ANN_score_regression.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import KFold\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout\n",
    "from keras import optimizers\n",
    "import keras as K\n",
    "import skimage.measure\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.utils.class_weight import compute_sample_weight\n",
    "import tensorflow as tf\n",
    "from keras.callbacks import EarlyStopping, ReduceLROnPlateau, ModelCheckpoint, TensorBoard\n",
    "from sklearn.externals import joblib\n",
    "from sklearn.metrics import r2_score, mean_absolute_error, mean_squared_error, cohen_kappa_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_score = \"fibrosis_score_pathologist\"\n",
    "# target_score = \"ballooning_score_pathologist\"\n",
    "# target_score = \"inflammation_score_pathologist\"\n",
    "# target_score = \"steatosis_score_cv_based\"\n",
    "\n",
    "score_max = {}\n",
    "score_max[\"fibrosis_score_pathologist\"] = 4\n",
    "score_max[\"ballooning_score_pathologist\"] = 2\n",
    "score_max[\"steatosis_score_cv_based\"] = 3\n",
    "score_max[\"inflammation_score_pathologist\"] = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Base path\n",
    "base_path = \".//model_data//Human_liver_score_for_external_OSF_repo//\"\n",
    "\n",
    "# Pathologist ground truth\n",
    "ground_truth_full_file_name = os.path.join(base_path, \"ground_truth.csv\")\n",
    "\n",
    "# Path where ANN models and scaler files for normalization will be saved\n",
    "model_path = os.path.join(base_path, \"model//\")\n",
    "\n",
    "# Path where results from CNN classifications (per tile) are stored\n",
    "results_path = os.path.join(base_path, \"CNN//result\")\n",
    "\n",
    "# Location and filenames of individual CNN classification results\n",
    "if target_score == \"fibrosis_score_pathologist\":\n",
    "    full_results_file_name = \"Combined_Train_Liver_fibrosis_model_PT_V0.20_Fibrosis_score.csv\"\n",
    "    model_name = \"HS_Liver_fibrosis_model_PT_V0.20\"\n",
    "    target_score_column = \"fibrosis_score\"\n",
    "elif target_score == \"ballooning_score_pathologist\":\n",
    "    full_results_file_name = \"Combined_Train_Liver_ballooning_model_PT_V0.41_Ballooning_score.csv\"\n",
    "    model_name = \"Liver_ballooning_model_PT_V0.41\"\n",
    "    target_score_column = \"ballooning_score\"   \n",
    "elif target_score == \"inflammation_score_pathologist\":\n",
    "    full_results_file_name = \"Combined_Train_HS_Liver_inflammation_model_PT_V0.6_Inflammation_score.csv\"\n",
    "    model_name = \"Liver_inflammation_model_PT_V0.6\"\n",
    "    target_score_column = \"inflammation_score\"\n",
    "elif target_score == \"steatosis_score_cv_based\":\n",
    "    full_results_file_name = \"Combined_Train_HS_Liver_steatosis_model_IN_V1.2_Steatosis_score.csv\"\n",
    "    model_name = \"Liver_steatosis_model_IN_V1.2\"\n",
    "    target_score_column = \"steatosis_score\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "ground_truth = pd.read_csv(ground_truth_full_file_name, sep = \";\", decimal = \",\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Keep only columns of interest\n",
    "ground_truth = ground_truth[[\"id\", target_score_column]]\n",
    "\n",
    "ground_truth = ground_truth.dropna(axis = 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>id</th>\n",
       "      <th>fibrosis_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>462</th>\n",
       "      <td>462</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>463</th>\n",
       "      <td>463</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>464</th>\n",
       "      <td>464</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>465</th>\n",
       "      <td>465</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>466</th>\n",
       "      <td>466</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      id fibrosis_score\n",
       "462  462            2.0\n",
       "463  463            0.0\n",
       "464  464            2.0\n",
       "465  465            2.0\n",
       "466  466            0.0"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ground_truth.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "full_results = pd.read_csv(os.path.join(results_path, full_results_file_name), sep = \";\", decimal = \".\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>ignore</th>\n",
