Dimensionality_reduction_with_statistical_model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Dimensionality reduction with statistical model.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "rK9kW_7xHORf",
        "kpVHuEf5hUTU",
        "4Km6do3bvRKw",
        "cWEMb1ZjvUTq",
        "Pph6i9sHvurx",
        "x_5iiCDpvBag",
        "yQTM0ZTP-fNQ",
        "G9wEr6hlxUC8",
        "P_QU-17t4Opz",
        "r3H2ftvS3m1e",
        "DPeC32Ai38Ck"
      ]
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "verBJk4CIwf6"
      },
      "source": [
        "\n",
        "\n",
        "*   Data: Resized 12-band or RGB images (resized ratio: 0.5, 0.5, 1)\n",
        "*   Dimensionality reduction: PCA and LLE\n",
        "*   Statistical models: Linear Regression and SVR\n",
        "*   Methods:\n",
        "      * Resized images are used to predict cases directly\n",
        "      * Resized images first do PCA then use statistical models to predict cases\n",
        "      * Resized images first do LLE then use statistical models to predict cases\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rK9kW_7xHORf"
      },
      "source": [
        "# Mount Drive"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7i7zUCFHbp80",
        "outputId": "eb50adcb-4e40-4145-9f5a-9dac41e79748"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kpVHuEf5hUTU"
      },
      "source": [
        "# Setup Environment and Load Functions"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gTCYitoTuicA"
      },
      "source": [
        "### Install library"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FPSXSZi1E0Lm",
        "outputId": "bea7cdb3-bf54-4a2c-dbfd-5063fb42d90a"
      },
      "source": [
        "!pip install scikit-learn==0.24"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting scikit-learn==0.24\n",
            "  Downloading scikit_learn-0.24.0-cp37-cp37m-manylinux2010_x86_64.whl (22.3 MB)\n",
            "\u001b[K     |████████████████████████████████| 22.3 MB 1.3 MB/s \n",
            "\u001b[?25hRequirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.24) (1.1.0)\n",
            "Requirement already satisfied: scipy>=0.19.1 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.24) (1.4.1)\n",
            "Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.24) (3.0.0)\n",
            "Requirement already satisfied: numpy>=1.13.3 in /usr/local/lib/python3.7/dist-packages (from scikit-learn==0.24) (1.19.5)\n",
            "Installing collected packages: scikit-learn\n",
            "  Attempting uninstall: scikit-learn\n",
            "    Found existing installation: scikit-learn 1.0.1\n",
            "    Uninstalling scikit-learn-1.0.1:\n",
            "      Successfully uninstalled scikit-learn-1.0.1\n",
            "Successfully installed scikit-learn-0.24.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Nf6xQA_1nLAv"
      },
      "source": [
        "# sklearn library\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "from sklearn.metrics import mean_absolute_error\n",
        "from sklearn.decomposition import PCA\n",
        "from sklearn.manifold import LocallyLinearEmbedding\n",
        "from sklearn.metrics import mean_absolute_percentage_error\n",
        "from sklearn import linear_model  \n",
        "from sklearn.svm import SVR"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cTsy3_Iq21K-"
      },
      "source": [
        "from random import randint, randrange\n",
        "from skimage import io\n",
        "from skimage.transform import rescale, resize, downscale_local_mean\n",
        "\n",
        "import skimage\n",
        "import cv2\n",
        "import os\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import pickle\n",
        "import plotly.express as px\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "gyQb5j_zCjoy",
        "outputId": "72962baf-3ebe-4cde-a938-6219c0a40552"
      },
      "source": [
        "skimage.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'0.18.3'"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KlpIqYTQDeKu",
        "outputId": "e1a5b951-89d2-46f2-b25b-c8564d423d7f"
      },
      "source": [
        "!pip install epiweeks\n",
        "from epiweeks import Week, Year\n",
        "from datetime import date"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Collecting epiweeks\n",
            "  Downloading epiweeks-2.1.3-py3-none-any.whl (5.9 kB)\n",
            "Installing collected packages: epiweeks\n",
            "Successfully installed epiweeks-2.1.3\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iHu-vQ3Dus7s"
      },
      "source": [
        "### Get epi weeks"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zn-z5IIbxNFq"
      },
      "source": [
        "def getEpiWeek(origin_str):\n",
        "  \"\"\"Get epi week from string\n",
        "  \"\"\"\n",
        "  date_ls = origin_str.split('-')\n",
        "  return Week.fromdate(date(int(date_ls[0]), int(date_ls[1]), int(date_ls[2])))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OZv16vrjuwXf"
      },
      "source": [
        "### Load CSV file"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n1Q7rP1Jdn5S"
      },
      "source": [
        "def loadStructuredData(csv_path):\n",
        "  df = pd.DataFrame()\n",
        "  if os.path.isdir(csv_path):\n",
        "    for filename in os.listdir(csv_path):\n",
        "      file_path = os.path.join(csv_path, filename)\n",
        "      df = df.append(pd.read_csv(file_path))\n",
        "  elif os.path.isfile(csv_path):\n",
        "    df = pd.read_csv(csv_path)\n",
        "  else:\n",
        "    print('Error: Not folder or file')\n",
        "  return df"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ri0pWvK-u0Hw"
      },
      "source": [
        "### Plot graphs "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jveGkUZr99De"
      },
      "source": [
        "def plotData(x_asix, y_axis, title):\n",
        "  plt.figure(figsize = (12, 8))\n",
        "  plt.plot(x_axis, y_axis, color ='black')\n",
        "  plt.ylabel('Dengue Cases')\n",
        "  plt.title(title)\n",
        "  plt.tick_params(axis='both',labelsize=3)\n",
        "  plt.setp(plt.gca().get_xticklabels(), rotation=30, horizontalalignment='right')\n",
        "\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M6IdJ8O33K7s"
      },
      "source": [
        "def plotDistribution(corr_list, X_train, y_train):\n",
        "  sorted_corr_list = list(corr_list)\n",
        "  sorted_corr_list.sort(reverse=True)\n",
        "\n",
        "  max_index = corr_list.index(sorted_corr_list[0])\n",
        "  sec_max_index = corr_list.index(sorted_corr_list[1])\n",
        "\n",
        "  color_list = []\n",
        "  max_cases = max(y_train)\n",
        "\n",
        "  for i in y_train:\n",
        "    color = [i / max_cases, 0, 0]\n",
        "    color_list.append(color)\n",
        "\n",
        "  color_list = np.array(color_list)\n",
        "\n",
        "  plt.subplot(1, 2, 1)\n",
        "  plt.scatter(X_train[:,max_index], X_train[:,sec_max_index], c=color_list)\n",
        "  plt.subplot(1, 2, 2)\n",
        "  plt.scatter(np.full(y_train.shape, 0, dtype=int), y_train, c=color_list)\n",
        "  plt.tight_layout()\n",
        "  plt.show()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "S0SyMfFGu3i8"
      },
      "source": [
        "### Image"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZvyE1Bzo7HKL"
      },
      "source": [
        "def readImg(img_path, resize_ratio=None):\n",
        "  img = io.imread(img_path)\n",
        "\n",
        "  if resize_ratio:\n",
        "    img_rescale = rescale(img, resize_ratio, anti_aliasing=True)\n",
        "\n",
        "  print(os.path.basename(img_path), '(origin shape:', img.shape, '-> rescale:', str(img_rescale.shape) + ')')\n",
        "  return img_rescale"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "N5EYXGzvu6Kp"
      },
      "source": [
        "### Split traing and testing set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wcq2BFi38MW5"
      },
      "source": [
        "def splitTrainTestSet(ratio):\n",
        "  # Split the data into training (ratio) and testing (1 - ratio)\n",
        "  train_val_ratio = ratio\n",
        "  train_num = int(len(info_dict['Image']) * train_val_ratio)\n",
        "\n",
        "  # Change list to array\n",
        "  origin_dimension_X = np.array(info_dict['Image'])\n",
        "  labels = np.array(info_dict['cases_medellin'])\n",
        "\n",
        "  print(''.center(60,'-'))\n",
        "\n",
        "  origin_X_train = origin_dimension_X[:train_num,:,:,:]\n",
        "  y_train = labels[:train_num]\n",
        "  origin_X_test = origin_dimension_X[train_num:,:,:,:]\n",
        "  y_test = labels[train_num:]\n",
        "\n",
        "  print('Total number of weeks:'.ljust(30), len(origin_dimension_X), 'weeks')\n",
        "  print('Training input:'.ljust(30), origin_X_train.shape)\n",
        "  print('Training output:'.ljust(30), y_train.shape)\n",
        "  print('Testing input:'.ljust(30), origin_X_test.shape)\n",
        "  print('Testing output:'.ljust(30), y_test.shape) \n",
        "\n",
        "  return origin_X_train, y_train, origin_X_test, y_test"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Qk6IPZ-Xu9rF"
      },
      "source": [
        "### Load csv and image data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_nucXPOKEo_b"
      },
      "source": [
        "# Load data from one of the source\n",
        "def loadData(csv_folder, img_folder, option=None, resize_ratio=None):\n",
        "  if option is None:\n",
        "    # Get data by combining from csv and images\n",
        "    df = loadStructuredData(csv_folder)\n",
        "    info_dict = combineData(img_folder, df, resize_ratio)\n",
        "    \n",
        "    print(len(info_dict['LastDayWeek']), len(info_dict['Image']), len(info_dict['cases_medellin']))\n",
        "\n",
        "  else:\n",
        "    # Load data from previous pickle file\n",
        "    info_dict = loadDataFromPickle(option)\n",
        "  return info_dict"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GPhprhk73Yj1"
      },
      "source": [
        "def combineData(img_folder, df, resize_ratio=None):\n",
        "  info_dict = {'LastDayWeek':[], 'cases_medellin':[], 'Image':[], 'epi_week':[]}\n",
        "  img_list = os.listdir(img_folder)\n",
        "\n",
        "  for index, row in df.iterrows():\n",
        "    name = row['LastDayWeek']\n",
        "    week_df = str(getEpiWeek(name))\n",
        "    case = row['cases_medellin']\n",
        "    for img_name in img_list:\n",
        "      \n",
        "      # If image name is image_2017-12-24.tiff -> get 2017-12-24\n",
        "      # Reference Links: https://www.w3schools.com/python/ref_string_join.asp, \n",
        "      #                  https://stackoverflow.com/questions/13174468/how-do-you-join-all-items-in-a-list/13175535\n",
        "      new_img_name = ''.join(i for i in img_name if i.isdigit() or i == '-')      \n",
        "\n",
        "      week_img = str(getEpiWeek(new_img_name))\n",
        "\n",
        "      if week_df == week_img:\n",
        "        img_path = os.path.join(img_folder, img_name)\n",
        "        img = readImg(img_path, resize_ratio)\n",
        "\n",
        "        info_dict['Image'].append(img)\n",
        "        info_dict['LastDayWeek'].append(name)\n",
        "        info_dict['cases_medellin'].append(case)\n",
        "        info_dict['epi_week'].append(week_df)\n",
        "        break\n",
        "\n",
        "  return info_dict"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iLc9X_o2vDww"
      },
      "source": [
        "### Dimensionality reduction - PCA"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aJg4PwES7L1T"
      },
      "source": [
        "def dimension_reduct_with_PCA(origin_X_train, origin_X_test, y_train):\n",
        "  print(' PRINCIPAL COMPONENT ANALYSIS  '.center(100, '='))\n",
        "\n",
        "  reshape_X_train = origin_X_train.reshape(origin_X_train.shape[0], -1)\n",
        "  reshape_X_test = origin_X_test.reshape(origin_X_test.shape[0], -1)\n",
        "\n",
        "  pca = PCA(n_components=0.95) \n",
        "  pca_X_train = pca.fit_transform(reshape_X_train)\n",
        "\n",
        "  pca_X_test = pca.transform(reshape_X_test)\n",
        "  print('Origin shape'.ljust(15), reshape_X_train.shape)\n",
        "  print('Resize shape'.ljust(15), pca_X_train.shape)  \n",
        "\n",
        "  return pca_X_train, pca_X_test"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Aioj_XnOSpaw"
      },
      "source": [
        "def pca_per_band(origin_X_train, origin_X_test, y_train):\n",
        "  pca_X_train = None\n",
        "  pca_X_test = None\n",
        "  for band in range(origin_X_train.shape[3]):\n",
        "    pca_band_train, pca_band_test = dimension_reduct_with_PCA(origin_X_train, origin_X_test, y_train)\n",
        "    if pca_X_train is None and pca_X_test is None:\n",
        "      pca_X_train = pca_band_train\n",
        "      pca_X_test = pca_band_test\n",
        "    else:\n",
