"/content/niw.csv\").sort_values(by=['region','date'])[\"prediction_confirmed\"].values.reshape((85,-1)).T,columns = labels,index = data_list)\n",
        "Matrix = {}\n",
        "Matrix[\"Среднее\"] = pd.DataFrame(np.zeros((days_prog, len(labels))),columns = labels,index = data_list)\n",
        "members = [\"niw\",\"tyz910\",\"zfturbo\",\"SG_da \",\"topspin26\",\"ksetherg\", \"therealroman\",\"solemn_leader\",\"LastEmperor\",\"supertoha\",\"macroslav\",\"Shapovalov\",\"submission\"]\n",
        "#members = [\"tyz910\",\"zfturbo\",\"SG_da \",\"topspin26\",\"solemn_leader\"]\n",
        "for member in members:\n",
        "  x, ers =  open_sub(member)\n",
        "  Matrix[member + \" (\" + str(ers) + \")\"] = x\n",
        "  Matrix[member + \" (\" + str(ers) + \")\"][\"RUS\"] = Matrix[member + \" (\" + str(ers) + \")\"].sum(axis=1)\n",
        "  Matrix[\"Среднее\"] += Matrix[member + \" (\" + str(ers) + \")\"]\n",
        "\n",
        "Matrix[\"Среднее\"] = Matrix[\"Среднее\"]//(len(Matrix) - 1)\n",
        "scores = []\n",
        "for i,(index, row) in enumerate(Matrix[\"Среднее\"].iterrows()):\n",
        "  score = pred_score2(np.abs(row),i)\n",
        "  if score:\n",
        "    scores.append(score)\n",
        "#print(\"Среднее\", sum(scores), scores) \n",
        "Matrix[\"Среднее\"][\"RUS\"] = Matrix[\"Среднее\"].sum(axis=1) \n",
        "Matrix[\"Среднее (\"+str(round(sum(scores),3))+\")\"] = Matrix[\"Среднее\"].copy()\n",
        "del Matrix[\"Среднее\"]\n",
        "\n",
        "from collections import OrderedDict\n",
        "Matrix = OrderedDict(sorted(Matrix.items(), key=lambda t: t[0]))\n",
        "\n",
        "# Количество графиков в строке\n",
        "subplot_x = 1\n",
        "# Отмечать точки\n",
        "point = True\n",
        "\n",
        "#За сколько дней до начала прогноза\n",
        "days_before = 3\n",
        "\n",
        "\n",
        "!mkdir img\n",
        "plt.rcParams['figure.figsize'] = [20, 10]\n",
        "plt.rcParams['font.size'] = 14\n",
        "labels = labels.tolist()\n",
        "labels.append(\"RUS\")\n",
        "for i in labels:\n",
        "  fig, ax = plt.subplots()\n",
        "  plot_new(i, Matrix, death_bool, subplot_x, days_before, point)\n",
        "  fig.savefig(\"/content/img/\"+i+'.png')\n",
        "  plt.clf()\n",
        "!zip -r img.zip img"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kPpDAwWbeqaV",
        "colab_type": "text"
      },
      "source": [
        "# Мусор параметры"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zDoQ95bb1Bdo",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "param_list =['countries', 'confirmed', 'deaths', 'iso_alpha3', 'delta_conf', 'delta_deaths', 'region',\n",
        "              'pred_conf', 'pred_deaths', 'delta_conf1', 'delta_deaths1',\n",
        "             \"delta_conf_count\",\n",
        "              \"isolation\",\n",
        "             'testpop',\n",
        "             'Health_GDP',\n",
        "             'Physicians', \n",
        "             'urban_pop_rate',             \n",
        "             'hospibed',\n",
        "             'Nurse',\n",
        "              'pred_conf-population_log',              \n",
        "             'population_log',\n",
        "             'tests',\n",
        "             'population',\n",
        "             'gatheringlimit',\n",
        "             'world_share',\n",
        "             'fertility_rate', \n",
        "             'migrants', \n",
        "             'land_area',\n",
        "             'healthperpop',\n",
        "             'Age_old/new', \n",
        "             'Health_USD', \n",
        "             'median_age',             \n",
        "             'Smoking', \n",
        "             'density',\n",
        "             'pred_recovered',\n",
        "             'recovered',\n",
        "             'days',\n",
        "             'days_10',\n",
        "             'days_100',\n",
        "             'days_1000',\n",
        "             'days_1_1000',\n",
        "             'days_after_Quarantine', \n",
        "             'delta_conf2', \n",
        "             'delta_deaths2',\n",
        "             'delta_conf3', \n",
        "             'delta_deaths3',\n",
        "             'delta_conf4', \n",
        "             'delta_deaths4',\n",
        "             'delta_conf5', \n",
        "             'delta_deaths5',\n",
        "             'delta_conf6', \n",
        "             'delta_deaths6',\n",
        "             \"delta_conf_sum\",\n",
        "             'days_mart']"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}