       "      <th>filenames</th>\n",
       "      <th>x</th>\n",
       "      <th>y</th>\n",
       "      <th>id</th>\n",
       "      <th>...</th>\n",
       "      <th>1_norm</th>\n",
       "      <th>1_norm_bin</th>\n",
       "      <th>2_norm</th>\n",
       "      <th>2_norm_bin</th>\n",
       "      <th>3_norm</th>\n",
       "      <th>3_norm_bin</th>\n",
       "      <th>4_norm</th>\n",
       "      <th>4_norm_bin</th>\n",
       "      <th>weighted_class</th>\n",
       "      <th>uncertainty</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>140049</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.98</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>466_9_5.png</td>\n",
       "      <td>9</td>\n",
       "      <td>5</td>\n",
       "      <td>466</td>\n",
       "      <td>...</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.98</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.98</td>\n",
       "      <td>0.02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140050</th>\n",
       "      <td>0.03</td>\n",
       "      <td>0.21</td>\n",
       "      <td>0.74</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>466_9_6.png</td>\n",
       "      <td>9</td>\n",
       "      <td>6</td>\n",
       "      <td>466</td>\n",
       "      <td>...</td>\n",
       "      <td>0.21</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.74</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.77</td>\n",
       "      <td>0.26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140051</th>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.73</td>\n",
       "      <td>0.21</td>\n",
       "      <td>0.05</td>\n",
       "      <td>0.0</td>\n",
       "      <td>466_9_7.png</td>\n",
       "      <td>9</td>\n",
       "      <td>7</td>\n",
       "      <td>466</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.73</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.21</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.05</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.30</td>\n",
       "      <td>0.27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140052</th>\n",
       "      <td>0.04</td>\n",
       "      <td>0.13</td>\n",
       "      <td>0.75</td>\n",
       "      <td>0.06</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>466_9_8.png</td>\n",
       "      <td>9</td>\n",
       "      <td>8</td>\n",
       "      <td>466</td>\n",
       "      <td>...</td>\n",
       "      <td>0.13</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.75</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.06</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.87</td>\n",
       "      <td>0.25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140053</th>\n",
       "      <td>0.83</td>\n",
       "      <td>0.17</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>466_9_9.png</td>\n",
       "      <td>9</td>\n",
       "      <td>9</td>\n",
       "      <td>466</td>\n",
       "      <td>...</td>\n",
       "      <td>0.17</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.17</td>\n",
       "      <td>0.17</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "           0     1     2     3     4  ignore    filenames  x  y   id  \\\n",
       "140049  0.01  0.01  0.98  0.00  0.00     0.0  466_9_5.png  9  5  466   \n",
       "140050  0.03  0.21  0.74  0.01  0.01     0.0  466_9_6.png  9  6  466   \n",
       "140051  0.00  0.00  0.73  0.21  0.05     0.0  466_9_7.png  9  7  466   \n",
       "140052  0.04  0.13  0.75  0.06  0.01     0.0  466_9_8.png  9  8  466   \n",
       "140053  0.83  0.17  0.00  0.00  0.00     0.0  466_9_9.png  9  9  466   \n",
       "\n",
       "           ...       1_norm  1_norm_bin  2_norm  2_norm_bin  3_norm  \\\n",
       "140049     ...         0.01         0.0    0.98         1.0    0.00   \n",
       "140050     ...         0.21         0.0    0.74         1.0    0.01   \n",
       "140051     ...         0.00         0.0    0.73         1.0    0.21   \n",
       "140052     ...         0.13         0.0    0.75         1.0    0.06   \n",
       "140053     ...         0.17         0.0    0.00         0.0    0.00   \n",
       "\n",