        "      pca_X_train = np.append(pca_X_train, pca_band_train, axis = 1)\n",
        "      pca_X_test = np.append(pca_X_test, pca_band_test, axis = 1)\n",
        "  \n",
        "  print(pca_X_train.shape, pca_X_test.shape)\n",
        "  return pca_X_train, pca_X_test"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4Km6do3bvRKw"
      },
      "source": [
        "### Dimensionality reduction - LLE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e77g9RiC6EHn"
      },
      "source": [
        "def dimension_reduct_with_LLE(origin_X_train, origin_X_test, y_train):\n",
        "  print(' LOCALLY LINEAR EMBEDDING '.center(100, '='))\n",
        "\n",
        "  reshape_X_train = origin_X_train.reshape(origin_X_train.shape[0], -1)\n",
        "  reshape_X_test = origin_X_test.reshape(origin_X_test.shape[0], -1)\n",
        "\n",
        "  embedding = LocallyLinearEmbedding(n_components=2, neighbors_algorithm='kd_tree')\n",
        "  lle_X_train = embedding.fit_transform(reshape_X_train, y_train)\n",
        "  lle_X_test = embedding.transform(reshape_X_test)\n",
        "\n",
        "  return lle_X_train, lle_X_test"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cWEMb1ZjvUTq"
      },
      "source": [
        "### Model - LR"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NEF-AWAf5ka7"
      },
      "source": [
        "def classified_with_LR(origin_X_train, origin_X_test, y_train, y_test):\n",
        "  print('[Linear Regression]'.center(100, '-'))\n",
        "\n",
        "  reshape_X_train = origin_X_train.reshape(origin_X_train.shape[0], -1)\n",
        "  reshape_X_test = origin_X_test.reshape(origin_X_test.shape[0], -1)\n",
        "\n",
        "  regr = linear_model.LinearRegression()\n",
        "  regr.fit(reshape_X_train, y_train)\n",
        "\n",
        "  float_y_pred = regr.predict(reshape_X_test)\n",
        "  int_y_pred = [int(i) for i in float_y_pred]\n",
        "\n",
        "  print('Predicted')\n",
        "  print(' '.ljust(3, ' '), 'List =', int_y_pred)\n",
        "  print(' '.ljust(3, ' '), 'Mean =', round(np.mean(int_y_pred), 4))\n",
        "  print('')\n",
        "\n",
        "  print('Real')\n",
        "  print(' '.ljust(3, ' '), 'List =', y_test)\n",
        "  print(' '.ljust(3, ' '), 'Mean =', round(np.mean(y_test), 4))\n",
        "  print('')\n",
        "\n",
        "  MAE = get_MAE_score(y_test, int_y_pred)\n",
        "  MAPE = get_MAPE_score(y_test, int_y_pred)\n",
        "\n",
        "  print('- MAE: ', str(MAE).rjust(8), '(cases different in average)')\n",
        "  print('- MAPE:', str(MAPE).rjust(8), '(times different in average)')\n",
        "\n",
        "  return MAE, MAPE"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Pph6i9sHvurx"
      },
      "source": [
        "### Model - SVR"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QqTl3o547XQ6"
      },
      "source": [
        "def classified_with_SVR(origin_X_train, origin_X_test, y_train, y_test):\n",
        "  print('[SVR]'.center(100, '-'))\n",
        "\n",
        "  reshape_X_train = origin_X_train.reshape(origin_X_train.shape[0], -1)\n",
        "  reshape_X_test = origin_X_test.reshape(origin_X_test.shape[0], -1)\n",
        "\n",
        "  regressor = SVR(C=1.0, epsilon=0.2)\n",
        "  regressor.fit(reshape_X_train, y_train)\n",
        "\n",
        "  float_y_pred = regressor.predict(reshape_X_test)\n",
        "  int_y_pred = [int(i) for i in float_y_pred]\n",
        "\n",
        "  print('Predicted')\n",
        "  print(' '.ljust(3, ' '), 'List =', int_y_pred)\n",
        "  print(' '.ljust(3, ' '), 'Mean =', round(np.mean(int_y_pred), 4))\n",
        "  print('')\n",
        "\n",
        "  print('Real')\n",
        "  print(' '.ljust(3, ' '), 'List =', y_test)\n",
        "  print(' '.ljust(3, ' '), 'Mean =', round(np.mean(y_test), 4))\n",
        "  print('')\n",
        "  \n",
        "  MAE = get_MAE_score(y_test, int_y_pred)\n",
        "  MAPE = get_MAPE_score(y_test, int_y_pred)\n",
        "  \n",
        "  print('- MAE: ', str(MAE).rjust(8), '(cases different in average)')\n",
        "  print('- MAPE:', str(MAPE).rjust(8), '(times different in average)')\n",
        "\n",
        "  return MAE, MAPE"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x_5iiCDpvBag"
      },
      "source": [
        "### Evaluation metrics"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eFu2p9ERQohZ"
      },
      "source": [
        "def calCorrelation(data, label):\n",
        "  corr = np.corrcoef(data, label)[0, 1]\n",
        "  return round(corr, 4)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Fr76lZu5d-ZB"
      },
      "source": [
        "def get_MAPE_score(y_true, y_pred):\n",
        "  \"\"\"Get Mean Absolute Percentage Error (MAPE)\n",
        "  \n",
        "  Calculate the MAPE score based on the prediction. \n",
        "  The lower MAPE socre is, the better the predictions are.\n",
        "\n",
        "  \"\"\"\n",
        "  return round(mean_absolute_percentage_error(y_true, y_pred), 4)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bJb-HT_4dxmh"
      },
      "source": [
        "def get_MAE_score(y_test, y_pred):\n",
        "  return round(mean_absolute_error(y_test, y_pred), 4)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yQTM0ZTP-fNQ"
      },
      "source": [
        "# 12-band Images Run Here"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IkaBRXTRhMdI",
        "outputId": "cee920b5-b996-440f-83b7-96f6f1514870"
      },
      "source": [
        "# Define main folder\n",
        "dataset_folder = '/content/drive/MyDrive/Dengue forecasting with Satellite Images/dataset'\n",
        "\n",
        "# Define image folder\n",
        "image_name = 'image/image_12_band'\n",
        "img_folder = os.path.join(dataset_folder, image_name)\n",
        "\n",
        "# Define csv folder\n",
        "csv_name = 'csv/merge_cases_timeseries.csv'\n",
        "csv_folder = os.path.join(dataset_folder, csv_name)\n",
        "info_dict = loadData(csv_folder, img_folder, resize_ratio=(0.5, 0.5, 1))\n",
        "\n",
        "print('INFO_DICT'.center(50, '-'))\n",
        "print('keys:', info_dict.keys())\n",
        "print('')\n",
        "\n",
        "print('DENGUE CASES'.center(50, '-'))\n",
        "print('Max weekly dengue cases:', max(info_dict['cases_medellin']))\n",
        "print('Min weekly dengue cases:', min(info_dict['cases_medellin']))\n",
        "print('')\n",
        "\n",
        "print('WEEKS'.center(50, '-'))\n",
        "print('Max week:', max(info_dict['LastDayWeek']))\n",
        "print('Min week:', min(info_dict['LastDayWeek']))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "image_2015-11-01.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-11-08.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-11-15.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-11-22.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-11-29.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-12-06.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-12-13.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-12-20.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2015-12-27.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-01-03.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-01-10.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-01-17.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-01-24.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-01-31.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-02-07.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-02-14.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-02-21.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-02-28.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-03-06.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-03-13.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-03-20.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-03-27.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-04-03.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-04-10.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-04-17.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-04-24.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-05-01.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-05-08.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-05-15.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-05-22.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-05-29.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-06-05.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-06-12.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-06-19.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-06-26.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-07-03.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-07-10.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-07-17.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-07-24.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-07-31.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-08-07.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-08-14.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-08-21.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-08-28.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-09-04.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-09-11.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-09-18.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-09-25.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-10-02.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-10-09.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-10-16.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-10-23.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-10-30.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-11-06.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-11-13.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-11-20.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-11-27.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-12-04.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-12-11.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2016-12-18.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-01.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-01.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-08.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-15.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-22.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-01-29.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-02-05.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-02-12.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-02-19.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-02-26.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-03-05.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-03-12.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-03-19.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-03-26.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-04-02.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-04-09.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-04-16.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-04-23.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-04-30.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-05-07.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-05-14.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-05-21.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-05-28.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-06-04.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-06-11.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-06-18.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-06-25.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-07-02.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-07-09.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-07-16.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-07-23.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-07-30.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-08-06.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-08-13.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-08-20.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-08-27.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-09-03.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-09-10.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-09-17.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-09-24.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-10-01.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-10-08.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-10-15.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-10-22.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-10-29.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-11-05.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-11-12.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-11-19.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-11-26.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-12-03.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-12-10.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "image_2017-12-17.tiff (origin shape: (1205, 765, 12) -> rescale: (602, 382, 12))\n",
            "112 112 112\n",
            "--------------------INFO_DICT---------------------\n",
            "keys: dict_keys(['LastDayWeek', 'cases_medellin', 'Image', 'epi_week'])\n",
            "\n",
            "-------------------DENGUE CASES-------------------\n",
            "Max weekly dengue cases: 592\n",
            "Min weekly dengue cases: 14\n",
            "\n",
            "----------------------WEEKS-----------------------\n",
            "Max week: 2017-12-23\n",
            "Min week: 2015-11-07\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HXXLV5Z59YWw",
        "outputId": "4f75339c-279f-43be-e08b-951293e437ea"
      },
      "source": [
        "print(info_dict['epi_week'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['201544', '201545', '201546', '201547', '201548', '201549', '201550', '201551', '201552', '201601', '201602', '201603', '201604', '201605', '201606', '201607', '201608', '201609', '201610', '201611', '201612', '201613', '201614', '201615', '201616', '201617', '201618', '201619', '201620', '201621', '201622', '201623', '201624', '201625', '201626', '201627', '201628', '201629', '201630', '201631', '201632', '201633', '201634', '201635', '201636', '201637', '201638', '201639', '201640', '201641', '201642', '201643', '201644', '201645', '201646', '201647', '201648', '201649', '201650', '201651', '201701', '201701', '201702', '201703', '201704', '201705', '201706', '201707', '201708', '201709', '201710', '201711', '201712', '201713', '201714', '201715', '201716', '201717', '201718', '201719', '201720', '201721', '201722', '201723', '201724', '201725', '201726', '201727', '201728', '201729', '201730', '201731', '201732', '201733', '201734', '201735', '201736', '201737', '201738', '201739', '201740', '201741', '201742', '201743', '201744', '201745', '201746', '201747', '201748', '201749', '201750', '201751']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6KMN9fBm-O4R"
      },
      "source": [
        "## Data Distrubtion Visualization"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EdXTQWwKWDQ4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 985
        },
        "outputId": "e8d493df-f198-46ab-8905-7ebdc5415ad4"
      },
      "source": [
        "x_axis = info_dict['LastDayWeek']\n",
        "y_axis = info_dict['cases_medellin']\n",
        "plotData(x_axis, y_axis, 'Images & Cases Intersection')\n",
        "\n",
        "dataset_folder = '/content/drive/MyDrive/Dengue forecasting with Satellite Images/dataset'\n",
        "csv_name = 'csv/merge_cases_timeseries.csv'\n",
        "csv_folder = os.path.join(dataset_folder, csv_name)\n",
        "df = loadStructuredData(csv_folder)\n",
        "\n",
        "x_axis = df['LastDayWeek']\n",
        "y_axis = df['cases_medellin'].astype(int)\n",
        "plotData(x_axis, y_axis, 'Cases')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAscAAAHkCAYAAADSJoIrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeXjU5b3//9edBQhLCFtC9qDIqhUICQWlQC2iotVjD2q1ot2sX1tqOfUcLafWpWoPatVTtQfBX60bFSouta64sAgoIUEQSHBhyQqEJSSEkPX+/TEzacg6WWY+k8nzcV25Qj7LfN6TuLy4877v21hrBQAAAEAKcboAAAAAIFAQjgEAAAA3wjEAAADgRjgGAAAA3AjHAAAAgBvhGAAAAHAjHAMAgooxZpEx5mmn6wDQPRGOAfiNMWafMeY7TtfRGcaY24wxRcaYEmPMR8aYCC/uSTfGvOW+56gxZrMx5of+qLctxpg1xpifeHntX40x9/m6pvYwxsw0xuQ3PGatfcBa69V7AoDGCMcA4CVjzBhJ90m6UNJQSfdIqmvjnqmSPpS0VtJISUMk/T9JF/u02ABkjAlzugYAaAvhGIAjjDE3GmM2GGMedY+o7jHGTHMfzzPGHDLG3NDg+rnGmK3GmFL3+bsbvd58Y8x+Y8wRY8ydDUepjTEhxpg7jDFfu8+vNMYMdp/rY4x5wX28xBiTYYyJaaHsGkm1kvZba2ustWustZVtvNWHJD1rrV1srT1sXTKttVe5nz/IGPNPY0yxMeaY+88Jjb5Pe4wxZcaYvcaY6xqc+5ExJtt937vGmGT3ceP+vh5yf78+N8ac7cXPZKYxJt8Y82v3vUWeEW5jzE2SrpP0X8aYE8aYN9zH44wxq9z17zXG/LLB691tjHnZ/f0tlXSjexR9i7uug8aYRxpc/01jzEb3z2GbMWZmg3ODjTHPGGMK3e/3NWNMP0lvS4pz13TCXc/dxpgXGtz7XWPMTvfrrjHGjG1wbp/7twHbjTHHjTErjDF92vpeAQhehGMATpoiabtco6nLJb0kKU2uEdYfSHrCGNPffW25pPmSoiTNlfT/jDFXSJIxZpykP8sV3mIlDZQU3+A5CyRdIWmGpDhJxyQ96T53g/v6RHcdN0uqaKHeQ+6Pl70JUMaYvpKmSnq5lctCJD0jKVlSkvvZT7jv7yfpT5IuttYOkDRN0mfuc5dLWiTpSknDJK2X9Df3a14o6VuSRrnf21WSjrRVr9tw/ev792NJTxpjBllrl0p6UdKD1tr+1trLjDEhkt6QtM19/QWSfmWMmdPg9S53v/8o9/3/K+l/rbWRks6UtNL9fuIlvSnXyPxgSbdJWmWMGeZ+necl9ZU0XlK0pEetteVyjcAXumvqb60tbPhmjDGj3N+XX7m/T29JesMY06vBZVdJukjSCEnfkHSjl98rAEGIcAzASXuttc9Ya2slrZAroN5rra201r4nqUquoCz3KO3n1to6a+12uQLPDPfr/LukN6y1H1trqyT9TpJt8JybJf23tTbfPdJ7t6R/d/+av1quUDzSWlvrHtUtbaHelZKWSvpS0muegOweGV3QzPWD5PrvbFFL3wBr7RFr7Spr7UlrbZmk+xu8L8nVtnG2MSbCWltkrd3Z4D39wVqbba2tkfSApAnu0eNqSQMkjZFk3Ne0WEMj1XL9DKqttW9JOiFpdAvXpkkaZq2911pbZa3dI2mZpGsaXLPJWvua++dW4X79kcaYodbaE9baT9zX/UDSW9bat9zXrpa0RdIlxphYuULwzdbaY+7a1nr5fq6W9Ka1drW1tlrSw5Ii5PqLhsefrLWF1tqjcoX9CV6+NoAgRDgG4KSDDf5cIUnW2sbH+kuSMWaKcU2AKzbGHJcrHA51XxcnKc9zk7X2pE4fKU2W9Kr71+olkrLlao+IkWtE8l1JL7l/Zf+gMSa8caHGmNGSzpcrXC2QdFSugOwZHf6wmfd3TK5wG9vSN8AY09cY85S7JaRU0jpJUcaYUPfI6NXu91pkjHnTuPqePe/pfxu8p6OSjKR4a+2Hco0+PynpkDFmqTEmsqUaGjniDtseJ+X+GTQjWa6WhpIGdSyS6/vqkdfonh/LNaKd425hubTBa81r9Frny/W9S5R01Fp7zMv30FCcpP2eL6y1de6aGv5m4UCDP7f2fgH0AIRjAN3Fckn/kJRorR0oaYlcYVByjcw27NONkGs02CNPrtaEqAYffay1Be5RyHustePkGk28VK72jcbCJIXKNRJbJ1c7Rq2krZKyG4zo1nOH9E2SvtfK+/q1XCOzU9ytBt/yvA33a7xrrZ0tV0jMkWtk1vOeftboPUVYaze67/uTtTZV0ji5wuh/tlKDt2yjr/PkGv1vWMMAa+0lLd1jrf3SWvt9uVojFsvVotLP/VrPN3qtftba/3GfG2yMifKipsYK5Qreklz92HKF7QIv3i+AHohwDKC7GCDX6OEpY0y6pGsbnHtZ0mXGNaGvl1xtE6bB+SWS7m8wYW2Yu2dXxphZxphzjDGhkkrl+rV/cytQ5MjVTvFnY8xASeGSVssVPE+4Q1dz/kuuiWj/aYwZ4n7mucaYlxq8rwpJJcY1SfAuz43GmBhjzOXu8FgpV4uDp7Ylkn5jjBnvvnagMWae+89p7pH2cLl6tU+18J7a66CkMxp8vVlSmTHmdmNMhDEm1BhztjEmraUXMMb8wBgzzP0XjBL34TpJL8j1M5zjfp0+xjVBMMHdEvK2XN/7QcaYcGOM5y8RByUNcf9MmrNS0lxjzAXu78ev5fpebuzoNwFAcCMcA+gubpF0rzGmTK6e4pWeE+5R2wVyTegrkitEHpIrBEmuSWD/kPSe+/5P5JoMKLkmoL0sVzDOlmvJtecbP9zdF32pXBPLvpZr5PF8SamSJsk1kawJ90jut90fe4wxR+XqW37LfcljcvXAHnbX9U6D20Mk/Ydco59H5epF/n/u131VrpHXl9ztGDv0r+XhIuUaYT4mV0vBEblWzeis/0/SOHfbw2sNvicTJO11v4en5ZrQ15KLJO00xpyQ6+dyjbW2wlqbJ9fkvUWSiuUaLf5P/ev/U9fL9ReXHLl+tr+SJGttjlz953vcdcU1fJi1drdc/cyPu+u7TNJl7t50AGjCWNvWb6QAoHtxr3BRIuksa+1ep+sBAHQfjBwDCArGmMvck9v6yTVp7nNJ+5ytCgDQ3RCOAQSLy+VqPyiUdJZcv67nV2MAgHahrQIAAABwY+QYAAAAcCMcAwAAAG5hThfgMXToUJuSkuJ0GQAAAAhymZmZh621w5o7FzDhOCUlRVu2bHG6DAAAAAQ5Y8z+ls7RVgEAAAC4EY4BAAAAN8IxAAAA4EY4BgAAANwIxwAAAIAb4RgAAABwIxwDAAAAboRjAAAAwI1wDAAAALgRjgEAAAA3wjEAAADgFuaLFzXG/Jukb0j6TFKdpBBr7eu+eBYAAADQVXw1cpwoyUo6JanS/bkJY8xNxpgtxpgtxcXFPioFAAAA8I6vwvEXkvZL6tPgowlr7VJr7WRr7eRhw4b5qBQAAADAOz5pq7DWvuOL1wUAAAB8iQl5AAAAgBvhGOhBPvjgAxUWFjpdBgAAAYtwDPQQtbW1mjt3rh566CGnSwEAIGARjoEeori4WJWVldqzZ4/TpQAAELAIx0APUVBQIEnav3+/w5UAABC4CMdAD+HpNd63b5+zhQAAEMAIx0AP4Rk5Pn78uI4fP+5wNQAABCbCMdBDNFylgtYKAACaRzgGegjPyLFEawUAAC0hHAM9RGFhoZKSkiQxcgwAQEsIx0APUVBQoHPPPVcRERGEYwAAWkA4BnqIgoICJSQkKDk5mbYKAABaQDgGeoBTp07p6NGjiouLU3JyMiPHAAC0gHAM9ACelSri4+OVkpLCyDEAAC0gHAM9gCcce0aODx8+rPLycoerAgAg8BCOgR7As4xbfHy8kpOTJUm5ublOlgQAQEAiHAM9QOO2Com1jgEAaA7hGOgBCgoK1KdPH0VFRdWPHDMpDwCApgjHQA9QWFio+Ph4GWMUGxur8PBwRo4BAGgG4RjoAQoKChQXFydJCgkJUVJSEiPHAAA0g3AM9AAFBQWKj4+v/5q1jgEAaB7hGAhy1loVFhbWjxxLYq1jAABaQDgGglxJSYkqKiqajBwXFRWpsrLSwcoAAAg8hGMgyDVcxs3Ds5wbax0DAHA6wjEQ5DwbgDRsq2A5NwAAmkc4BoJccyPHhGMAAJpHOAaCnGfkODY2tv5YQkKCQkNDmZQHAEAjhGMgyBUWFmrw4MGKiIioPxYWFqb4+HhGjgEAaIRwDAS5hhuANJScnMzIMQAAjRCOgSDXeAMQj5SUFEaOAQBohHAMBLnGG4B4JCcnq6CgQDU1NQ5UBQBAYCIcA0GspqZGBw4caHHkuLa2Vvn5+Q5UBgBAYCIcA0Hs0KFDqqurazYcs5wbAABNEY6BINbcBiAehGMAAJoiHANBrLkNQDySkpIkiRUrAABogHAMBLHWRo579+6t2NhYRo4BAGiAcAwEscLCQoWGhio6OrrZ86x1DADA6QjHQBArKCjQ8OHDFRoa2ux5J9Y6Li0tVVlZmV+fCQCAtwjHQBBraQMQj+TkZOXm5qqurs5vNV199dW6/vrr/fY8AADag3AMBLHCwsJWw3FKSoqqq6tVVFTkl3qstdq8ebN27drll+cBANBehGMgiBUUFDQ7Gc/D38u5HT58WEePHlVeXp6stX55JgAA7UE4BoLUyZMnVVJS0mZbheS/5dyys7MlSadOndLRo0f98kwAANqDcAwEKc8ax4E0cuwJx5LYthoAEJAIx0CQam0DEI9+/fpp6NChfgvHOTk59X/Oy8vzyzMBAGgPwjEQpFrbAKQhf651nJ2drZiYGEmMHAMAAhPhGOhGrLXNfjTHm5Fjyb9rHefk5GjmzJkKDQ1l5BgAEJAIx0A3cffddys0NFQhISGnffTq1UtLly5tcn1BQYH69u2ryMjIVl83OTlZ+/fv9/nqEeXl5dq/f7/Gjx+vuLg4Ro4BAAEpzOkCALRt/fr1uvfee3XxxRcrPT39tHPvvfeeFi5cqDlz5tRPsJP+tQGIMabV105JSVFFRYUOHTpU3/LgC7t375YkjR07VgkJCYRjAEBAIhwDAa68vFw33nijRowYoRUrVqh///6nnb/xxhs1btw4LViwQK+//np9GG5rAxCPlJQUSa4VK3wZjj2T8caOHavExERt3brVZ88CAKCjaKsAAtztt9+uvXv36plnnmkSjCVXW8Q999yjN954Q6+99lr98bY2APHwhGNfT8rLzs5WSEiIRo4cWT9yzEYgAIBAQzgGAtiHH36oJ598Urfeequ+9a1vtXjdrbfeqnPPPVcLFixQWVmZrLVejxz7a63jnJwcnXnmmerdu7cSEhJUUVGhY8eO+fSZAAC0F+EYCFClpaX60Y9+pLPOOkv3339/q9eGh4frqaeeUmFhoe68804dPXpUlZWVXo0cR0ZGatCgQX4ZOR4zZowkKTExURJrHQMAAg/hGAhQt912m/Ly8vTss8+qb9++bV4/ZcoU3XzzzXr88cf1xhtvSGp7GTePlJQUn4bjmpoaffHFFxo7dqwkKSEhQRJrHQMAAg/hGAhA7777rpYtW6bbbrtNU6dO9fq+Bx54QNHR0VqwYIGktjcA8fB1ON67d6+qq6sJxwCAgEc4BgLMqVOn9JOf/ETjxo3TPffc0657o6Ki9Nhjj+nEiROS2j9y7KsJctnZ2ZJU31YRGxvLRiAAgIBEOAYCzNatW5Wfn697771Xffr0aff9V111lS666CKFhYUpNjbWq3tSUlJ08uRJHT58uN3P80bjcBwaGqrY2FhGjgEAAYdwDASYrKwsSa4e4o4wxuhvf/ubPvzwQ/Xu3durexqudewLOTk5Gj58uKKiouqPJSYmEo4BAAGHcAwEmMzMTA0bNszrlojmREVFafr06V5f7+u1jrOzs+v7jT0SEhJoqwAABBzCMRBgsrKylJqa2ua2z13Js9axL8KxtVY5OTnNhmM2AgEABBrCMRBATp06pZ07d2rSpEl+fe7AgQMVFRXlk3B84MABHT9+vL7f2CMxMVEnT55kIxAAQEAhHAMB5PPPP1dNTY1SU1P9/mxfLefmmYzX3MixxHJuAIDAQjgGAkhmZqYk+X3kWPJdOM7JyZGkZkeOJcIxACCwEI6BAJKVlaXBgwfX9wD7k6/WOs7OztaAAQOaTDD0jBwzKQ8AEEgIx0AAyczM1KRJk/w6Gc8jJSVF5eXlOnr0aLvvPXz4sMrKypo9l5OTozFjxjR5T8OHD1dISAgjxwCAgOKTcGyMucEYs8AYc7kx5jJjzOW+eA4QTKqqqvT555870m8sdXw5N2utzj//fF144YWqra1tcj47O7tJS4UkhYWFKS4ujnAMAAgovho5PiQpQtIpSZXuz00YY24yxmwxxmwpLi72USlA97Bjxw5VV1c70m8sdXw5t82bN2v37t365JNP9NRTT512rrS0VAUFBU0m43mw1jEAIND4KhyXSMqR1KfBRxPW2qXW2snW2snDhg3zUSlA9+DZGa+7jRwvX75cvXv31vnnn6/f/OY3Kioqqj+3e/duSU0n43l41joGACBQ+CQcW2s3WWv/Ya193fPZF88BgklmZqYGDhyoM844w5HnR0VFaeDAge0KxzU1NVqxYoUuvfRS/eUvf1FlZaUWLlxYf76lZdw8PFtIsxEIACBQMCEPCBBZWVmOTcbzaO9ybh999JEOHjyoa6+9VmeddZYWLVqkFStW6J133pHkmowXFhamM888s9n7ExISVF5erpKSkq4oHwCATiMcAwGgurpa27Ztc6zf2KO94Xj58uWKjIzUJZdcIkm6/fbbNXr0aN1yyy06efKksrOzNXLkSIWHhzd7P2sdAwACDeEYCAC7du1SZWWlY/3GHikpKdq/f79XbQ4VFRVatWqVvve976lPH9e0gt69e2vJkiXau3ev7rvvPmVnZ7fYUiGx1jEAIPAQjoEA4JmMFwgjx2VlZTp27Fib17755psqKyvTtddee9rxmTNn6oYbbtBDDz2kr776qsXJeBJbSAMAAg/hGAgAmZmZ6t+/v8466yxH62jPihXLly9XTEyMZs2a1eTcww8/rMjISNXW1rY6chwbG8tGIACAgEI4BgJAVlaWJk6cqJAQZ/+V9Hat45KSEr355pu65pprFBoa2uT80KFD9eijj8oY02qrSFhYmGJjY2mrAAAEDMIx4LCamhp99tlnjvcbS96PHL/yyiuqqqpq0lLR0Pz583Xo0CGNGzeu1ddirWMAQCAhHAMO2717tyoqKhzvN5Zcax1HRka2GY6XL1+uM888U2lpaa1eN3To0Daf6VnrGACAQEA4BhyWmZkpybmd8RoyxrS5nFtRUZE+/PBDXXfddV2yJrNnC2k2AgEABALCMeCwrKws9e3bV6NHj3a6FEn/Ws6tJStWrJC1Vt///ve75HmJiYkqLy/X8ePHu+T1AADoDMIx4LDMzExNmDCh2YltTvCMHLc0krt8+XJNmjSp1SXa2oPl3AAAgYRwDDiorq5OW7duDYh+Y4+UlBSVlpY2u6Xzl19+qYyMjFYn4rUXG4EAAAIJ4Rhw0BdffKHy8vKA6Df2aG3Fij//+c8KDQ3VNddc02XPYwtpAEAgIRwDDgqUnfEaammt4+LiYi1dulTXXXed4uPju+x5w4cPZyMQAEDAIBwDDrHW6t1331WfPn3aXAvYn1oaOf7Tn/6kiooK3XHHHV36vPDwcA0fPpy2CgBAQCAcAw6orq7Wz372Mz333HO68cYbFRYW5nRJ9QYNGqQBAwacFo5LS0v1xBNP6N/+7d9a3Q66o1jrGAAQKAjHgJ+Vlpbqsssu07Jly7Ro0SI9+eSTTpd0Gs9axw2Xc1uyZIlKSkr0m9/8xifP9Kx1DACA0wjHgB/l5+dr+vTpev/99/X000/r/vvvV0hI4P1r2HAjkIqKCj3yyCOaPXu2Jk+e7JPnJSYmshEIACAgBN7/lYEg9dlnn2nKlCnat2+f3n77bf34xz92uqQWNQzHzzzzjA4ePKhFixb57HkJCQkqLy9XaWmpz54BAIA3CMeAD9XV1Wnt2rX60Y9+pGnTpik0NFQff/yxZs+e7XRprUpJSdHx48dVXFysBx98UFOnTtWMGTN89jzWOgYABIrAmQUEBJGvvvpKzz//vJ577jnt27dP/fv31zXXXKP77rtPcXFxTpfXJs+KFX/4wx+0f/9+PfHEEzLG+Ox5nrWO8/LydPbZZ/vsOQAAtIVwDHSxO+64Q4sXL5YxRt/5znd033336YorrlC/fv2cLs1rnrWO//SnP+mcc87R3Llz/fK8hpMAAQBwAuEY6EK5ubn64x//qHnz5umRRx6pbxfobjwjx7W1tfrNb37j01FjSYqNjVV4eDjhGADgOMIx0IUefvhhGWP0xz/+sdsGY0kaPHiw+vfvr5iYGM2bN8/nzwsJCVFSUlKzW1YDAOBPhGOgixw6dEjLli3T9ddfX99D210ZY/TII49o9OjRftugJDk5mZFjAIDjCMdAF3nsscdUWVmp22+/3elSusRPf/pTvz4vOTlZ7777rl+fCQBAYyzlBnSBkpISPfnkk5o3b55GjRrldDndUkpKigoLC1VZWel0KQCAHoxwDHSBP//5zyotLfXZ9so9gWfFCtY6BgA4iXAMdNLJkyf16KOP6pJLLtGECROcLqfb8qyQwaQ8AICTCMdAJz399NM6fPgwo8adxFrHAIBAQDgGOqGqqkoPPfSQpk+frvPPP9/pcrq1+Ph4hYSEEI4BAI5itQqgE1544QXl5+dr2bJlTpfS7YWHhyshIYG2CgCAoxg5BjqotrZWixcv1sSJEzVnzhynywkKrHUMAHAaI8dAB3300Uf64osv9NJLL/l8e+WeIjk5WevWrXO6DABAD8bIMdBB7777rnr16qVLL73U6VKCRkpKigoKClRTU+N0KQCAHopwDHTQ6tWrNW3aNPXr18/pUoJGcnKyamtrVVBQ4HQpAIAeinAMdMDBgwe1bds2zZ492+lSggprHQMAnEY4Bjrggw8+kCRdeOGFDlcSXFjrGADgNMIx0AGrV6/W4MGDNXHiRKdLCSqJiYmSCMcAAOcQjoF2stZq9erVuuCCCxQaGup0OUGlT58+io2Npa0CAOAYwjHQTjk5OSooKKDf2EdY6xgA4CTCMdBOq1evliTCsY+kpKQwcgwAcAzhGGin1atXa+TIkfUrK6BrJScnKy8vT3V1dU6XAgDogQjHQDtUVVVpzZo1jBr7UHJysqqqqnTgwAGnSwEA9ECEY6AdPvnkE504cYJw7EOsdQwAcBLhGGiH1atXKyQkRLNmzXK6lKDFWscAACcRjoF2WL16tdLT0xUVFeV0KUHLE44ZOQYAOIFwDHjp2LFjysjIoKXCx/r166ehQ4cycgwAcAThGPDSRx99pLq6OsKxH7DWMQDAKYRjwEurV69W//799c1vftPpUoIeax0DAJxCOAa8tHr1as2cOVPh4eFOlxL0PCPH1lqnSwEA9DCEY8ALe/fu1ddff01LhZ8kJyeroqJChw8fdroUAEAPQzgGvMCW0f7FWscAAKcQjgEvvP/++4qPj9eYMWOcLqVHYK1jAIBTCMeAF3bt2qW0tDQZY5wupUdgrWMAgFMIx4AX8vPzlZiY6HQZPUZUVJQGDhzIyDEAwO8Ix0AbysrKdPz4cSUkJDhdSo/CWscAACcQjoE2FBQUSBLh2M9Y6xgA4ATCMdCG/Px8SYRjf2OtYwCAEwjHQBsIx85ISUlRaWmpSkpKnC4FANCDEI6BNnjCcVxcnMOV9Cws5wYAcALhGGhDfn6+hg0bpj59+jhdSo9COAYAOIFwDLQhPz+flgoHsEseAMAJhGOgDYRjZwwZMkR9+/Zl5BgA4FeEY6ANhGNnGGOUnJzMyDEAwK/CfPGixpiRks6TVCQpSlKJtfY9XzwL8KWKigodOXKEcOyQESNGKDs72+kyAAA9iK9GjhMlHZNkJJW4PzdhjLnJGLPFGLOluLjYR6UAHccGIM6aO3eucnJytG3bNqdLAQD0EL4Kx0MljZVkJQ1yf27CWrvUWjvZWjt52LBhPioF6DjWOHbW1VdfrfDwcD333HNOlwIA6CF80lZhrf27L14X8DfCsbOGDBmiSy+9VC+++KIWL16ssDCf/CcLAIB6TMgDWuEJx/Hx8Q5X0nPNnz9fBw8e1OrVq50uBQDQAxCOgVbk5+dr0KBB6tevn9Ol