       "        3_norm_bin  4_norm  4_norm_bin  weighted_class  uncertainty  \n",
       "140049         0.0    0.00         0.0            1.98         0.02  \n",
       "140050         0.0    0.01         0.0            1.77         0.26  \n",
       "140051         0.0    0.05         0.0            2.30         0.27  \n",
       "140052         0.0    0.01         0.0            1.87         0.25  \n",
       "140053         0.0    0.00         0.0            0.17         0.17  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "full_results.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_scoring_ANN_feature_matrix(classification_result, scaler_file = \"\"):\n",
    "    \"\"\" Will take a full results dataframe (processed by renormalize_and_process_CNN_results)\n",
    "        and compute ANN features\n",
    "\n",
    "        Input:\n",
    "            classification_result (dataframe)\n",
    "            Optional: scaler_file: Full path to min max scaler. If empty result will not be scaled.\n",
    "\n",
    "        Result:\n",
    "            X, X_scaled (normal and scaled feature matrix)\n",
    "            slide_id: Dataframe with list of slide_id in exact order as X, X_scaled\n",
    "    \"\"\"\n",
    "\n",
    "    epsilon = 0.0000001\n",
    "\n",
    "    # Define input features to be used from classification_result\n",
    "    features = []\n",
    "    # All columns ending with \"_norm\"\n",
    "    for column in classification_result.columns:\n",
    "        if column.endswith(\"_norm\"):\n",
    "            features.append(column)\n",
    "\n",
    "    # Define output features (in X and X_scaled)\n",
    "    features_X = []\n",
    "    features_X.append(\"average_weighted_class\")\n",
    "    for feature in features:\n",
    "        features_X.append(feature+ \"_Shannon_entropy\")\n",
    "        features_X.append(feature + \"_mean\")\n",
    "    features_X.append(\"log_average_weighted_class\")\n",
    "\n",
    "    # Create empty dataframes\n",
    "    X = pd.DataFrame(columns= features_X)\n",
    "    X_scaled = pd.DataFrame(columns=features_X)\n",
    "\n",
    "    # Working list to build dataframe in loop as a list of dicts\n",
    "    X_list = []\n",
    "    slide_id_list = []\n",
    "\n",
    "    # Iterate over classification_result\n",
    "    for slide_id in classification_result[\"id\"].unique():\n",
    "        print(slide_id)\n",
    "        \n",
    "        slide_id_list.append({\"id\": slide_id})\n",
    "\n",
    "        classification_result_subset = classification_result[classification_result[\"id\"] == slide_id]\n",
    "\n",
    "        x_min = int(np.min(classification_result_subset[\"x\"]))\n",
    "        x_max = int(np.max(classification_result_subset[\"x\"]))\n",
    "        y_min = int(np.min(classification_result_subset[\"y\"]))\n",
    "        y_max = int(np.max(classification_result_subset[\"y\"]))\n",
    "\n",
    "        features_X_dict = {}\n",
    "        for feature in features:\n",
    "            # Create temporary 2d feature image to compute entropy with skimage\n",
    "            feature_img = np.empty(shape=(int(x_max - x_min + 1), int(y_max - y_min + 1)))\n",
    "            feature_img[:] = epsilon\n",
    "\n",
    "            for index_full_results, row_full_results in classification_result_subset.iterrows():\n",
    "                x = int(row_full_results[\"x\"] - x_min)\n",
    "                y = int(row_full_results[\"y\"] - y_min)\n",
    "\n",
    "                feature_img[x, y] = row_full_results[feature]\n",
    "\n",
    "                if feature_img[x, y] == 0:\n",
    "                    feature_img[x, y] = epsilon\n",
    "\n",
    "                if np.isnan(feature_img[x, y]):\n",
    "                    feature_img[x, y] = epsilon\n",
    "\n",
    "            # Entropy of p1, p2, ...\n",
    "            shannon_entropy = skimage.measure.shannon_entropy(feature_img)\n",
    "            features_X_dict[feature + \"_Shannon_entropy\"] = shannon_entropy\n",
    "\n",
    "            # Average feature of (p0, p1, ...\n",
    "            feature_mean = np.mean(classification_result_subset[feature])\n",
    "            features_X_dict[feature + \"_mean\"] = feature_mean\n",
    "\n",
    "        # Average of weighted class\n",