9FiXXHKJBg8erGeffdbpUgAAPQDhGGgFy7g5r1evXvr+97+v1157TSUlJU6XAwAIcoRjoBWE48Awf/58VVZW6uWXX3a6FABAkCMcA60gHAeGtLQ0jRkzhlUrAAA+RzgGWlBVVaWDBw8SjgOAMUbz58/X+vXrtWfPHqfLAQAEMcIx0ILCwkJJLOMWKK677joZY/TCCy84XQoAIIgRjoEWsMZxYElKStKsWbP03HPPydpm9xUCAKDTCMdACwjHgWf+/Pn6+uuvtXHjRqdLAQAEKcIx0ALCceC58sor1bdvXybmAQB8hnAMtCA/P18DBgxQZGSk06XAbcCAAbryyiu1YsUKnTp1yulyAABBiHAMtIBl3ALT/Pnzdfz4cb3zzjtOlwIACEKEY6AFhOPANGPGDPXu3Vsff/yx06UAAIIQ4RhoAeE4MPXq1UuTJ09mUh4AwCcIx0AzampqVFRURDgOUNOmTVNmZqYqKyudLgUAEGQIxwh6GzZs0MiRI+s39fDGgQMHVFdXRzgOUFOnTlVVVZWysrKcLgUAEGQIxwh6GzZs0Ndff60lS5Z4fQ/LuAW2qVOnShKtFQCALkc4RtDLzc2VJD311FNe/xqecBzYhg8frhEjRmjTpk1OlwIACDKEYwS93Nxc9e7dW4cOHdLKlSu9uodwHPimTZumjRs3spU0AKBLEY4R9HJzc/Wd73xHo0eP1uOPP+7VPfn5+YqIiNCgQYN8XB06aurUqSoqKqr/zQAAAF2BcIygl5ubq5SUFP3iF79QRkaGPv300zbv8SzjZozxQ4XoiGnTpl3w/88AACAASURBVEmi7xgA0LUIxwhqZWVlOnbsmJKSknTDDTdowIABXo0es8Zx4DvnnHPUr18/+o4BAF2KcIyglpeXJ0lKSkrSgAEDdOONN2rlypU6cOBAq/cRjgNfWFiY0tPTGTkGAHQpwjGCmqcfNSkpSZL0i1/8QtXV1Vq6dGmL99TV1amgoIBw3A1MnTpVn332mcrLy50uBQAQJAjHCGqNw/GoUaM0Z84cLVmyRFVVVc3ec+jQIdXU1BCOu4Fp06aptrZWW7ZscboUAECQIBwjqOXm5io0NFSxsbH1xxYsWKCioiK98sorzd7DMm7dxze/+U1Jou8YANBlCMcIavv371dCQoJCQ0Prj1188cU688wzW5yYRzjuPoYMGaLRo0fTdwwA6DKEYwS13Nzc+pYKj5CQEP385z/Xxo0blZWV1eQewnH3MnXqVG3atKldm4Fs2rRJf/zjH31YFQCguyIcI6g1F44l6Yc//KH69eunu+++u0moys/PV69evTR06FB/lYlOmDZtmg4fPqyvvvrK63vuvvtu3XHHHaqrq/NhZQCA7qjNcGyM6WeMCXH/eZQx5rvGmHDflwZ0Tm1trfLz85sNx1FRUbr33nv1xhtv6IEHHjjtXH5+vuLj4xUSwt8du4OpU6dK8r7vuKSkRB9++KFqamp07NgxX5YGAOiGvPm//zpJfYwx8ZLek3S9pL/6siigKxw4cEA1NTXNhmNJWrhwoa677jrdeeed+uc//1l/nDWOu5dx48YpMjLS677jt99+WzU1NZKkgwcP+rI0AEA35E04Ntbak5KulPRna+08SeN9WxbQeY2XcWvMGKNly5Zp4sSJuvbaa5WTkyOJcNzdhISE1Pcde+P111+v//OhQ4d8VRYAoJvyKhwbY6ZKuk7Sm+5joa1cDwSEtsKxJEVEROjVV19Vnz59dPnll6ukpIRw3A1NnTpVn3/+uUpLS1u9rrKyUm+99ZamTZsmiZFjAEBT3oTjX0n6jaRXrbU7jTFnSPrIt2UBnedNOPacX7Vqlfbs2aPvfve7qqysVHx8vD9KRBeZNm2arLXavHlzq9d99NFHKisr00033SSJcAwAaKrNcGytXWut/a6kx91f77HW/tLnlQGdlJubq4EDByoyMrLNa6dPn67HH39c69evl8Qybt3NlClTZIxps+/4tddeU79+/TRv3jyFhoYSjgEATXizWsVUY8wuSTnur881xvzZ55UBrTh27NhpvaPNaWkZt5bcfPPN9SOKI0aM6FR98K/IyEidffbZrYbjuro6/eMf/9BFF12kvn37Kjo6mp5jAEAT3rRVPCZpjqQjkmSt3SbpW74sCmjL008/rSuuuEL79+9v8Zr2hmNJeuKJJ7RmzRpNnDixsyXCzy644AJ9+OGH+vLLL5s9n5GRoaKiIl1xxRWSpOjoaEaOAQBNeLWQq7U2r9GhWh/UAnht3759ktRqj2lHwnF4eLhmzJghY0xnyoMDbr/9dvXu3Vu33XZbs+dfe+01hYaGau7cuZKkmJgYwjEAoAlvwnGeMWaaJGuMCTfG3CYp28d1Aa3yTLbLyMho9vyJEyd09OjRdodjdF/Dhw/XnXfeqX/84x967733mpx//fXXNWPGDA0aNEgS4RgA0DxvwvHNkn4uKV5SgaQJ7q8Bx7QVjvPyXL/sIBz3LLfeeqvOPPNMLVy4UNXV1fXHd+/erezs7PqWCskVjg8dOtRk+3AAQM/mzWoVh62111lrY6y10dbaH1hrj/ijOKAlnvCbmZmpurq6Jue9XcYNwaV379565JFHtGvXLi1ZsqT+uGfy5uWXX15/LDo6WhUVFTpx4oTf6wQABC5vVqt40BgT6W6p+MAYU2yM+YE/igOac+LECR07dkxjx45VWVmZdu/e3eQawnHPddlll2n27Nn63e9+p8OHD0tyheOJEyee9s9DTEyMJNY6BgCczpu2iguttaWSLpW0T9JISf/py6KA1nhGja+88kpJzbdW5ObmKiQkRHFxcX6tDc4zxujRRx9VWVmZ7rrrLh04cECbNm06raVCIhwDAJrnTTgOc3+eK+nv1trjPqwHaJNnVPjCCy9Uv379WgzH8fHxCgsLa3IOwW/8+PG65ZZbtGTJEj3wwAOy1rYYjlnrGADQkDfh+J/GmBxJqZI+MMYMk3TKt2UBLfOE45SUFKWmprYYjmmp6NnuvvtuRUVF6fHHH1dKSorOOeec085HR0dLYuQYAHA6bybk3SFpmqTJ1tpqSeWSLm/9LsB38vLy6lsm0tPTtXXrVlVVVZ12DeEYgwcP1u9//3tJ0hVXXNFk7ephw4ZJIhwDAE7n7e+c4yR9xxjTp8Gx53xQD9Cm3NxcxcXFKSwsTGlpaaqqqtLnn3+u1NRUSa5tgvPy8jRv3jyHK4XTbrrpJh09elTXX399k3Ph4eEaMmQI4RgAcJo2w7Ex5i5JMyWNk/SWpIslfSzCMRySl5dXPyqclpYmyTUpzxOODx48qOrqakaOobCwMP32t79t8TwbgQAAGvOm5/jfJV0g6YC19oeSzpU00KdVAa3Izc1VYmKiJFff8ZAhQ07rO2YZN3grOjqaCXkAgNN4E44rrLV1kmqMMZGSDklK9G1ZQPOstaeNHBtjlJaWRjhGhzByDABozJtwvMUYEyVpmaRMSVmSNvm0KqAFxcXFqqysPC34pqWlaefOnSovL5dEOIb3CMcAgMa8Wa3iFmttibV2iaTZkm5wt1cAfucJvp62CskVjuvq6rR169b6awYMGKCBA+n+QetiYmJUWlqqU6dYnRIA4NJiODbGzDHG/HvDY9bafZJGGWNm+7owoDme3fEajxxL/9opz7OMW+Olu4DGPGsd03cMAPBobeT4d5LWNnN8jaR7fVIN0IbmRo6HDx+uxMTEJuEYaAtbSAMAGmstHPe21hY3PmitPSypn+9KAlqWl5eniIgIDRky5LTjaWlp2rx5syTCMbxHOAYANNZaOI40xjRZB9kYEy4pwnclAS1rqWUiLS1NX3/9tfLz83X48GHCMbxCOAYANNZaOH5F0jJjTP0osTGmv6Ql7nOA3zVc47ghT9/xq6++KomVKuAdeo4BAI21Fo5/K+mgpP3GmExjTKakvZKK3ecAv2u4xnFDnt3xVq1aJYlwDO9ERERowIABjBwDAOq1uH20tbZG0h3GmHskjXQf/spaW+GXyoBGqqqqVFRU1OzIcVRUlEaNGqX169dLIhzDe6x1DABoyJt1jiustZ+7PwjGcExhYaGstS0GX896x8YYxcfH+7k6dFeEYwBAQ97skAcEhLZ2vvP0HcfFxSk8PNxvdaF7i46OpucYAFCPcIxuo7k1jhtKT0+XREsF2oeRYwBAQ22GY+PyA2PM79xfJxlj0n1fGnA6z+54LYXjCRMmKCwsjHCMdomJidGRI0dUU1PjdCkAgADQ4oS8Bv4sqU7St+XaGa9M0ipJaT6sC2giNzdXQ4YMUd++fZs9HxERoT/84Q+aNGmSnytDdxYTEyNrrYqLixUbG+t0OQAAh3kTjqdYaycZY7ZKkrX2mDGmV1s3GWOmSxoh6YCkKEkl1tr3OlUterSWlnFr6LbbbvNTNQgWDdc6JhwDALzpOa42xoRKspJkjBkm10hyWwoknZJkJJW4P5/GGHOTMWaLMWZLcXGTnaqB07S0AQjQGeySBwBoyJtw/CdJr0qKNsbcL+ljSQ94cV+CXIHYShrk/nwaa+1Sa+1ka+3kYcOGeV81eiTP1tFAVyIcAwAaarOtwlr7ont3vAvkCrtXWGuzvbhvXRfUB0iSSktLdfz4ccIxuhzhGADQUJvh2BiTJOmkpDcaHrPW5vqyMKChtlaqADpqwIAB6t27N+EYACDJuwl5b8rVEmEk9ZFrkt1uSeN9WBdwGk84ZuQYXc0Yo5iYGDYCAQBI8q6t4pyGXxtjJkm6xWcVAc1oawMQoDPYCAQA4NHuHfKstVmSpvigFqBFubm5Cg0NZakt+AThGADg4U3P8X80+DJE0iRJhT6rCGhGXl6e4uPjFRbmTScQ0D7R0dHKzMx0ugwAQADwJmkMaPDnGrl6kFf5phygeaxxDF+KiYlRcXGx6urqFBLS7l+oAQCCiDc9x/f4oxCgNXl5eUpPT3e6DASpmJgY1dTU6NixYxoyZIjT5QAAHORNW8UbarqBx3FJWyQ9Za095YvCAI+6ujrl5eXpe9/7ntOlIEg1XOuYcAwAPZs3vz/cI+mEpGXuj1JJZZJGub8GfOrQoUOqqqpiGTf4THR0tCQ2AgEAeNdzPM1am9bg6zeMMRnW2jRjzE5fFQZ4sMYxfM0zcsxaxwAAb0aO+7t3yZNUv2Nef/eXVT6pCmiANY7ha2whDQDw8Gbk+NeSPjbGfC3XLnkjJN1ijOkn6VlfFgdIjBzD9wYPHqzQ0FDCMQDAq9Uq3jLGnCVpjPvQ7gaT8B7zWWWAW25urvr27atBgwY5XQqCVEhIiIYNG0Y4BgB4NXIsSamSUtzXn2uMkbX2OZ9VBTSQm5urpKQkGWOcLgVBLCYmhp5jAEDbPcfGmOclPSzpfElp7o/JPq4L7bB9+3aNGTOmvje3I26++WbNnTtXZWVlXVhZ18jLy6OlAj7HFtIAAMm7CXmTJZ1nrb3FWrvA/fFLXxcG7/3zn//U7t279cwzz3To/hUrVuipp57SW2+9pTlz5qi0tLTLasvKytKzz3auNZ3d8eAPhGMAgORdON4habivC0HHZWRkSJKee+45Wdt4v5bWHThwQLfccovS09O1cuVKZWRkaM6cOTp+/Hin66qsrNS8efN08803t7suj8OHD+vAgQMaOXJkp+sBWhMdHa2DBw92+J9VAEBw8CYcD5W0yxjzrjHmH54PXxcG72VkZCgqKkp79uzRhg0bvL7PWqubb75Z5eXl+utf/6p58+bp73//uzIzM3XhhReqpKSkU3U99thj2rNnj06dOtXhsL1x40ZJ0rRp0zpVC9CWmJgYnTp1SidOnHC6FACAg7wJx3dLukLSA5L+2OADAaCoqEgFBQX69a9/rX79+um557yfJ/nCCy/o9ddf1/3336+xY8dKkq644gqtWrVKW7du1ezZs3Xs2LEO13Xfffdp4MCB9V93xIYNGxQeHq60tLS2LwY6gbWOAQCSF+HYWrtW0j5J4e4/Z0jK8nFd8JKnpWLWrFn63ve+p5UrV6qioqLN+woKCrRgwQKdd955+tWvfnXaucsuu0yvvvqqtm/fru985zsd6kFetGiRKisr9fDDD0vqXDhOTU1VREREh+4HvEU4BgBI3q1W8VNJL0t6yn0oXtJrviwK3svIyFBoaKgmTpyo+fPn6/jx43rjjTdavcdaq5/+9KeqqqrSM888o9DQ0CbXzJ07Vy+//LKysrL04osvtrumv/71r1q4cKHOP/98SR0Lx5WVldqyZYvOO++8dt8LtFd0dLQkwjEA9HTetFX8XNJ5kkolyVr7paRoXxYF72VkZGj8+PHq27evZs6cqYSEhDZbK/7yl7/o7bff1uLFi3XWWWe1eN2ll16qhIQEffTRR17XY63VrbfeqpiYGP33f/+3YmNjJXUsHGdmZqqyspJwDL/wjBy///772rRpk4qKilRXV+dwVQAAf/MmHFdaa6s8XxhjwiQxnTsAWGuVkZFR348bGhqqH/zgB3rnnXdaHP3as2ePFi5cqJkzZ+rnP/95q69vjNGsWbO0Zs0ar2fw/+1vf9OmTZv0hz/8QZGRkYqMjFRERESHwrFnciGT8eAP0dHRiomJ0f/93/9p2rRpiouLU9++fTVq1Chde+21OnLkiNMlAgD8wJtwvNYYs0hShDFmtqS/S2r99/bwi7179+ro0aOnTVabP3++amtrtXz58ibXnzhxQldccYVCQ0P1l7/8RSEhbf/4Z86cqeLiYu3atavNa8vLy/Vf//VfSk1N1Q033CDJFbBjY2M7FI4//vhjjRw5sn5ED/ClsLAw7d27Vzt27NA///lPPfnkk/rlL3+pc889V6+88oqmTp2qr776yukyAQA+5s320XdI+rGkzyX9TNJbkp72ZVHwjmcyXsNwPHbsWKWlpem5557TwoUL649ba/XDH/5QO3fu1Ntvv60RI0Z49YxZs2ZJkj766CONHz++1WsXL16sgoICrVy58rTg3ZFwbK3Vxo0bdemll7brPqAzIiIiNH78+Cb/rG/YsEGXX365pk6dqtdff53fZgBAEPNmtYo6uSbg3WKt/Xdr7TLLKvkBISMjQ71799Y555xz2vH58+frs88+0/bt2+uPPfDAA3r55Ze1ePFiXXjhhV4/Y8SIEUpOTm6z77i4uFgPPfSQrr322ibBoSPh+IsvvtDhw4fpN0ZAOO+887Rp0yZFRUXp29/+tv7+9787XRIAwEdaDMfG5W5jzGFJuyXtNsYUG2N+57/y0JrNmzdr4sSJCg8PP+34Nddco7CwMD3//POSXNtL33nnnbruuuv061//ut3PmTVrltauXdvq5KRVq1bp1KlTuv3225uc60g49vQbE44RKM466yxt2rRJqampuuqqq/Tggw+ymx4ABKHWRo4XyrVKRZq1drC1drCkKZLOM8YsbOU++EFtba2ysrKa3Rxj6NChmjt3rl544QXt3LlT1157rSZOnKhly5bJGNPuZ82cOVNHjhzRjh07Wrxm5cqVGjNmTJNRbMkVjktLS3Xy5Emvn7lhwwYNHjxYo0ePbne9gK8MHTpUH3zwga6++mrdfvvtevLJJ50uCQDQxVoLx9dL+r61dq/ngLV2j6QfSJrv68LQuuzsbJWXl7e4c9z8+fN14MABnX/++YqIiNCrr77a4Y00GvYdN+fAgQNau3atrrrqqmbDd0eWc9uwYYOmTZvm1aRBwJ/69Omj5cuXa9KkSc1OfAUAdG+tJY9wa+3hxgettcWSwpu5Hn7U3GS8hubOnatBgwbpxIkTevnll5WUlNThZyUlJemMM85oMRyvWrVKdXV1uuqqq5o97wnHBw4c8Op5xcXF2r17d/0GIkCgCQkJ0dy5c/Xpp592eIt1AEBgai0cV3XwHPwgIyNDkZGRGjVqVLPne/furRdffFFvvvmmpk+f3unnzZw5U+vWrWu273jlypXNzvD3aO/I8caNGyXRb4zAdtFFF6murk7vv/++06UAALpQa+H4XGNMaTMfZZKaNpbCrzIyMpSamtpq28HFF1/crpUpWjNr1iwdO3ZM27ZtO+14YWGh1q9f3+KosdT+cLxhwwb16tVLkydP7njBgI+lp6dr4MCBevfdd50uBQDQhVpMVtbaUGttZDMfA6y1tFU4qLKyUtu2bWuxpcIXZs6cKalp3/HLL78sa22r4XjIkCEKCwtrVzhOTU1Vnz59Olwv4GthYWGaPXu23nnnHVatAIAgwmynbmj79u2qrq5Wenq6356ZkJCgkSNHas2aNacdX7lypb7xjW9ozJgxLd4bEhKi4cOHexWOT506pS1bttBSgW5hzpw5Kigo8GoHSQBA90A47obamoznK7NmzdK6detUW1srScrPz9eGDRtaHTX28Hat48zMTFVVVTEZD93CnDlzJInWCgAIIoTjbigjI0PR0dFKTEz063NnzZql48ePa+vWrZJcLRWSujQcezb/YHtedAeJiYkaN26c3nnnHadLAQB0EcJxN5SRkaG0tLQObejRGY37jlesWKGJEyfqrLPOavNeb8Pxxx9/rFGjRmnYsGGdqhXwlzlz5mjdunXt2uQGABC4CMfdTFlZmXbt2uX3lgrJFXBHjx6tNWvWaP/+/frkk0+8GjWWpOHDh6u4uFjV1dUtXmOt1caNG+k3Rrdy0UUXqbKyUmvXrnW6FABAFyAcdzNZWVmy1joSjiVXa8X69ev10ksvSfKupUL613JuBw8ebPGa3bt368iRI4RjdCvTp09Xnz596DsGgCBBOO5mnJqM5zFr1iyVlZXpwQcf1OTJk3XGGWd4dZ83ax17+o2ZjIfuJCIiQjNmzKDvGACCBOG4m8nIyFBycrJjPbkzZsyQJB09etTrUWPJu3CcmZmpgQMHtrjrHxCoLrroIu3evVv79+93uhQAQCcRjruRkpISvf/++5o6dapjNcTExGjcuHGSpHnz5nl9nzfheOfOnRo/frzfJxoCncWSbgAQPAjH3ci9996rY8eO6fbbb3e0jp/97Ge64YYblJKS4vU9MTExMsa0GI6ttfXhGOhuxowZo6SkJForACAIhDldALyTk5Ojxx9/XD/5yU80YcIER2v55S9/2e57wsPDNXTo0BbD8aFDh3TkyBHCMbolY4zmzJmjFStWqLq6WuHh4U6XBADoIEaOu4n/+I//UN++fXXfffc5XUqHtbbW8c6dOyWJcIxua86cOSotLdUnn3zidCkAgE4gHHcDb731lt5++23dddddio6OdrqcDiMcI5hdcMEFCg0Npe8YALo5wnGAq6qq0sKFCzVq1Cj94he/cLqcTmktHO/YsUODBw/W8OHD/VwV0DWioqL0zW9+k75jAOjmCMcB7oknntAXX3yhRx99VL169XK6nE6JjY3VwYMHVVdX1+QcK1UgGMyZM0eZmZk6evSo06UAADqIcBzADh06pHvuuUcXX3yxLrnkEqfL6bTY2FjV1NToyJEjpx1npQoEi2nTpklyrdkNAOieCMcB7Le//a1OnjypRx55xOlSukRLax0XFRWppKSEcIxub+LEiZJc27wDALonwnGA2r59u55++mktWLBAY8aMcbqcLtFSOGYyHoLF4MGDNWLECEaOAaAbIxwHqJUrVyokJES//e1vnS6lyxCO0RNMmjSJkWMA6MYIxwFq3bp1mjRpkgYPHux0KV3GsxJF43C8Y8cODR06tFsvUwd4pKam6uuvv1ZJSYnTpQAAOoBwHIAqKir06aefasaMGU6X0qX69u2ryMjIZkeOzz77bIeqArrWpEmTJElbt251uBIAQEcQjgPQp59+qqqqKn3rW99yupQu13itY2utdu3aRUsFgoYnHNN3DADdE+E4AK1du1bGGE2fPt3pUrpc43Ccn5+v0tJSwjGCxrBhw5SYmEjfMQB0U4TjALRu3Tqde+65ioqKcrqULtc4HDMZD8EoNTWVkWMA6KYIxwGmqqpKmzZtCrp+Yw9POLbWSiIcIzhNmjRJX3zxhUpLS50uBQDQToTjAJORkaGKioqgDscVFRX1oWHHjh2KiYnRkCFDHK4M6DqpqamSpM8++8zhSgAA7UU4DjBr166VpKDsN5aarnXMttEIRp5JefQdA0D3QzgOMGvXrtX48eM1dOhQp0vxiYbhuK6ujpUqEJSGDx+uuLg4+o4BoBsiHAeQmpoabdiwIWhbKqTTw3Fubq7Ky8tZ4xhBiZ3yAKB7IhwHkKysLJWXl/eYcMxkPASz1NRU5eTkqLy83OlSAADtQDgOIJ5+42Dc/MNj4MCB6tOnD+EYQW/SpEmqq6vTtm3bnC4FANAOhOMAsnbtWo0aNUrDhw93uhSfMcbUL+e2Y8cOxcXFBeV6zoBnxQr6jgGgewnz1QsbY2ZISpZ0QFKUpBJr7XuNrrlJ0k2SlJSU5KtSuoXa2lp9/PHHuuqqq5wuxec84fj48eOMGiNoxcXFKTo6mr5jAOhmfDlyvFXScUlGUon782mstUuttZOttZOHDRvmw1IC3/bt23X8+PGg7jf2iI2NVWFhobKzswnHCFrGGHbKA4BuyJfheIGkCZKspEHuz2iBp9+4p4TjL774QhUVFYRjBLVJkyZp165dqqiocLoUAICXfNZWYa2931evHYzWrl2rM844QwkJCU6X4nOxsbH120cTjhHMUlNTVVtbq+3bt2vKlClOlwMA8AIT8gJAXV2d1q9fH9SrVDTkWc5NIhwjuLFTHgB0P4TjALBr1y4dOXKkR7RUSKpfjSMxMVGRkZEOVwP4TlJSkoYMGeJ133FFRYWuv/563X333b4tDADQIp+1VcB7PanfWPrXyDGjxgh2xhivd8o7efKkLr/8cr3//vuSpG984xu68sorfV0iAKARRo4DwNq1a5WYmKiUlBSnS/ELwjF6ktTUVO3YsUOVlZUtXlNeXq7LLrtMH3zwgZYuXaq0tDT9+Mc/1v79+/1YKQBAIhw7zlpb329sTJPV7oJSdHS0fve73+lHP/qR06UAPjdp0iRVV1drx44dzZ4vLy/XpZdeqjVr1ujZZ5/VT3/6U7300kuqq6vT97//fVVXV/u5YgDo2QjHDisqKtKBAwd61Ex2Y4zuuecejRs3zulSAJ9rbae8EydO6OKLL9a6dev0/PPP6/rrr5cknXHGGVq6dKk2bdqku+66y6/1AkBPR8+xwzz/w/TMagcQXEaMGKHBgwfr5ptv1u9//3slJycrJSVFycnJWrNmjT799FO9+OKLuuaaa0677+qrr9YHH3yg//mf/9GsWbM0e/Zsh94BAPQshGOHZWVlyRijCRMmOF0KAB8wxuiVV17Rhx9+qP3792vfvn36+OOP9dJLLyk0NFTLly9vcdv4xx57TBs2bND111+vbdu2KSYmxs/VA0DPYzybMTht8uTJdsuWLU6X4Xff/e539dVXX2nXrl1OlwLAj2pqalRVVaW+ffu2et2OHTuUlpam6dOn65133lFICN1wANBZxphMa+3k5s7xX1mHZWVl1fckAug5wsLC2gzGknT22Wfr4Ycf1urVq7V+/Xo/VAYAPRvh2EEHDx5UQUEB/cYAWuXpR/7kk08crgQAgh/h2EGejQEYOQbQmiFDhujMM8/U5s2bnS4FAIIe4dhBnpUqmIwHoC1TpkzRp59+6nQZABD0CMcOysrK0qhRoxQZGel0KQACXHp6ugoKClRQUOB0KQAQ1AjHDsrMzKTfGIBXPBsF0VoBAL5FOHbI4cOHlZubS78xAK9MmDBB4eHhhGMA8DHCsUM8k/EYOQbgjT59+ujcc8+l7xgAfIxw7BDC8MDNSwAAIABJREFUMYD2Sk9P15YtW1RbW+t0KQAQtAjHDsnMzNQZZ5yhqKgop0sB0E1MmTJFZWVlysnJcboUAAhahGOHsDMegPZKT0+XxKQ8APAlwrEDjh07pj179tBSAaBdRo0apYEDB9J3DAA+RDh2wNatWyWxMx6A9gkJCVF6ejojxwDgQ4RjB3h2xps4caLDlQDobtLT07V9+3adPHnS6VIAICgRjh2QlZWlpKQkDR061OlSAHQzU6ZMUW1tbf1voAAAXYtw7IDMzExaKgB0iGdSHn3HAOAbhGM/Ky0t1ZdffslkPAAdEhMTo+TkZPqOAcBHCMd+xmQ8AJ2Vnp7OyDEA+Ajh2M/YGQ9AZ02ZMkX79u3ToUOHnC4FAIIO4djPMjMzFR8fr5iYGKdLAdBNsRkIAPgO4djPsrKyGDUG0CmTJk1SaGgo4RgAfIBw7EcnTpxQTk4O/cYAOqVfv346++yz6TsGAB8gHPvRtm3bZK1l5BhAp02ZMkWbN2+WtdbpUgAgqBCO/Wjt2rWSWKkCQOelp6erpKREX375pdOlAEBQIRz7yalTp/TEE0/o29/+tuLi4pwuB0A3N2XKFElMygOArkY49pNnn31WRUVFWrRokdOlAAgCY8eOVf/+/ek7BoAuRjj2g5qaGi1evFjp6en69re/7XQ5AIJAaGio0tPTtW7dOqdLAYCgQjj2gxUrVmjv3r1atGiRjDFOlwMgSFx88cXavn27cnNznS4FAIIG4djH6v7/9s48vMriauC/yQIhJGxJWMK+hQBhURAEF1BAUaS2KKJoRYPWfhWVqkXrbitaq4iKGwoKilKkYsvWWkFAUCRAgQIS2XcIEAIEhITknu+PmZtcwk3IfkM4v+fJg8573pkzZ86cOe+8c+/1ePjLX/5C+/btGThwYKDVURSlEnHDDTcAMHv27ABroiiKUnnQ5LiMmT17NuvWreOPf/wjQUFqbkVRSo82bdrQqlUrTY4VRVFKEc3WyhARYfTo0TRv3pwhQ4YEWh1FUSoZxhgGDhzIN998w4kTJwKtjqIoSqVAk+MSMmvWLK666ioWL1581rUFCxaQlJTEY489RkhISAC0UxSlsnPDDTeQkZHB119/HWhVFEVRKgWaHJeQjz/+mIULF3LllVeSmJjIoUOHcq69+OKL1K9fn2HDhgVQQ0VRKjNXXHEFNWvW1KMViqIopYRuZ5aQpKQkBg4cSLt27RgzZgwzZ87klVdeoW3btsyfP59XXnmFsLCwQKupKEolJTQ0lP79+zN79mw8Ho9+tkFRFKWEaBQtAfv372fnzp307t2bv/zlL6xatYq2bduSmJhInz59qF27Nvfdd1+g1VQUpZIzcOBAUlJSWLFiRaBVURRFOe/R5LgEeH+21fszrgkJCSxatIiJEycSERHBk08+SWRkZCBVVBTlAqB///4EBQUxa9asQKuiKIpy3qPJcQlYtmwZwcHBXHTRRTllQUFBJCYmsn//fh555JEAaqcoyoVCVFQUl112mZ47VhRFKQU0OS4BSUlJdOzYkfDw8LOu6S/hKYpSngwcOJDVq1eza9euQKuiKIpyXqPJcTHxeDwkJSXRrVu3QKuiKIpyzl/L27NnD/Pnz+f48ePlqZaiKMp5hybHxWTjxo0cO3Ys57yxoihKIImPj6dly5Z+zx2vWrWKjh070rdvX2rVqsWll17KY489xty5czl27FgAtFUURam4XNDJscfj4e9//ztHjhwp8r3Lli0D0J1jRVEqBPn9Wt7KlSvp06cP1atXZ/r06Tk/SjR27FgGDBhAdHQ0//jHPwKouaIoSsXigk6ON2zYwODBgxk7dmyR701KSiIyMpL4+Pgy0ExRFKXoDBw4kIyMDObNmwfAihUr6Nu3LzVq1GDRokXcfPPNjB49miVLlnDkyBHmzZtHQkIC9957LwcOHAiw9oqiKBWDCzo5bt++PTfddBNjx47l8OHDRbp32bJldO3aleDg4DLSTlEUpWhcfvnl1KhRg1mzZpGUlJRzjGLhwoU0b978DNnw8HD69OnDlClTSE9P57e//S0iEiDNFUVRKg4XdHIM8Oyzz5Kens5rr71W6HtOnTrFmjVr9LyxoigViipVqtC/f39mzJhBv379qFOnDosWLaJZs2b53tOuXTv+/Oc/8+WXXzJ16tTyU1ZRFKWCcsEnxx06dOCWW27hjTfe4NChQ4W6Z/Xq1WRlZel5Y0VRKhwDBw4kLS2NmJgYFi1aRJMmTc55z8MPP0yPHj0YMWIEe/fuLQctFUVRKi4XfHIMdvf4xIkTjBkzplDy3g/j6c6xoigVjZtuuonRo0ezcOFCGjduXKh7goODmTRpEidPnuS+++7T4xWKolzQaHKMfa04ZMgQxo0bx8GDB88pn5SURKNGjYiNjS0H7RRFUQpPtWrVeOKJJ2jUqFGR7ouLi+Oll15i9uzZTJ48uYy0UxRFqfhocux49tlnOXnyJK+88so5ZZctW6ZHKhRFqXQ8+OCDXHHFFTz00EPs3r070OooiqIEBE2OHfHx8dx22228/fbbpKSk5CuXmprKli1b9EiFoiiVjqCgID766COysrL47W9/G2h1FEVRAoImxz4888wznDp1ir/+9a/5yiQlJQH64x+KolROWrZsyRNPPMGcOXPYuHFjoNVRFEUpdzQ59iEuLo477riDd955h3379vmVSUpKIigoiK5du5azdoqiKOVDYmIiwcHBfPTRR4FWRVEUpdzR5DgPTz/9NKdPn+Yvf/mL3+vLli2jXbt2RERElLNmiqIo5UODBg24/vrrmTx5MllZWYFWR1EUpVzR5DgPrVq14q677uKtt97ik08+OeOaiJCUlKTnjRVFqfQkJiayb98+/v3vfwdaFUVRlHJFk2M/vPnmm/Tu3Zthw4YxadKknPKtW7eSmpqq540VRan0DBgwgLp16zJx4sRAq6IoilKuaHLsh/DwcGbNmkWfPn1ITEzkww8/BHI/jKc7x4qiVHZCQ0O58847mT17doHf4KMoilLZKLPk2BhzkzHmGWPMNcaYW4wx15RVW2VBeHg4M2fOpF+/fgwfPpwJEyawbNkywsPDad++faDVUxRFKXOGDx9OVlZWzhGzLVu24PF4AqyVoihK2VKWO8crgSOA8fn3DIwxvzHGrDDGrCjML9OVN9WqVeOf//wn1113Hffeey+ffPIJXbp0ISQkJNCqKYqilDnx8fH07NmTiRMnkp6eTtu2bXn11VcDrZaiKEqZUpbJcSvgKCBAbffvGYjI+yLSVUS6xsTElKEqxScsLIwvv/ySAQMGcPjwYT1vrCjKBUViYiLJycksWLCA06dP8/777+vusaIolRojclbOGhC6du0qK1asCLQa+ZKRkcEbb7zBbbfdRuPGjQOtjqIoSrmQnp5OgwYN6NOnDzNnzgRg4cKF9OrVK8CaKYqiFB9jzEoR8fujFfqBvEJStWpVRo0apYmxoigXFJGRkQwZMoQ5c+bklOk3WCiKUpnR5FhRFEUpkMTERLKzswH7YeW///3vHD16NMBaKYqilA2aHCuKoigF0rNnT9q0aQPA4MGDOXnyJH/7298CrJWiKErZoMmxoiiKUiDGGIYPHw5A165dSUhI0KMViqJUWjQ5VhRFUc7JnXfeSVRUFE2aNGH48OEsX76ctWvXBlotRVGUUkeTY0VRFOWc1KtXj5SUFH7xi19wxx13EBoamvProYqiKJUJTY4VRVGUQhEcHAxAdHQ0N954I5988gmZmZkB1kpRFKV00eRYURRFKTLDhw8nNTU157uPFUVRKguaHCuKoihFpl+/fjRq1EiPViiKUunQ5FhRFEUpMsHBwdx111189dVX7N69O9DqKIqilBqaHCuKoijF4u6778bj8TB58uRAq6IoilJqaHKsKIqiFIsWLVpw9dVXM378eNLS0gKtjqIoSqmgybGiKIpSbJ5//nlSUlK48cYbOXnyZKDVURRFKTGaHCuKoijF5vLLL+eTTz5hyZIl3H777WRnZwdaJUVRlBKhybGiKIpSIm655RZef/11vvzyS0aMGIGIBFolRVGUYhMSaAUURVGU858HH3yQvXv38vLLL9OwYUOeeuqpQKukKIpSLDQ5VhRFUUqFl156iX379vH0009Tv3597rnnnkCrpCiKUmQ0OVYURVFKBWMMEyZM4MCBA9x33300atSI/v37B1otRVGUIqFnjhVFUZRSIzQ0lOnTp9O+fXvuvvtuDh8+HGiVFEVRioQmx4qiKEqpEhERwccff8yhQ4d44IEHAq2OoihKkdDkWFEURSl1OnfuzNNPP81nn33GjBkzAq2OoihKoTEV5St3unbtKitWrAi0GoqiKEopcfr0aS699FJ27drF+vXriYmJCbRKiqIoABhjVopIV3/XdOdYURRFKRNCQ0OZPHkyR44c4f777w+0OoqiKIVCk2NFURSlzEhISOD5559n+vTpTJs2LdDqKIqinBNNjhVFUZQy5Q9/+APdunXjd7/7Hfv37w+0OoqiKAWiybGiKIpSpoSEhDBp0iROnDjB8OHD+fnnnwOtUrmxevVq/UCiopxnaHKsKIqilDlt27bl1VdfZe7cuSQkJPCvf/0r0CoVmzfffJORI0dyrg+0Z2RkcNNNNzFkyBD27t1bTtopilJSNDlWFEVRyoURI0awYMECqlSpwvXXX8/gwYPZs2dPoNUqEtu2bePRRx/ljTfe4IsvvihQ9vXXX2fr1q1kZWUxfvz4ctJQUZSSol/lpiiKopQrGRkZvPrqq7zwwguEhITwwgsvcP/99xMSEhJo1c7JHXfcwYwZM2jevDlpaWls2LCBmjVrniW3b98+4uLiuPrqq8nOzmbFihXs3LmTKlWqBEBrRVHyol/lpiiKolQYqlatypNPPsm6deu4/PLLGTlyJN26dWP58uUF3ufxeJg9ezZTpkxhyZIl7N69m+zs7HLSGlatWsWnn37KyJEjmTRpEvv37+epp57yK/vEE0+QkZHBmDFjeOCBB0hJSWH69OnlpmsgWb16NZMmTWLBggVs27aN06dPB1olRSkaIlIh/rp06SKKoijKhYXH45Fp06ZJgwYNxBgj999/vxw5cuQMmaysLJk6daq0a9dOgDP+QkJCpEWLFtKvXz957rnnZMGCBXLy5Mky0fXaa6+VOnXqSFpamoiIjBgxQowxkpSUdIZcUlKSADJq1CgREcnOzpa4uDjp3r17mehVkUhNTZWYmJgzxigoKEgaN24s/fr1k82bNwdaRUURERFgheSTk+qxCkVRFCXgHDt2jKeeeoq3336bunXr8vrrr3PTTTcxbdo0XnjhBZKTk2nfvj1PP/00nTp1YseOHWzfvp3t27ezY8cONmzYwJo1axARqlatSvfu3enVqxd33HEHcXFxJdZv/vz59O3blzFjxvDwww8DcPToUdq2bUu9evVYvnw5ISEhiAg9e/Zk27ZtbNy4kRo1agAwbtw4HnzwQZKSkrjkkktKrE9F5d577+Wjjz5izpw5hISEnDFOM2fOpF27dixevJjg4OBAq6pc4BR0rCLgO8beP905VhRFUZYvXy5dunQRQOrUqSOAJCQkyOeffy7Z2dkF3puWliazZs2SRx55RLp27SpBQUFSq1YtWb58eYl0ys7Oli5dukiTJk3O2pWePn26APLaa6+JiMiUKVMEkA8//PAMuaNHj0pERIT8+te/LpEuvhw/flzWr18vc+bMkXfeeUdGjRolQ4cOlQ8++OCctioLFi9eLIA8+uijfq9/+umnAsjLL79czpopytlQwM5xwJNi758mx4qiKIqIPUbx5ptvysCBA+WLL74odqK3detWadasmdSsWVOWLVtWbH2mTZsmgEyePPmsax6PR66//nqpXr26/PjjjxIbGytdunTxq/OIESOkSpUqkpKSUmxdRGxSnJiYeNYRkypVqkiDBg0EkB49esjq1atL1E5RyMjIkPbt20uTJk3k+PHjfmU8Ho8MGjRIqlSpIuvXry833RTFH5ocK4qiKBck27dvl+bNm0uNGjVk6dKlRb4/MzNTWrZsKR06dJCsrCy/Mtu2bZNq1apJrVq1BJDvvvvOr1xycrIA8uc//7nIenhZu3attG3bVowxMmLECPnss8/k+++/lz179kh2drZ4PB6ZPHmyREdHS3BwsDzyyCOSnp5e7PYKy0svvSSAzJo1q0C5lJQUiY6Olq5du8rp06fLXK+8vPTSS7Jq1apyb1epeGhyrCiKolyw7Ny5U1q2bCmRkZF+E9fs7GxZt26dLF26VPbt2ycejyfn2ttvvy2AzJkzp8A2Xn75ZQFk6NChBcpdc801EhsbK5mZmUXqg8fjkffff1/CwsKkXr168vXXXxcon5qaKvfee68A0rhxY/nHP/5RpPaKwpYtWyQsLEwGDRpUKHnvUZQXXnihzHTyx9atWwWQzp07l8uxk8zMTFm7dm2Zt6MUD02OFUVRlAuaXbt2SevWrSUiIkK+/fZbWbNmjbzxxhsyaNAgiYqKOuN4QtWqVSUuLk6uueYaqVOnjvTq1euMhNkfmZmZMn78eDl06FCBcrNmzRJApk2bVmjdjx49KrfeeqsA0rdvX9m/f3+h712yZIkkJCT4PQddGng8Hunfv79ERETIrl27Cn3frbfeKqGhoeV69GP8+PE5YzxlypQybeu7776TDh06CCCTJk0q07aU4lFQcqzfVqEoiqJcEOzdu5errrqKjRs35pQ1a9aMXr160atXL2JiYnK+XcH7b2pqKp9//jldunQpFR2ys7OJi4sjNjaWxYsXFyi7bds2Pv74YyZOnMiePXv485//zOOPP05QUNF+oiAzM5MBAwawYMEC/vnPfzJgwIB8ZT0eD3v37qVRo0aFqvvzzz9nyJAhvP766zz00EOF1ik1NZX27dtTv359kpKSyuXHUQYPHswPP/xATEwMaWlpJCcnU7Vq1SLV4fF42L59O02aNPH7ozWHDx/m8ccf54MPPqBRo0bUrVuXn376iZUrV9KmTZvS6opSCui3VSiKoiiKiOzdu1eeffZZmTx5smzfvj0gOowZMybn7PGMGTNk5cqVkpqaKh6PR44ePSoTJ06UK6+8UgAxxsjVV18tixcvLlGbx44dk4svvliqVasmP/zwg1+ZlJQUueaaa8QYIwsXLjxnnUeOHJH69etLly5d8j2PXRD/+Mc/BJC7775bli1blu8ZZI/HI8nJyTJ+/Hj59NNPi9yOiP2QZ+3ateXuu++Wr776SgB5/fXXi1RHZmamDB48WACJjIyU/v37y0svvSRLly6VjIwMmTx5ssTExEhwcLA8/PDDkp6eLrt375aoqCjp1KlTmX3/tlI80GMViqIoilIxSEtLkzZt2pz1bRORkZESFhYmgLRu3VpeeOEF2bFjR6m1u3//fmnRooVERUVJcnLyGde++eYbqV+/voSFhUlUVJR069btnEdJHn74YTHGlOir8u6///6c/kdEROQknPPnz5d33nlHhgwZIvXr1z/DTjNnzixyO94fZvnss8/E4/FInz59JCoq6qwfnMmPjIwMGTRokADy8MMPy//93/+d8aM0oaGhAkj37t3P+sDf7NmzBZARI0YUWW+l7NDkWFEURVEqEB6PRw4dOiQrV66UL774Ql577TV58MEH5aGHHpKlS5eeMzEtLps2bZKYmBhp2rSp7NmzR7KysuSZZ54RY4zEx8fL//73P/nwww8FkOnTp+dbT3JysoSEhMg999xTYp327t0rf/vb385KOAFp2LCh3H777fL+++/LunXrpHPnzhIVFVWk880iIqNHjxYg52v0VqxYIYA89dRT57w3IyNDbrzxRr+7zSkpKTJ9+nQZOXKkTJgwId8P+v3+978XQGbMmFEkvZWyo6DkWM8cK4qiKMoFxIoVK+jduzctW7akdu3aLFq0iGHDhvH2229TvXp1srOz6dSpE5mZmaxfv57Q0NCz6hgwYABLlixh48aN1KtXr1T1O3DgACtXriQuLo4WLVpgjMm59tNPP9GlSxe6dOnC/Pnz/Z779Ufv3r05evQoq1atyim77bbbmDlzJps3b6ZBgwZ+78vIyGDw4MHMmjWLcePGMWLEiGL1KTMzk549e7JlyxZWr15N06ZNi1WPUnromWNFURRFUXL46quvJCQkRMLDw/1+m8LMmTMFkHffffesa3PnzhVAXn311fJQ9SwmT54sgDz77LOFkk9PT5fQ0FD5wx/+cEb55s2bJSQkRH7729/6ve/kyZMyYMAAAeSdd94pqdqyadMmiYyMlJ49ewbkO56VM0F3jhVFURRF8WXFihXUqVOHFi1anHVNRLjyyivZtGkTmzdvJiIiAoDTp0/ToUMHPB4P69atK5dvmfDHsGHDmDJlCvPnz6d3794Fys6dO5cBAwbwn//8h379+p1x7YEHHuDdd9/lxx9/JC4ujuzsbFavXs2iRYv4/PPPWbZsGePHj+c3v/lNqeg9depUhg4dynXXXcegQYPo1asXrVq1OmN3/HzB4/EU+ZtTKhK6c6woiqIoSpH47rvvzvpFv7Fjxwogs2fPDqBmdjc4Li5OYmNj5cCBAwXKjhw5UqpWrSo///zzWddSUlIkIiJCunfvLgMGDJAaNWrknHdu1aqV358MLynPPfec1K1bN6ed2NhYue222+S9996T5OTkMjtvXpps2LBBGjZsKK+88kqgVSk26M6xoiiKoihFZdCgQcybN48tW7YA0Lp1a3r06MHcuXMDvtu5evVqunfvTt++fZk9e3a++iQkJNCgQQO+/vprv9dffPFFnnzySdq0aUPv3r1zvvc6Nja2zHQXEX766ScWLVrEokWLWLhwIfv27QOgXr16OTr06tWLNm3anNU3Y0yp79ru37+f559/nltuuYWrrroqX7k9e/bQs2dPdu7cSWRkJNu3b6dOnTqlqkt5oDvHiqIoiqIUmQ0bNkhQUJA89NBDct9990lISIj8+OOPgVYrh3Hjxgkgb731lt/re/bsEUBefvnlfOvweDyF/kq3ssLj8cjGjRvlgw8+kNtvv10aNmx41lf9+f4FBwdL//79ZerUqX53xLOysuTf//633HbbbRIVFSX33HOPpKam5tv+119/LfXq1RNAQkJC5KOPPvIrl5aWJh06dJCIiIics99PPvlkaZmhXEF3jhVFURRFKQ6/+c1vmDRpEtnZ2Tz44IOMHTs20CrlICJce+21/PDDDyQnJ5+12/vxxx8zbNgw/vvf/3LRRRcFSMuiIyJs3bqVRYsWsWfPnrOuHzlyhOnTp7Nr1y5q1qzJLbfcwrBhw6hduzaTJ09mypQp7N27l9q1a3PFFVcwZ84cateuzZgxY/j1r3+dsxOdlZXFc889x4svvkjbtm354IMPePbZZ5k3bx5PPfUUf/rTn3JkT506Rf/+/fn++++ZO3cuffv2ZciQIcydO5dt27YRHR1drjYqKbpzrCiKoihKsdi9e7dUq1ZNoqOj5fDhw4FW5yw2bdokVatWlZtvvvmsa3fccYfExMTk+/3D5zPZ2dkyf/58ufPOOyU8PPyMXeUbbrhBpk+fLqdOnRIRkTVr1kiPHj0EkN69e8uGDRtk165dcvnllwsgiYmJcvz4cRGxvwQ4fPhwAWTo0KFy6tQpycrKkptvvjnnh1S8rF+/Xowx8vjjjwfEBiUB3TlWFEVRFKW4zJs3jxo1atCtW7dAq+KX0aNH89RTTzFnzhyuv/56wG7+xcbG0rt3b6ZOnRpgDcuW48ePM2PGDNLT07n55pv9fve0x+NhwoQJPPbYY5w4cYLq1auTlZXFe++9x+23336GrIjw0ksv8eSTT3LFFVcQHx/PBx98wGuvvcbvf//7M2SHDh3KzJkz2bZtGzExMWe1KyKMGzeO8PBwBg8eTM2aNUu388WkoJ1jTY4VRVEURTmvyczMpHPnzpw8eZL169cTHh7O2rVr6dixIxMnTiQxMTHQKlYYDhw4wKhRo9i6dSsTJkwgLi4uX9mpU6dy1113kZmZyaOPPsorr7xylkxycjLt27fnkUce4a9//esZ1zweDyNGjODdd98FICwsjF/+8pcMGzaMfv36ERwcXLqdKwIFJcfn7xfUKYqiKIqiAFWqVOG9995j+/bt/OlPfwLI+XaKvN9tfKFTt25dJk2axLfffltgYgz2VwQXLVrE2LFjefnll/3KxMfHM3ToUN566y1SUlJyyj0eD7/73e949913GTVqFElJSQwfPpz//Oc/XHfddTRu3JhRo0adcU9FQXeOFUVRFEWpFCQmJvLJJ5/w3//+l1GjRrFt2zaSk5MDrValZ+PGjbRt25aRI0cyZswYPB4P9913HxMmTOCPf/wjo0ePzvlgX0ZGBnPmzGHy5MnMmzePHTt2BOTDfHqsQlEURVGUSk9qairx8fG0bNmS//3vfwwfPpxx48YFWq0Lgrvuuotp06axefNmnnnmGT788MOzvvEiL+np6URGRpazphY9VqEoiqIoSqUnKiqKV199lWXLlnHy5Ek9UlGOPP3005w+fZpu3brx4Ycf8swzzxSYGAMBS4zPhSbHiqIoiqJUGu6880569+5NaGgovXv3DrQ6FwwtW7Zk2LBh7N27l+eee47nn38+4L+iWFz0WIWiKIqiKJWKQ4cOsWnTJnr06BFoVS4oTpw4wcqVK7nyyisDrco5KehYRUh5K6MoiqIoilKWREdHn3e/2FYZqF69+nmRGJ8LPVahKIqiKIqiKA5NjhVFURRFURTFocmxoiiKoiiKojg0OVYURVEURVEUhybHiqIoiqIoiuLQ5FhRFEVRFEVRHOWSHBtjBhpjbiyPthRFURRFURSluJTXznEGcCpvoTHmN8aYFcaYFQcPHiwnVRRFURRFURTFP+WVHIe5vzMQkfdFpKuIdI2JiSknVRRFURRFURTFP+XyC3kiMrM82lEURVEURVF8duJJAAAbPklEQVSUkqAfyFMURVEURVEUhybHiqIoiqIoiuLQ5FhRFEVRFEVRHJocK4qiKIqiKIpDk2NFURRFURRFcWhyrCiKoiiKoigOTY4VRVEURVEUxWFEJNA6AGCMOQjsCFDz0cChQpRVFtmKrNv5JluRdTvfZCuybuebbEXW7XyTrci6nW+yFVm38022IutWGrLlQVMR8f8LdCJywf8BKwpTVllkK7Ju55tsRdbtfJOtyLqdb7IVWbfzTbYi63a+yVZk3c432YqsW2nIBvpPj1UoiqIoiqIoikOTY0VRFEVRFEVxaHJseb+QZZVFtrzbq8yy5d1eZZYt7/Yqs2x5t1eZZcu7vcosW97tVWbZ8m6vvGUDSoX5QJ6iKIqiKIqiBBrdOVYURVEURVEUhybHiqIoiqIoiuK44JJjY0wDY0wPP+UN/ZQ1zqeO+vldy6e9WnnKYo0xjfKpN6Iw9RYFY0yMMaZVCeuoV9g68rNxAbKtS6JbSSmufYwxpiz0qWiU1H+K4jtOvtD+UxFw/Su2D/uLEa78rDiRX4zIb4yMMU3zxipjTJN84k9sXp8uq3hpjInOR4em+ejlT7aRMSa4uPXmV7e/el15Ez9l9Y0x7f3JGmOC8pTVM8bU9iPbwJ/O+ehbJnE4n/tj8xn7s2zszzauvLztc16tt2U4v86aywXUW98YE+anvG4+sv7GKL9Y5W8e1fPXj4rGBZMcG2MijDEJwGWAuDJjjKlljKkC9DXG9HXl1Y0xLYFBvk7uytsCtwA9fR3KGNPQGFPd5/9rOSfqCLTyKasBdAGuc3V5620BDAIuNcZU9amnrjGmo59Fq64xprtvgHHBJdrn/8NdYLoCaGqMqe9zrYExpr2feqN8J5oxpqYxJhJoDcT5OnbeOnzau9zXxu7f+saYKD/1NgFijDE1fa7FGmM6+NEtxrd/AbJPtDHmZmNMsLgD+64vl/oJ9o383F/PGHOxn/Jqxpj+fupo7Ee2ft468to3gPbJ8Z+CfMdfHefwn9p5bez07ZpP327wY8safuo4a4x86s5r40bmzDkebmxM6QbUMD6LshsPf3bPiRPGT4zwKT8jThgbI5qTJ0YYG9faYeNazhi5OqoCbYFOxph4V1YT6IGdd94YWN3YOHcbdqzLLF766HsV0Mb136tvGHCFu+5rh6uAq9x133g5FLjM6VXoevOr21+9PrZoAwxx133tPhC4xBjTwKfeKkBf4Gof+3Z0Zb6xtbqr41pnT1/9yioORxljOufjl+2cvjWdbeKAKj5j7623r7NxLX+2KSf7nNfrrSm9+dWOM+dXDWMfBG91dee1z2XGmHifsibAjdh54Dt2CcAvjDGdXFkNY+PPlUAjn37UMMY0A65249nO1Rvp6uySp976wEBjTI5uFZWQQCtQjrTBTvbDQJQxJkhEPMaYnsBJ4N/AA8aYLUAkUB2YB/QxxhwXkf1Ac6AmsAbrJO2MMRuBRtgJvNoY818RSQcuAqq5Ou43xuwAOgBVga+xzutxTtYQaABsw07uk8aY1U7f2sBx7GK2wxgT7vrSyN0T5mSzsZMyxBgzV0SOAbGuHweAOlinPur60Rk4BWQBP/nU2xk4ZozJBA4CCUA48APW2RsbY44DTX3qOG2M2ebaqA6kANEukIQZY+KAS4FkY8w6NwYJQATwLXAXsMfNt2wgHruohwHLnW7NgV7APmPMN8Bpp2+x7OPur+10KIx9wlxZttM/1hhzGIhx/hLt9P7RGFMNG6CvBWY424Q5vS7Ckg5scvVeDBwBjE8d1Z1+1wGLnA7VgShsMlbHjdMe5yeXA2uNMT+KyCEfOxTXPkFY/y2sffL6TxbQwo/vHHS2aJbHf7a7NiLy+E8VY0xnbKyKBC42xmxw9m4L1HC29NonBqgLnABCgQw3Hm19+hPv2muFDeqzjTFbRER8+tHB3Z9ujDng+tkLOGiM+R4bM6KcTgeBS4BDbt40d/dHYOf7cmN3qM6IE9g5YYD5uBghIgexyasHnzjhbFmbM2PEUmfHjk6HakBDp+/FzjbfAz2dzS92fZoJXOuz0DfCxrUlwA3GmN2urVKPl0Amdu6nA52AbGPMt9hFNBiYDtzt5lY7bHLmjRHtnK82cvp97eo44mxUqHqdbhfnqbs9dp6cUa8xZq2zfVtgLTbJ9jjbxmPnRlugtTEmxfmBx7V3s/ODU67eVcA1xpj9InLA9SMKSHL3ZfuLwyKS4vQNBpZRQBwGNgL18R+H47FzLMi1kWzsw1K260OMq6Mhdr4tB670Gft6zscOYONYMLDX2c7XNvux8SYhj31S3X1BPvbp6nzsDPu4cazvxz5nzaMC1tsErN/7rrfBlMF6C/xMbvw6Z7wkN0akcub86ur8x3d+VcHGu7zzq6Frx3d+VcfGnN3YOb5XRLZiY0028E9guDFmL3YdqerG+TpjzGoR2Yf1fe987meM2Qe0BGo521zlxvigG/tabgyjXRtdXdl+oJubc5tdvUFuPH5pjFkjIsepoFTqnWNjTB1jzH0AIrISO+EAjgLe8rnYoHA5dnLtEZH/YSdDFWALcJOTXYdNOH8GpmADQnPsoG9y9zxojGknIguwE6UtdvGoLyLfuP+OwwaCjkC8iGzEOtZR4EusszYjdzFphZ3kuPbauL5swgaxbthF4HtsEH3CGBPiHLIGdpxXYSdRrNPzAHZi1DX2STXWtZOGDfr/B1wmIt9hF9QEIBlYhw0gkcB2bMAYCHQXkd2u3ANscPaLdfpud+0mAle6eoOcfXY4fRKwicEy7OQOcrrVcf3f5mx3JTYwtPe1j0uC6rs6vfYJwwahS5x92mEXw8tFZA82CAQ7+2x0fWuSxz5Brp8tnb1TyV0AGrj+1gS6On1ru779hA3aHVyZuDGOBFq4QF3PjXUDN1ZXGLuT0diV/QAkGLuz0tjJb3e6tXC2wdmrJnCRMaYO1qfz2qcmNnH0tU+Ek+vq7JMAPAT0dIHS13+2uzby2qeq07U1NtkbANyNXdRCyPWd1e7/65DrP7G4XQ7nPxHk+k8YdnFs6Wx30PUr1v3tduVN3UNGfVfuDbiZrjzc9aEmdp51wyYVbYEfsfO3gxvnWKyPpbr7Wrjr3YF/Ofs/BvQQkV1OBqzvHHT6RgA7nY2C3QNwPNYP9gG3uzq/dvZtgY0zQwBE5F/YxaWrs/UWEVnrxu4oMBc7l9q4mHQUO0d3ubriXKwJwfpWSyDbldV047QZG9faiMhPrp122LmwO0+8zHbj6Rsvr3Ft5o2X+7HxwDdeVnP9znJl6Vi/3ICdF21EZL7rc3dnk7pO3xpYf5nvdIx3+jZy9hCfttKxyd2Prq54V29NZ4Nd2J3Cdj6xuCU2iczyqbcK1t+bAo1d/467ca0LBLuyE0AG1s9rAQ1F5GvnA12cbidcfI9x9W52OjR17dVz9S4Btro2W2LjcFvgcWNMDbeeVOHMueS9dzs2Hv7aGBPhkqEa5M4j47Wds08m9m1dEDY+Xera7o6N1d+6ttpgfWuQMaauiGxz/azl/CTF2f0ENulLwz6ItRCRH7Hr5CnsetYIaOn6EY317epApohscX0Jxc6Dfq6OLeQm+kuw8b8ZuettHDDKGNPF1VsdO7djgOY+a3Cc1+5AZzceUeSutytdOwmcGS+DsXHNN16GYNdt73qSADwO9Hfxy3c9WYf1XW+8rAEkGmMucvM5zY3NIeAeY0wnEZmHnRPdsPEv2vlJfWy83QD8ytWx0dktyI3rWuyacxrr6wux8xQR+TfWdy92+pwSkf84+zQCFgO/cLLfOX+q63TIdv5x3NW9G7uzXdOVe8c/yvnEPGenw8B32DcHNVy9rZxN/uV8o8JSKZNjY0ywMWYYdodji7GvFCOwi1dL