    "        average_weighted_class = np.mean(classification_result_subset[\"weighted_class\"])\n",
    "        features_X_dict[\"average_weighted_class\"] = average_weighted_class\n",
    "\n",
    "        # Log of average weighted class\n",
    "        log_average_weighted_class = np.log(average_weighted_class)\n",
    "        features_X_dict[\"log_average_weighted_class\"] = log_average_weighted_class\n",
    "\n",
    "        X_list.append(features_X_dict)\n",
    "\n",
    "    X = pd.DataFrame(X_list)\n",
    "\n",
    "    # Reorder columns\n",
    "    X = X[features_X]\n",
    "\n",
    "    # Load scaler\n",
    "    if (scaler_file):\n",
    "        scaler = joblib.load(scaler_file)\n",
    "\n",
    "        # transform X with previously saved min-max scaler\n",
    "        X_scaled = scaler.transform(X)\n",
    "    else:\n",
    "        X_scaled = []    \n",
    "    \n",
    "    slide_id_df = pd.DataFrame(slide_id_list)\n",
    "\n",
    "    return X, X_scaled, slide_id_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2\n",
      "3\n",
      "5\n",
      "6\n",
      "8\n",
      "9\n",
      "10\n",
      "11\n",
      "13\n",
      "14\n",
      "15\n",
      "17\n",
      "22\n",
      "20\n",
      "23\n",
      "27\n",
      "29\n",
      "31\n",
      "32\n",
      "33\n",
      "34\n",
      "37\n",
      "38\n",
      "41\n",
      "43\n",
      "44\n",
      "45\n",
      "47\n",
      "48\n",
      "50\n",
      "52\n",
      "55\n",
      "58\n",
      "60\n",
      "62\n",
      "135\n",
      "63\n",
      "64\n",
      "65\n",
      "66\n",
      "67\n",
      "68\n",
      "69\n",
      "70\n",
      "71\n",
      "73\n",
      "74\n",
      "75\n",
      "76\n",
      "77\n",
      "79\n",
      "80\n",
      "82\n",
      "83\n",
      "84\n",
      "85\n",
      "86\n",
      "87\n",
      "90\n",
      "92\n",
      "93\n",
      "94\n",
      "95\n",
      "96\n",
      "97\n",
      "98\n",
      "99\n",
      "102\n",
      "104\n",
      "105\n",
      "106\n",
      "110\n",
      "112\n",
      "113\n",
      "115\n",
      "116\n",
      "117\n",
      "118\n",
      "119\n",
      "120\n",
      "123\n",
      "124\n",
      "125\n",
      "126\n",
      "127\n",
      "128\n",
      "129\n",
      "130\n",
      "131\n",
      "132\n",
      "133\n",
      "134\n",
      "159\n",
      "160\n",
      "161\n",
      "162\n",
      "163\n",
      "167\n",
      "168\n",
      "169\n",
      "170\n",
      "171\n",
      "172\n",
      "173\n",
      "174\n",
      "151\n",
      "152\n",
      "153\n",
      "155\n",
      "156\n",
      "176\n",
      "177\n",
      "181\n",
      "183\n",
      "184\n",
      "187\n",
      "188\n",
      "194\n",
      "195\n",
      "198\n",
      "200\n",
      "204\n",
      "205\n",
      "209\n",
      "210\n",
      "215\n",
      "216\n",
      "212\n",
      "221\n",
      "223\n",
      "225\n",
      "228\n",
      "229\n",
      "230\n",
      "231\n",
      "232\n",
      "236\n",
      "237\n",
      "238\n",
      "240\n",
      "241\n",
      "243\n",
      "244\n",
      "246\n",
      "247\n",
      "249\n",
      "252\n",
      "253\n",
      "255\n",
      "256\n",
      "258\n",
      "259\n",
      "260\n",
      "262\n",
      "263\n",
      "264\n",
      "266\n",
      "267\n",
      "268\n",
      "269\n",
      "270\n",
      "271\n",
      "272\n",
      "273\n",
      "274\n",
      "275\n",
      "280\n",
      "282\n",
      "283\n",
      "285\n",
      "286\n",
      "288\n",
      "287\n",
      "291\n",
      "292\n",
      "293\n",
      "295\n",
      "297\n",
      "301\n",
      "302\n",
      "303\n",
      "304\n",
      "305\n",
      "307\n",
      "308\n",
      "312\n",
      "314\n",
      "316\n",
      "318\n",
      "321\n",
      "322\n",
      "323\n",
      "325\n",
      "326\n",
      "334\n",
      "352\n",
      "354\n",
      "357\n",
      "358\n",
      "360\n",
      "368\n",
      "369\n",
      "374\n",
      "375\n",
      "376\n",
      "377\n",
      "378\n",
      "379\n",
      "381\n",
      "382\n",
      "384\n",
      "385\n",
      "387\n",
      "389\n",
      "390\n",
      "392\n",
      "393\n",
      "394\n",
      "397\n",
      "396\n",
      "398\n",
      "399\n",
      "400\n",
      "401\n",
      "402\n",
      "405\n",
      "406\n",
      "407\n",
      "408\n",
      "410\n",
      "411\n",
      "416\n",
      "417\n",
      "418\n",
      "420\n",
      "421\n",
      "422\n",
      "423\n",
      "424\n",
      "425\n",
      "427\n",
      "429\n",
      "430\n",
      "431\n",
      "432\n",
      "433\n",
      "434\n",
      "435\n",
      "436\n",
      "437\n",
      "438\n",
      "440\n",
      "441\n",
      "442\n",
      "443\n",
      "444\n",
      "445\n",
      "446\n",
      "448\n",
      "449\n",
      "450\n",
      "451\n",
      "452\n",
      "453\n",
      "454\n",
      "455\n",
      "456\n",
      "457\n",
      "458\n",
      "459\n",
      "460\n",
      "463\n",
      "464\n",
      "465\n",
      "466\n"
     ]
    }
   ],
   "source": [
    "X, X_scaled, slide_id_df = get_scoring_ANN_feature_matrix(full_results)\n",
    "# Note: X_scaled is empty, since no scaler_file is available during training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "X[\"id\"] = slide_id_df[\"id\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(275, 13)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save unscaled weights, so that features do not need to be created each time\n",
    "X.to_csv(os.path.join(base_path, \"ANN//X\", full_results_file_name[:-4] + \"_X\" + \".csv\"), index = False, sep = \";\", decimal=\",\", float_format='%.4f')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Keep only X values where we have available ground truth for\n",
    "X = X[X[\"id\"].isin(ground_truth[\"id\"])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(225, 13)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Construct Y (ground truth) in the same order as X (features)\n",
    "Y_data = []\n",
    "\n",
    "for index, row in X.iterrows():\n",
    "    \n",
    "    slide_id_str = row[\"id\"]\n",
    "    score = int(float(ground_truth[ground_truth[\"id\"] == slide_id_str][target_score_column]))\n",
    "        \n",
    "    Y_data.append({\"id\": row[\"id\"], target_score_column : score})\n",
    "    \n",
    "Y = pd.DataFrame(Y_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_Y = X.merge(Y, on = [\"id\"])\n",
    "\n",
    "# Optional\n",
    "# X_Y.to_csv(os.path.join(model_path, full_results_file_name[:-4] + \"_X_Y\" + \".csv\"), index = False, sep = \";\", decimal=\",\", float_format='%.4f')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>average_weighted_class</th>\n",
       "      <th>0_norm_Shannon_entropy</th>\n",
       "      <th>0_norm_mean</th>\n",
       "      <th>1_norm_Shannon_entropy</th>\n",
       "      <th>1_norm_mean</th>\n",
       "      <th>2_norm_Shannon_entropy</th>\n",
       "      <th>2_norm_mean</th>\n",
       "      <th>3_norm_Shannon_entropy</th>\n",
       "      <th>3_norm_mean</th>\n",
       "      <th>4_norm_Shannon_entropy</th>\n",
       "      <th>4_norm_mean</th>\n",
       "      <th>log_average_weighted_class</th>\n",
       "      <th>id</th>\n",
       "      <th>fibrosis_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1.171035</td>\n",
       "      <td>7.899933</td>\n",
       "      <td>0.286520</td>\n",
       "      <td>8.404122</td>\n",
       "      <td>0.422907</td>\n",
       "      <td>7.411710</td>\n",
       "      <td>0.184978</td>\n",
       "      <td>6.709523</td>\n",
       "      <td>0.042863</td>\n",
       "      <td>6.362523</td>\n",
       "      <td>0.061784</td>\n",
       "      <td>0.157888</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.864512</td>\n",
       "      <td>6.838021</td>\n",
       "      <td>0.485793</td>\n",
       "      <td>6.611893</td>\n",
       "      <td>0.240549</td>\n",
       "      <td>5.904274</td>\n",
       "      <td>0.227134</td>\n",
       "      <td>4.216028</td>\n",
       "      <td>0.013293</td>\n",
       "      <td>3.881828</td>\n",
       "      <td>0.031098</td>\n",
       "      <td>-0.145590</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.064637</td>\n",
       "      <td>7.306966</td>\n",
       "      <td>0.189055</td>\n",
       "      <td>7.891524</td>\n",
       "      <td>0.272066</td>\n",
       "      <td>7.206297</td>\n",
       "      <td>0.125626</td>\n",
       "      <td>7.357105</td>\n",
       "      <td>0.108176</td>\n",
       "      <td>7.731327</td>\n",