7GTY5FPe0ZX9C5u0hGEndi/s4P7kZMOwkyQN62StnJP/jN2BWg98jn3FGoF7+gc+Aaq7p9AD2ORuK7ACqOmSup+x43GZiGSKyHrsAl0d68A9AVz5EWyguxhY5AJyBPY10SZsMGpu7PmiEKxzXykip13AScM6+RHshL3EJdLp2AU8DfgUu6PXyJUfwybmXXySmG7YADwLyDL21VkYdrI0B7q5eo842erYwBtqjLkIG5xOYgNSVzd5vMEnBbuo9XDt1XL22YANqJucLSOwAaYuNqHbig02TbCTMFtElji9BmEX6TVAbWPMFa7ONOwifomI7HV+chobDBoBUW5hSHG63QDUFpEN2ADQ2MlmAbVcHaGuv82BVFdWxdniv06fJiLifTCo6epKAWJFJNnZvD02mTnlU9ba2Wm7S/BDnf9sxyYil7h6vf4TjA3wl4rIdtdOU2ygyhaR750dfwXswSYi613SW9W12dGN0X5Xb6Ybv8bkLq4tsIvXFtenDq6/m5xPDHJj5Os/p7EJVTNjX+EFO59IAAa6/u3DJm6XOvtucmPstXskcKPz7YNu7COwrwt/KSKpuCQKG/S3u7E74mx3GJvMXOrjr2HYRepnEfnW2XUkdtGbAzQw9hVqqNOlEXbh3eXaDnE2ucgdv1mFffDLwMaUI+Qmk32xiXKyizPVnb13YhfgVj4xorsboyRs7Ih049GM3JhSy8Wfi509ZnBm/OmFfSD1xp9wrJ+dwiZdjV29KdhE5jR2B72l0625G6OVTraaG4fO7t+jTraaqzfa9T/U1XuQ3MR2JxDu6m2JTUbmO9nqzj4dsXFqF1DN6Zvg6j2AncstnJ/EYGNSBjYBbIbd3ezi+uFb3tzpscvZoSnWp9tjk46a2Iemltg5J66NeFe2F5sQhDj9Orp6m7p6s5yst96LsPOmBvaB11sONr5XcX553Om7B5sId3a+loadR92Bqj5J8BXYGL0b+2YlHhvXTrgxqeNiRyo2jnqwyVYd7FupK7BvqLz9SMDOidZYH43HHj1p7/QMxyZhh12MOODslenaTHflXv8JdeXHjTGhzu71nN1TjTEhTv+LyY27R135cez61RNI8HkI6kZuDGrnxqO569ch7LGPls6Gnch90G7nfCUd6+/dgatd7EjBxkvvhtO1rvwQuW9owkRksavvNmysWwrUMcZ438wcwx5puNrFr0PYne4dwMfYtzvtnN3inS0XYudSWzf2P2G//7eesW9ejwN9gP9h50dDY4+zXIaN+z9g4+9x99cDu9YmG3ucIxgbj3dhd6Vbu7Lt2Iegpdg8p717aGrn+rEMG0+PYuNPB9f+FtfeMezDzBZsXGvtNgP2YOPaVnetDZZtwAER2SgiWVRgKtX3HLtgHC0iu1zylYVdbC8HvhaRn3zKw7GvB97CBoGCZIOxk2gqNoGYgZ30TbATqgZ28TyKnVyDgCexjg92st0ITBaRzca+JvaWDwBexgaoKq5sAzaI18Uu5qfzlA/ABrpT2Cdi7+5eBjbJM06Pr8h9fRiC3VHJwgZI7wcovLs+2e6a97VwEPaJexx2ooX7yJ5yuqWR++TaBnjTyXp3imLctX3YgNwUu6hmYJ+qL/KRrYUNImGuD6ddeU3sYnodNph4d3cOYgN4M+yidcinvDU2QGx3171Pvh5XtwebwCRhF+Rwd39DV+d+V+611R6ndxdXfxXsQpTp+tLYR3Y3diHo5+7L9imLJvfVVJzr305Xd2MnG4dddPdiA9Rh16/8yrY7G9V29ezD+mQqNpDVI3dR95bf7f7d7MbQu4N9wtnAgw2E32ADXLjP/WlurL075g2xyelCcncJM7H+ttjd0xebZHt9Zbe794TTvYqPfAdXxxFs8N2JnUeHXX9buPu9DyXiI5tf+XbsA817zk7NsX5U1Y3px86uzbGJ/C+ACa4/rV2dx5yeaa5/nbGLxHXA3939zbBzrImz0Xigt5Nfho0BO7EL82XAAhH50cUDj9OnH3YetXJlQXlkO2Hnd4jT81MXq7zl1bBnEP/odDiX7D3YNwXeHXPj7LcZWOBsEodd9HsCo7Fz3yvbxNksP9kY7INZsLPZt+6vmSuvCvR3OjTwkb0Im7gVJBuPjUXBTpd5TtZb7j026FvunWOf4naX/cg2d/avhk2S3sTG01aub7HYuf2vYsqGOhvOxK4rPzv9tmHn9Qbs2NfDxkEPdlcwr2x9rD+mY9cAf7KXuvFJdnZqg41h3jOfs7CxDmeHKs5Gc5y+zbHzP97ZrKqrI8vZ/BR2V7fZOcqNa+d9bMw4Vx1Z2LHv7/rR0fXVu+YZbLJbHRsz3iP3TUBVZ4/D2DWwuU95OPZBtAe5vlPXx+atsbHRt/wybGJ70o1nkLNnhhuD49h4swabP3jrbYAd6x/IXaN3Y+f4bmfji5zeB7BvNF/BzqnIQso2x8aX49i8Y43rX1dXngn8DhsP2hZB1vuW4WfsGfwl2PXA+5B5Gvg98CD2YbIg2Z+xb8XGiMhJzhMqzc6xsQe9W2EPe9cm9xygwQbZPj7lfbBOdgzrvAXJXoJ1Ju853jXYiRaHXYSuwQafLdiFMMz9dzVswnUxdqJ/jj1PVdunPBWbMJ/CJm/eT3Zeit2ZXI4NdMHY4N3f6fIVdqLWxj6RXocNGPtdfzYC37sdM996ewA1RWSVqzfI2eKX2ARmh9Mt2F2f6s4Ebckjex1212mnj+xnTnark/W+PumGndwHnewu7M7dIR9Z79EARGQpdvcqCLs4RGMXl+lOt9quj6dFZLVrW8g9a7odu8Clu3GKcbbe4WOfrcAPbudlq7u/NnZhyHC7E+udDunYILwT+A828B3G+kqWiPw3j6yvvmk+ssa92lpBbrJ+DPD41JHt2gnFLn7efvgr877mMiLyFfZtAc5mbbEPiUudfcAuSJe5MZqLTUzbYpOYX2KD/X6sn23BvsoX145vvXWcX/7obOZNZE5id8SzXJszsbss4vQW91cNu9isdmMVhk1Ul7tdaa9PRDjbrcH6+yns4rYROORsts7J5lfurWMt9hhUdWezqkCQiCRhd0xaYRc6727WP904VnNlPZ1NNmMDfiawVOy50TXu/uOuL1XEHotY6O4LBe7H+uEi7OIbjF2UfmHsOc5D2PkXhl2gq7uyrn5kU7H+7HF9+pVP+dVO3vvW5VyyWcB85xNHsfNwPzAb6z+nXXlL7MPRKhE5kkd21jlkjzk/2QVMBOo5We+bqJ0+OvjKTiik7HGcL/rIesvX5yn37nQdEHu2Oz/Zo26c9gFr3O7fUecT+7GJSkQxZaOxCVcy1jeTsf54Ahu/qjm/3IWNOzuwvplXth4QKfaYy8Z8ZIOx/pnl7LcWu9tf19myhtPR+2bkIDaOhPvo6z1XvUNEdjqb1cHGtvVAAx/75lfeEutzB0Tk50LUsdX1Q1w/QrEPuhFADRd/dpC7wfCT2CNgR7HzbD92/tXw6UeMs/suV5f32FaGqyPa2WoFufOwJjZGzXY2jHS+gts93kHu5kkWNi5764104+QRkTVujLpgk/O5Tjdx917lxnWNWxf3FEF2J3Zdrg5Mw85F7xuZXzldlro5UxTZXU62GvARdi6KK/+V02GBS3YLku0DHBWRF86nxBgqQXJs7KdWr8HuDP2Iddie5O7qdvEpH4gdrFPYnY6CZHtiHWA+dtC9SXcI8Gvs0+qD2EXTt96Ffurt6qfeBdhJf6tbsDLc/x/APmn29ymvi53Q4a792k7XBsBYd8/VWKdfiU2yzlVvHWxSEwU8ig2o3jpWOPPeWgzZ2k62OXanIa9uWXl0C8YGnB3YDxB4y6Pdv82cjLcfh7E7vpflsc9x18YJH/s0BN7ATv6CdPBX7ynsAnQKu/h5yN1ZScfuYvYppmw4diHaBPTyKW+IDeQNsQlYfmXe+zf63J9B7nnorXn6F+P0aOIzRhdjd0E+drYryD556+2EfWjsgX1gWIrd2amOXUh+xr4l8NZRC/uQVA/7JmL/OdqLdmXtXNtePzmJTRr75/Eff+V56zjkfKI6dhf3Wp/EtI/T0XsExTvHuwOvY33YG/wXOJlbfe7PW28n7IPEHdgHhm+crZq5vl/kxv449kNJfZ3NlpD7+tSfrG/5xfnU8V0hZU9ijzoccq9ta7i+dMb66yFjzOXkvr5eUQLZg64f/mRXlkD2EPYIwN5zlPvW8UMRZJP8yHYspmwcNjYtxs7hxi5xq4Y9ZtcU+0C9NY/syXxkPz+H7FHsfMtJeLFxspPzjcXYh84IH3075NO3ZYXoW1nap7fT/V1gWyF065BPvb72OYad8z9jE/BaeR7EjmMftg84212Gfbj6AvtBvfzqzU82ArsJcGke+1yCjS/rfPQtjmz3fGTXBkg2DrvmvOtseN5xXh+rcK8jvWdks7FJyTZyP+TkW+790NCeQshuwy6QjbAJ7kfYXdvmrp6t2Alc3HobYgPcROyuUip2dzUSm7B4yz3YiVcV+4r2fqwjZmMXO++rkN3YBTu2kPUWpY6yki1Kn8vbPiXVIT/ZotRRVvWW1D4vYheh/2F3XH4m981AGDYZLa5PlFWfD5O7U+YrK9gFLsxHNhL7KnALNskuSF9/9b6I3XFajX0wPuHur4eNJWlOx0xnmwbYHRaPT1v+ZEujDl/ZbHKP/ES7fqX5lHs/fLb7PJGt6LoZrI80w64nV2P9ri42ocsoRdntWN9ujV1rPsIen4nErkHeh+yKYvei9Lk06vVnn6uxsaw2dt3/oBTHqKL6ZVn6+34ROcx5zPmeHHvPxV6LfSVywJ3LO6sc+3RYWNkq+E+6D2EdpDTqLe6idwjrkMWtt6h1lJVsaS70FU2HkiY3ZZk0ldQ+e7DJX3fO9PfO+dRR2PYCmSiWh31KFJNKKa5VWtmKrBtnryfeBCIF+1antGX9JSx7sD5/aQW0e1H6XBr1FiX5K9EYVXC/LLO5SCUgJNAKlASx3wvYCbuVv8idc8m3vLCybsBPceaAb3R19CvlepON/a7kwjhjsmuvT3HrLWodZSVblD6Xp31KQ4f87F7YOsqq3lLynw3GmCHk8XfsTnKxfaIs+1wR7FPSmFRadVRm2YqqWz7+86Oro29py7ryA37KK6RfFqXPpVGvP/uQT/JX0jGqyH5ZlrKVgfN65xjsr8r4GxB/5UWUHYLdDfq777UyrPes8iLqUOh6S6nPZaJbRbBPWckWpY6yqreU7JOfv5eovTLuc0WwT4liR2nUUZllK7JuFSS2VmS7V5b1tiL4WsBlKwPnfXJcVpTVgJeGM5a03qLWUVayZTVZK4IOJa2jLBcnf1QE/wnAglyu9lEuXCpCbK3IVJb1Vqk8aHKsKIqiKIqiKI7z/qvcFEVRFEVRFKW00ORYURRFURRFURyaHCuKoiiKoiiKQ5NjRVEURVEURXFocqwoiqIoiqIojv8Hs4jE80OHZnUAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 864x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x576 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eFkbWi2aPqv3"
      },
      "source": [
        "## Test Different Combination"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CKJfHR7UnGgt"
      },
      "source": [
        "### Resized images without dimensionality reduction"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wdNlFDY22TAb"
      },
      "source": [
        "#### Split training and testing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b_JYXuRI68ZM",
        "outputId": "0aef920d-62be-47d4-a706-094ac2fd612f"
      },
      "source": [
        "origin_X_train, y_train, origin_X_test, y_test = splitTrainTestSet(0.8)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "------------------------------------------------------------\n",
            "Total number of weeks:         112 weeks\n",
            "Training input:                (89, 602, 382, 12)\n",
            "Training output:               (89,)\n",
            "Testing input:                 (23, 602, 382, 12)\n",
            "Testing output:                (23,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "voLlp1JBpqRH"
      },
      "source": [
        "#### First two bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rLqXB9P0DPgP",
        "outputId": "04320465-b14e-46d9-91e8-736e59117dde"
      },
      "source": [
        "first_n_band = 2\n",
        "print(' RESIZED IMAGE WITHOUT DIMENSIONALITY REDUCTION : FIRST TWO BANDS '.center(100, '='))\n",
        "\n",
        "selected_X_train = origin_X_train[:, :, :, 0 : first_n_band]\n",
        "selected_X_test = origin_X_test[:, :, :, 0 : first_n_band]\n",
        "\n",
        "# LR\n",
        "MAE, MAPE = classified_with_LR(selected_X_train, selected_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(selected_X_train, selected_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "================= RESIZED IMAGE WITHOUT DIMENSIONALITY REDUCTION : FIRST TWO BANDS =================\n",
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [1673354267770, -22868437251675, -22511634620422, 53247525538014, -34717885302216, 60361240856599, -1854869421725, 5505640077899, 5908017111515, 34384319654725, 30436979969638, 19855009194995, 71443527575209, 6359414614693, -5929351554176, 35048408419020, 34576089118802, 78356080951674, -5295705872468, 17230452648749, 5040743547260, 6505502744062, 6123983873921]\n",
            "    Mean = 16472974180080.998\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  24575398877693.0 (cases different in average)\n",
            "- MAPE: 939267873816.1606 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 188, 188, 188, 188, 188, 188, 188, 186, 188, 188, 188, 188, 188, 189, 187, 188, 187, 187, 187, 188, 187, 186]\n",
            "    Mean = 187.5652\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  158.9565 (cases different in average)\n",
            "- MAPE:   6.1741 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xxjcJrxXpzM9"
      },
      "source": [
        "#### All bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kEf4RzDGR87F",
        "outputId": "98a8c0ed-72bb-4695-cb85-67523b758def"
      },
      "source": [
        "print(' RESIZED IMAGE WITHOUT DIMENSIONALITY REDUCTION : ALL BANDS '.center(100, ' '))\n",
        "print(''.center(100, '='))\n",
        "\n",
        "# LR\n",
        "MAE, MAPE = classified_with_LR(origin_X_train, origin_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(origin_X_train, origin_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "                     RESIZED IMAGE WITHOUT DIMENSIONALITY REDUCTION : ALL BANDS                     \n",
            "====================================================================================================\n",
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [735232989636, -8526366323040, -7573519815685, -2014298661537, -5152719209312, -8184875211498, 2414704583652, 1990400237346, -739508613604, 1805136965051, -458509643996, 3625276638322, 199694233821, -12744980917998, -353607020319, -963618376628, 7141116572792, -8286923335743, -4222487707085, -7178973501072, -10382527796988, 4607948639208, 2255792122843]\n",
            "    Mean = -2261200571818.8696\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  4415574744230.435 (cases different in average)\n",
            "- MAPE: 176556141230.256 (times different in average)\n",
            "\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 188, 187, 187, 187, 188, 187, 187, 186, 187, 187, 187, 187, 188, 188, 187, 187, 186, 187, 187, 187, 187, 186]\n",
            "    Mean = 187.0\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  158.3913 (cases different in average)\n",
            "- MAPE:   6.1519 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4l04W9cRnNn9"
      },
      "source": [
        "### Resized images with PCA per bands"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LV22845f2AYq"
      },
      "source": [
        "#### Perform dimensionality reduction"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gQbyjnW2UN5X",
        "outputId": "a6a2f666-7981-449f-91f6-837be816b255"
      },
      "source": [
        "first_n_band = 2\n",
        "selected_X_train = origin_X_train[:, :, :, 0 : first_n_band]\n",
        "selected_X_test = origin_X_test[:, :, :, 0 : first_n_band]\n",
        "\n",
        "# First n bands\n",
        "n_band_pca_separate_X_train, n_band_pca_separate_X_test = pca_per_band(selected_X_train, selected_X_test, y_train)\n",
        "\n",
        "# All bands\n",
        "all_band_pca_separate_X_train, all_band_pca_separate_X_test = pca_per_band(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 459928)\n",
            "Resize shape    (89, 39)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 459928)\n",
            "Resize shape    (89, 39)\n",
            "(89, 78) (23, 78)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n",
            "(89, 324) (23, 324)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eJB3Unm2n_lJ"
      },
      "source": [
        "#### First two bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YtH4fUKLOojP",
        "outputId": "eb996d04-f092-4607-9027-e65a25e052cd"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(n_band_pca_separate_X_train, n_band_pca_separate_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(n_band_pca_separate_X_train, n_band_pca_separate_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [124, 263, 151, 170, 121, 133, 201, 187, 217, 218, 154, 151, 9, 118, 142, 92, 214, 84, 150, 230, 38, 93, 168]\n",
            "    Mean = 149.0435\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:     122.0 (cases different in average)\n",
            "- MAPE:   4.5055 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 189, 188, 188, 189, 189, 189, 188, 186, 189, 188, 188, 189, 189, 190, 188, 188, 187, 187, 187, 188, 188, 186]\n",
            "    Mean = 188.0\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  159.3913 (cases different in average)\n",
            "- MAPE:   6.1888 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dthF867voEDk"
      },
      "source": [
        "#### All bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9CGtUU53Dafp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "bbc52c5a-530e-4845-a74f-4b28f6f4c54c"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(all_band_pca_separate_X_train, all_band_pca_separate_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(all_band_pca_separate_X_train, all_band_pca_separate_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [204, 206, 214, 237, 121, 209, 275, 252, 225, 264, 209, 198, -24, 207, -24, -4, 213, 28, 161, -50, 110, 147, 205]\n",