       "      <td>0.303692</td>\n",
       "      <td>0.724955</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.278146</td>\n",
       "      <td>7.320232</td>\n",
       "      <td>0.169000</td>\n",
       "      <td>8.375098</td>\n",
       "      <td>0.618171</td>\n",
       "      <td>6.603034</td>\n",
       "      <td>0.063341</td>\n",
       "      <td>6.199967</td>\n",
       "      <td>0.063317</td>\n",
       "      <td>6.349997</td>\n",
       "      <td>0.085146</td>\n",
       "      <td>0.245411</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.625061</td>\n",
       "      <td>8.846545</td>\n",
       "      <td>0.230677</td>\n",
       "      <td>9.026727</td>\n",
       "      <td>0.164896</td>\n",
       "      <td>9.456843</td>\n",
       "      <td>0.410174</td>\n",
       "      <td>8.548836</td>\n",
       "      <td>0.135313</td>\n",
       "      <td>7.689018</td>\n",
       "      <td>0.057717</td>\n",
       "      <td>0.485545</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1.044597</td>\n",
       "      <td>7.739146</td>\n",
       "      <td>0.266650</td>\n",
       "      <td>8.332433</td>\n",
       "      <td>0.540220</td>\n",
       "      <td>6.788715</td>\n",
       "      <td>0.105477</td>\n",
       "      <td>6.164989</td>\n",
       "      <td>0.055648</td>\n",
       "      <td>5.665417</td>\n",
       "      <td>0.030660</td>\n",
       "      <td>0.043631</td>\n",
       "      <td>14</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1.501382</td>\n",
       "      <td>5.940131</td>\n",
       "      <td>0.315366</td>\n",
       "      <td>5.966079</td>\n",
       "      <td>0.235854</td>\n",
       "      <td>5.836206</td>\n",
       "      <td>0.235041</td>\n",
       "      <td>5.024997</td>\n",
       "      <td>0.056504</td>\n",
       "      <td>5.278880</td>\n",
       "      <td>0.155691</td>\n",
       "      <td>0.406386</td>\n",
       "      <td>15</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1.268579</td>\n",
       "      <td>8.265106</td>\n",
       "      <td>0.331241</td>\n",
       "      <td>8.507873</td>\n",
       "      <td>0.345432</td>\n",
       "      <td>7.540641</td>\n",
       "      <td>0.143561</td>\n",
       "      <td>6.960323</td>\n",
       "      <td>0.081259</td>\n",
       "      <td>6.926848</td>\n",
       "      <td>0.097230</td>\n",
       "      <td>0.237897</td>\n",
       "      <td>23</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1.312350</td>\n",
       "      <td>7.890483</td>\n",
       "      <td>0.345092</td>\n",
       "      <td>8.072578</td>\n",
       "      <td>0.299263</td>\n",
       "      <td>7.271297</td>\n",
       "      <td>0.156544</td>\n",
       "      <td>6.868453</td>\n",
       "      <td>0.095207</td>\n",
       "      <td>6.571061</td>\n",
       "      <td>0.103018</td>\n",
       "      <td>0.271820</td>\n",
       "      <td>31</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.958239</td>\n",
       "      <td>9.323522</td>\n",
       "      <td>0.388478</td>\n",
       "      <td>9.501269</td>\n",
       "      <td>0.386080</td>\n",
       "      <td>8.253264</td>\n",
       "      <td>0.132259</td>\n",
       "      <td>7.512133</td>\n",
       "      <td>0.063542</td>\n",
       "      <td>6.801179</td>\n",
       "      <td>0.028239</td>\n",
       "      <td>-0.042658</td>\n",
       "      <td>32</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.669716</td>\n",
       "      <td>8.220747</td>\n",
       "      <td>0.433721</td>\n",
       "      <td>8.353835</td>\n",
       "      <td>0.495685</td>\n",
       "      <td>5.462923</td>\n",
       "      <td>0.042067</td>\n",
       "      <td>5.288560</td>\n",
       "      <td>0.022429</td>\n",
       "      <td>4.301908</td>\n",
       "      <td>0.004832</td>\n",
       "      <td>-0.400902</td>\n",
       "      <td>33</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.896197</td>\n",
       "      <td>8.850508</td>\n",
       "      <td>0.388632</td>\n",
       "      <td>9.067475</td>\n",
       "      <td>0.437464</td>\n",
       "      <td>7.256276</td>\n",
       "      <td>0.096567</td>\n",
       "      <td>7.006479</td>\n",
       "      <td>0.042650</td>\n",
       "      <td>6.221441</td>\n",
       "      <td>0.033590</td>\n",