            "    Mean = 155.7826\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  145.4348 (cases different in average)\n",
            "- MAPE:   5.5311 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 189, 187, 187, 187, 188, 187, 187, 186, 187, 188, 187, 187, 189, 188, 186, 187, 186, 187, 187, 187, 187, 186]\n",
            "    Mean = 187.087\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  158.4783 (cases different in average)\n",
            "- MAPE:   6.1544 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4ZIGVp-xoGK1"
      },
      "source": [
        "### Resized images with PCA (all bands together)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tl4mBEl02Hic"
      },
      "source": [
        "#### Perform dimensionality reduction"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BygSiP4mx3GJ",
        "outputId": "fb7c4b40-f454-4f0a-e4ca-a950cf803d0b"
      },
      "source": [
        "first_n_band = 2\n",
        "selected_X_train = origin_X_train[:, :, :, 0 : first_n_band]\n",
        "selected_X_test = origin_X_test[:, :, :, 0 : first_n_band]\n",
        "\n",
        "# First n bands\n",
        "n_band_pca_together_X_train, n_band_pca_together_X_test = dimension_reduct_with_PCA(selected_X_train, selected_X_test, y_train)\n",
        "\n",
        "# All bands\n",
        "all_band_pca_together_X_train, all_band_pca_together_X_test = dimension_reduct_with_PCA(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 459928)\n",
            "Resize shape    (89, 39)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (89, 2759568)\n",
            "Resize shape    (89, 27)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NZDy3vjwo0HS"
      },
      "source": [
        "#### First two bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fwd7t9ZsozIk",
        "outputId": "2b3fca6c-c623-42e4-dec2-8f2872aa77a8"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(n_band_pca_together_X_train, n_band_pca_together_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(n_band_pca_together_X_train, n_band_pca_together_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [124, 263, 151, 170, 121, 133, 201, 187, 217, 218, 154, 151, 9, 118, 142, 92, 215, 84, 149, 229, 38, 93, 168]\n",
            "    Mean = 149.0\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  121.9565 (cases different in average)\n",
            "- MAPE:    4.504 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 189, 188, 188, 189, 189, 189, 188, 186, 189, 188, 188, 189, 189, 190, 188, 188, 187, 187, 187, 188, 188, 186]\n",
            "    Mean = 188.0\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  159.3913 (cases different in average)\n",
            "- MAPE:   6.1888 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7xxY5pgRo219"
      },
      "source": [
        "#### All bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "f_dLs34LyBi_",
        "outputId": "7ad582d8-f543-465c-d303-2c248271d016"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(all_band_pca_together_X_train, all_band_pca_together_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(all_band_pca_together_X_train, all_band_pca_together_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [204, 205, 214, 237, 122, 209, 275, 252, 225, 264, 208, 198, -24, 207, -24, -4, 213, 27, 162, -50, 110, 146, 205]\n",
            "    Mean = 155.6957\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  145.4348 (cases different in average)\n",
            "- MAPE:   5.5312 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [186, 189, 187, 187, 187, 188, 187, 187, 186, 187, 188, 187, 187, 189, 188, 186, 187, 186, 187, 187, 187, 187, 186]\n",
            "    Mean = 187.087\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  158.4783 (cases different in average)\n",
            "- MAPE:   6.1544 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7wc9-PMQpChm"
      },
      "source": [
        "### Resized images with LLE"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-Bk2cwhM_I1K"
      },
      "source": [
        "#### Perform LLE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JXw8X2TXWiqY",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "00efdc38-24ed-43c2-ad27-39d46d2f0b52"
      },
      "source": [
        "first_n_band = 2\n",
        "selected_X_train = origin_X_train[:, :, :, 0 : first_n_band]\n",
        "selected_X_test = origin_X_test[:, :, :, 0 : first_n_band]\n",
        "\n",
        "# First n bands\n",
        "n_band_lle_X_train, n_band_lle_X_test = dimension_reduct_with_LLE(selected_X_train, selected_X_test, y_train)\n",
        "\n",
        "# All bands\n",
        "all_band_lle_X_train, all_band_lle_X_test = dimension_reduct_with_LLE(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "===================================== LOCALLY LINEAR EMBEDDING =====================================\n",
            "===================================== LOCALLY LINEAR EMBEDDING =====================================\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Jxr87o8KwHqQ"
      },
      "source": [
        "#### First 2 bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OYPh6ifiNtCR",
        "outputId": "907c4993-0372-4311-f2ef-db4a45da6902"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(n_band_lle_X_train, n_band_lle_X_test, y_train, y_test)\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(n_band_lle_X_train, n_band_lle_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [224, 220, 224, 224, 221, 223, 224, 224, 224, 224, 224, 224, 223, 220, 221, 224, 224, 224, 219, 219, 224, 224, 224]\n",
            "    Mean = 222.8696\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  194.2609 (cases different in average)\n",
            "- MAPE:    7.527 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [188, 188, 187, 188, 188, 189, 187, 187, 188, 187, 187, 187, 189, 188, 188, 188, 188, 188, 188, 188, 187, 188, 188]\n",
            "    Mean = 187.7826\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  159.1739 (cases different in average)\n",
            "- MAPE:   6.1822 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sELdV_SRwKSl"
      },
      "source": [
        "#### All bands"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "eF2DYwI3wLak",
        "outputId": "d8968233-7a68-4437-e7d0-facbe01ab39c"
      },
      "source": [
        "# Linear Regression\n",
        "MAE, MAPE = classified_with_LR(all_band_lle_X_train, all_band_lle_X_test, y_train, y_test)\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(all_band_lle_X_train, all_band_lle_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [210, 239, 210, 215, 231, 228, 210, 210, 210, 210, 215, 210, 227, 239, 241, 219, 210, 206, 242, 242, 218, 215, 210]\n",
            "    Mean = 220.3043\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  191.6957 (cases different in average)\n",
            "- MAPE:    7.421 (times different in average)\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [184, 193, 184, 185, 192, 190, 184, 184, 184, 184, 185, 184, 190, 193, 193, 186, 184, 183, 192, 192, 186, 184, 184]\n",
            "    Mean = 186.9565\n",
            "\n",
            "Real\n",
            "    List = [36 38 27 14 44 29 31 39 37 42 31 29 27 27 24 27 22 28 21 30 14 22 19]\n",
            "    Mean = 28.6087\n",
            "\n",
            "- MAE:  158.3478 (cases different in average)\n",
            "- MAPE:   6.1455 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G9wEr6hlxUC8"
      },
      "source": [
        "# RGB Images Run Here"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "P_QU-17t4Opz"
      },
      "source": [
        "## Read images"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ch5FzZUnxWE3",
        "outputId": "4fc0cf5a-a1db-4713-be18-0dc578d5bf70"
      },
      "source": [
        "# Define main folder\n",
        "dataset_folder = '/content/drive/MyDrive/Dengue forecasting with Satellite Images/dataset'\n",
        "\n",
        "# Define image folder\n",
        "image_name = 'image/image_rgb'\n",
        "img_folder = os.path.join(dataset_folder, image_name)\n",
        "\n",
        "# Define csv folder\n",
        "csv_name = 'csv/merge_cases_timeseries.csv'\n",
        "csv_folder = os.path.join(dataset_folder, csv_name)\n",
        "info_dict = loadData(csv_folder, img_folder, resize_ratio=(0.5, 0.5, 1))\n",
        "\n",
        "print('INFO_DICT'.center(50, '-'))\n",
        "print('keys:', info_dict.keys())\n",
        "print('')\n",
        "\n",
        "print('DENGUE CASES'.center(50, '-'))\n",
        "print('Max weekly dengue cases:', max(info_dict['cases_medellin']))\n",
        "print('Min weekly dengue cases:', min(info_dict['cases_medellin']))\n",
        "print('')\n",
        "\n",
        "print('WEEKS'.center(50, '-'))\n",
        "print('Max week:', max(info_dict['LastDayWeek']))\n",
        "print('Min week:', min(info_dict['LastDayWeek']))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "image_2016-01-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-01-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-01-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-01-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-02-07.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-02-14.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-03-06.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-03-20.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-04-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-04-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-04-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-05-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-05-08.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-05-15.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-05-22.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-06-05.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-06-12.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-06-19.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-06-26.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-07-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-07-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-07-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-07-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-07-31.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-08-07.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-08-14.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-08-21.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-08-28.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-09-04.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-09-11.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-09-18.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-09-25.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-10-02.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-10-09.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-10-16.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-10-23.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-11-06.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-11-13.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-11-20.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-11-27.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-12-04.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-12-11.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2016-12-18.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-01-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-01-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-01-22.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-01-29.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-02-05.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-02-12.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-02-19.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-02-26.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-03-05.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-03-12.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-03-19.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-03-26.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-04-02.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-04-09.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-04-16.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-04-23.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-04-30.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-05-07.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-05-14.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-05-21.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-05-28.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-06-04.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-06-11.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-06-18.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-06-25.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-07-02.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-07-09.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-07-16.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-07-23.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-07-30.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-08-06.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-08-13.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-08-20.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-08-27.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-09-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-09-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-09-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-09-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-10-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-10-08.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-10-15.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-10-22.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-10-29.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-11-05.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-11-12.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-11-19.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-11-26.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-12-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-12-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2017-12-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-01-06.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-01-13.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-01-20.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-01-27.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-02-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-02-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-02-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-02-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-03-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-03-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-03-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-03-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-03-31.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-04-07.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-04-14.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-04-21.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-04-28.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-05-05.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-05-12.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-05-19.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-05-26.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-06-02.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-06-09.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-06-16.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-06-23.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-06-30.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-07-07.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-07-14.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-07-21.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-07-28.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-08-04.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-08-11.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-08-18.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-08-25.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-09-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-09-08.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-09-15.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-09-22.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-09-29.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-10-06.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-10-13.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-10-20.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-10-27.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-11-03.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-11-10.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-11-17.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-11-24.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-12-01.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-12-08.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-12-15.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-12-22.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "image_2018-12-29.tiff (origin shape: (1205, 765, 3) -> rescale: (602, 382, 3))\n",
            "145 145 145\n",
            "--------------------INFO_DICT---------------------\n",
            "keys: dict_keys(['LastDayWeek', 'cases_medellin', 'Image', 'epi_week'])\n",
            "\n",
            "-------------------DENGUE CASES-------------------\n",
            "Max weekly dengue cases: 592\n",
            "Min weekly dengue cases: 12\n",
            "\n",
            "----------------------WEEKS-----------------------\n",
            "Max week: 2018-12-29\n",
            "Min week: 2016-01-09\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5Ngmo1j42xNP",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "d0f15000-e562-4764-b56e-3b6ac02f1a20"
      },
      "source": [
        "origin_X_train, y_train, origin_X_test, y_test = splitTrainTestSet(0.8)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "------------------------------------------------------------\n",
            "Total number of weeks:         145 weeks\n",
            "Training input:                (116, 602, 382, 3)\n",
            "Training output:               (116,)\n",
            "Testing input:                 (29, 602, 382, 3)\n",
            "Testing output:                (29,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r3H2ftvS3m1e"
      },
      "source": [
        "## Resized images without dimensionality reduction"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EGiMfyXp2-cS",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "5cb660b3-3f64-4d76-8c54-beaaac66c12e"
      },
      "source": [
        "# LR\n",
        "MAE, MAPE = classified_with_LR(origin_X_train, origin_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(origin_X_train, origin_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [-263734540560, 5990493315302, 5087699576081, -316787402706, -10298217423965, -8051446612973, -2518859611867, 650306461787, 3167768225805, -3863527111602, -1989119808204, -5641552997999, 1029883611107, -1623059347266, 456420129330, 597979034726, -2472634171748, -1415306658159, 7807246345234, 4237288852588, -2358614228165, -4142858818420, -5168702371026, 1795326904732, -5858308221088, 3820330306055, 1927554290535, -5182760891799, 1772604048908]\n",
            "    Mean = -787054797081.2759\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:  3431254873096.1724 (cases different in average)\n",
            "- MAPE: 158604687757.0361 (times different in average)\n",
            "\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [43, 43, 42, 42, 43, 44, 44, 43, 44, 43, 43, 45, 41, 41, 41, 42, 44, 43, 41, 43, 43, 42, 42, 42, 43, 44, 44, 44, 42]\n",
            "    Mean = 42.7931\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:   20.1724 (cases different in average)\n",
            "- MAPE:   1.0293 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DPeC32Ai38Ck"
      },
      "source": [
        "## Resized images with PCA per band"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jJ8KkyPx3H2l",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "9ea2416a-16c9-440e-eb1a-ae90d77459e7"
      },
      "source": [
        "rgb_pca_separate_X_train, rgb_pca_separate_X_test = pca_per_band(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (116, 689892)\n",
            "Resize shape    (116, 49)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (116, 689892)\n",
            "Resize shape    (116, 49)\n",
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (116, 689892)\n",
            "Resize shape    (116, 49)\n",
            "(116, 147) (29, 147)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-PadvSP_43SC",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "4bf2b943-caf7-49a0-d4d1-8a8947940ddd"
      },
      "source": [
        "# LR\n",
        "MAE, MAPE = classified_with_LR(rgb_pca_separate_X_train, rgb_pca_separate_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(rgb_pca_separate_X_train, rgb_pca_separate_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [190, 303, -39, 122, 90, 192, 305, 216, 255, 186, 107, 89, 71, 86, 114, 118, 161, 123, 102, 183, 104, 51, 27, 31, -30, 102, 247, 137, 127]\n",
            "    Mean = 130.0\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:  116.3448 (cases different in average)\n",
            "- MAPE:   5.5302 (times different in average)\n",
            "\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [43, 44, 41, 42, 44, 44, 44, 43, 45, 44, 43, 45, 41, 41, 41, 42, 45, 43, 41, 43, 43, 41, 42, 42, 43, 44, 44, 44, 42]\n",
            "    Mean = 42.8966\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:   20.2759 (cases different in average)\n",
            "- MAPE:   1.0333 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iKaWaoOQ4U79"
      },
      "source": [
        "## Resized images with PCA (all bands together)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qrkEScq63eJp",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "ae6b1ef3-b604-4cb6-d02b-086b5c7677cd"
      },
      "source": [
        "rgb_pca_together_X_train, rgb_pca_pca_together_X_test = dimension_reduct_with_PCA(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "================================== PRINCIPAL COMPONENT ANALYSIS  ===================================\n",
            "Origin shape    (116, 689892)\n",
            "Resize shape    (116, 49)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ny0drT8B44HI",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "8eb52f2f-7d9a-4ccd-89d1-0a4b2537a26a"
      },
      "source": [
        "# LR\n",
        "MAE, MAPE = classified_with_LR(rgb_pca_together_X_train, rgb_pca_pca_together_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(rgb_pca_together_X_train, rgb_pca_pca_together_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [190, 304, -39, 122, 90, 192, 305, 217, 255, 187, 107, 89, 70, 86, 114, 118, 162, 123, 102, 183, 104, 51, 27, 31, -30, 102, 246, 137, 127]\n",
            "    Mean = 130.069\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:  116.4138 (cases different in average)\n",
            "- MAPE:   5.5339 (times different in average)\n",
            "\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [43, 44, 41, 42, 44, 44, 44, 43, 45, 44, 43, 45, 41, 41, 41, 42, 45, 43, 41, 43, 43, 41, 42, 42, 43, 44, 44, 44, 42]\n",
            "    Mean = 42.8966\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:   20.2759 (cases different in average)\n",
            "- MAPE:   1.0333 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n3PI5Fa04aO-"
      },
      "source": [
        "## Resized images with LLE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4mP8FI5z3c56",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "49ae5e9e-2a5f-4a27-df74-fd2584843c8b"
      },
      "source": [
        "rgb_lle_X_train, rgb_lle_X_test = dimension_reduct_with_LLE(origin_X_train, origin_X_test, y_train)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "===================================== LOCALLY LINEAR EMBEDDING =====================================\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c5_ZLdAG440O",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "dff0f591-fac1-40b1-c5f5-e08cf15985b8"
      },
      "source": [
        "# LR\n",
        "MAE, MAPE = classified_with_LR(rgb_lle_X_train, rgb_lle_X_test, y_train, y_test)\n",
        "print('')\n",
        "\n",
        "# SVR \n",
        "MAE, MAPE = classified_with_SVR(rgb_lle_X_train, rgb_lle_X_test, y_train, y_test)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "----------------------------------------[Linear Regression]-----------------------------------------\n",
            "Predicted\n",
            "    List = [147, 123, 123, 123, 191, 187, 189, 123, 207, 138, 123, 232, 165, 165, 151, 151, 238, 158, 162, 157, 119, 145, 135, 121, 204, 165, 174, 130, 123]\n",
            "    Mean = 157.5517\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:   134.931 (cases different in average)\n",
            "- MAPE:    6.386 (times different in average)\n",
            "\n",
            "-----------------------------------------------[SVR]------------------------------------------------\n",
            "Predicted\n",
            "    List = [42, 42, 42, 42, 48, 47, 46, 42, 47, 42, 42, 48, 44, 44, 42, 42, 48, 43, 44, 43, 42, 42, 41, 42, 49, 44, 45, 41, 42]\n",
            "    Mean = 43.7241\n",
            "\n",
            "Real\n",
            "    List = [22 23 23 27 35 22 19 20 22 18 25 27 25 12 22 22 27 23 24 24 17 20 15 41\n",
            " 28 12 27 17 17]\n",
            "    Mean = 22.6207\n",
            "\n",
            "- MAE:   21.1034 (cases different in average)\n",
            "- MAPE:    1.065 (times different in average)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kXH-KST05Nan"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}