       "      <td>-0.109595</td>\n",
       "      <td>34</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1.666501</td>\n",
       "      <td>8.263145</td>\n",
       "      <td>0.274544</td>\n",
       "      <td>8.546743</td>\n",
       "      <td>0.300066</td>\n",
       "      <td>7.731857</td>\n",
       "      <td>0.112239</td>\n",
       "      <td>7.556725</td>\n",
       "      <td>0.109171</td>\n",
       "      <td>7.767403</td>\n",
       "      <td>0.202902</td>\n",
       "      <td>0.510726</td>\n",
       "      <td>41</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1.549282</td>\n",
       "      <td>7.510776</td>\n",
       "      <td>0.235223</td>\n",
       "      <td>7.975706</td>\n",
       "      <td>0.359356</td>\n",
       "      <td>7.111262</td>\n",
       "      <td>0.168713</td>\n",
       "      <td>6.966382</td>\n",
       "      <td>0.091510</td>\n",
       "      <td>6.778723</td>\n",
       "      <td>0.143564</td>\n",
       "      <td>0.437792</td>\n",
       "      <td>43</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0.737426</td>\n",
       "      <td>8.547742</td>\n",
       "      <td>0.474317</td>\n",
       "      <td>8.520286</td>\n",
       "      <td>0.367287</td>\n",
       "      <td>6.362006</td>\n",
       "      <td>0.112040</td>\n",
       "      <td>5.753966</td>\n",
       "      <td>0.036832</td>\n",
       "      <td>4.438861</td>\n",
       "      <td>0.008000</td>\n",
       "      <td>-0.304590</td>\n",
       "      <td>44</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.597224</td>\n",
       "      <td>8.647000</td>\n",
       "      <td>0.524134</td>\n",
       "      <td>8.576402</td>\n",
       "      <td>0.405039</td>\n",
       "      <td>5.712645</td>\n",
       "      <td>0.038642</td>\n",
       "      <td>5.106374</td>\n",
       "      <td>0.013209</td>\n",
       "      <td>4.306042</td>\n",
       "      <td>0.018091</td>\n",
       "      <td>-0.515462</td>\n",
       "      <td>50</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>0.531950</td>\n",
       "      <td>7.902670</td>\n",
       "      <td>0.636454</td>\n",
       "      <td>7.571775</td>\n",
       "      <td>0.211241</td>\n",
       "      <td>6.433968</td>\n",
       "      <td>0.141348</td>\n",
       "      <td>3.182311</td>\n",
       "      <td>0.003333</td>\n",
       "      <td>4.376511</td>\n",
       "      <td>0.005780</td>\n",
       "      <td>-0.631205</td>\n",
       "      <td>52</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>0.879661</td>\n",
       "      <td>7.698441</td>\n",
       "      <td>0.409864</td>\n",
       "      <td>7.802700</td>\n",
       "      <td>0.351017</td>\n",
       "      <td>6.992631</td>\n",
       "      <td>0.206102</td>\n",
       "      <td>5.112219</td>\n",
       "      <td>0.012814</td>\n",
       "      <td>5.411873</td>\n",
       "      <td>0.018441</td>\n",
       "      <td>-0.128219</td>\n",
       "      <td>55</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>1.176135</td>\n",
       "      <td>8.029264</td>\n",
       "      <td>0.314214</td>\n",
       "      <td>8.332981</td>\n",
       "      <td>0.391004</td>\n",
       "      <td>7.289096</td>\n",
       "      <td>0.159061</td>\n",
       "      <td>6.709758</td>\n",
       "      <td>0.075000</td>\n",
       "      <td>6.126293</td>\n",
       "      <td>0.059716</td>\n",
       "      <td>0.162234</td>\n",
       "      <td>58</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0.885543</td>\n",
       "      <td>8.210233</td>\n",
       "      <td>0.474525</td>\n",
       "      <td>8.136216</td>\n",
       "      <td>0.284502</td>\n",
       "      <td>7.151931</td>\n",
       "      <td>0.171629</td>\n",
       "      <td>6.131654</td>\n",
       "      <td>0.018959</td>\n",
       "      <td>5.969771</td>\n",
       "      <td>0.049367</td>\n",
       "      <td>-0.121554</td>\n",
       "      <td>60</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>0.270529</td>\n",
       "      <td>7.633711</td>\n",
       "      <td>0.778365</td>\n",
       "      <td>7.130560</td>\n",
       "      <td>0.186010</td>\n",
       "      <td>4.871919</td>\n",
       "      <td>0.026298</td>\n",
       "      <td>2.841580</td>\n",
       "      <td>0.002404</td>\n",
       "      <td>2.753089</td>\n",
       "      <td>0.005144</td>\n",
       "      <td>-1.307377</td>\n",
       "      <td>62</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>1.083285</td>\n",
       "      <td>6.495710</td>\n",
       "      <td>0.404672</td>\n",
       "      <td>6.601895</td>\n",
       "      <td>0.230219</td>\n",
       "      <td>6.170636</td>\n",
       "      <td>0.285474</td>\n",
       "      <td>5.163017</td>\n",
       "      <td>0.033796</td>\n",
       "      <td>4.927565</td>\n",
       "      <td>0.044088</td>\n",
       "      <td>0.079998</td>\n",
       "      <td>135</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>0.654049</td>\n",
       "      <td>6.969939</td>\n",
       "      <td>0.646319</td>\n",
       "      <td>6.259954</td>\n",
       "      <td>0.111043</td>\n",
       "      <td>5.637852</td>\n",
       "      <td>0.196442</td>\n",
       "      <td>3.954726</td>\n",
       "      <td>0.032699</td>\n",
       "      <td>3.672541</td>\n",
       "      <td>0.012515</td>\n",
       "      <td>-0.424573</td>\n",
       "      <td>63</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>0.441067</td>\n",
       "      <td>6.083616</td>\n",
       "      <td>0.724400</td>\n",
       "      <td>5.653626</td>\n",
       "      <td>0.190400</td>\n",
       "      <td>3.064283</td>\n",
       "      <td>0.029067</td>\n",
       "      <td>2.673208</td>\n",
       "      <td>0.029200</td>\n",
       "      <td>2.080029</td>\n",
       "      <td>0.025200</td>\n",
       "      <td>-0.818559</td>\n",
       "      <td>64</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>1.122941</td>\n",
       "      <td>3.505183</td>\n",
       "      <td>0.375294</td>\n",
       "      <td>3.678104</td>\n",
       "      <td>0.328824</td>\n",
       "      <td>2.747662</td>\n",
       "      <td>0.161765</td>\n",
       "      <td>2.075510</td>\n",
       "      <td>0.058824</td>\n",
       "      <td>2.083333</td>\n",
       "      <td>0.071765</td>\n",
       "      <td>0.115951</td>\n",
       "      <td>65</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>0.697549</td>\n",
       "      <td>7.286471</td>\n",
       "      <td>0.588725</td>\n",
       "      <td>6.891501</td>\n",
       "      <td>0.204804</td>\n",
       "      <td>5.713699</td>\n",
       "      <td>0.144804</td>\n",
       "      <td>4.771375</td>\n",
       "      <td>0.040980</td>\n",
       "      <td>3.690693</td>\n",
       "      <td>0.019118</td>\n",
       "      <td>-0.360182</td>\n",
       "      <td>66</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>0.951235</td>\n",
       "      <td>5.671804</td>\n",
       "      <td>0.362716</td>\n",
       "      <td>5.906129</td>\n",
       "      <td>0.436667</td>\n",
       "      <td>4.188393</td>\n",
       "      <td>0.119012</td>\n",
       "      <td>4.147420</td>\n",
       "      <td>0.047901</td>\n",
       "      <td>2.143010</td>\n",
       "      <td>0.031975</td>\n",
       "      <td>-0.049995</td>\n",
       "      <td>67</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0.849106</td>\n",
       "      <td>6.306970</td>\n",
       "      <td>0.385366</td>\n",
       "      <td>6.494167</td>\n",
       "      <td>0.437967</td>\n",
       "      <td>4.872470</td>\n",
       "      <td>0.128618</td>\n",
       "      <td>4.127193</td>\n",
       "      <td>0.035610</td>\n",
       "      <td>1.579861</td>\n",
       "      <td>0.010894</td>\n",
       "      <td>-0.163572</td>\n",
       "      <td>68</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>0.721748</td>\n",
       "      <td>6.307881</td>\n",
       "      <td>0.491748</td>\n",
       "      <td>6.266904</td>\n",
       "      <td>0.375437</td>\n",
       "      <td>4.485850</td>\n",
       "      <td>0.072039</td>\n",
       "      <td>4.232194</td>\n",
       "      <td>0.036602</td>\n",
       "      <td>3.432663</td>\n",
       "      <td>0.021650</td>\n",
       "      <td>-0.326080</td>\n",
       "      <td>69</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>0.949704</td>\n",
       "      <td>7.430811</td>\n",
       "      <td>0.479593</td>\n",
       "      <td>7.230749</td>\n",
       "      <td>0.243741</td>\n",
       "      <td>6.093404</td>\n",
       "      <td>0.128889</td>\n",
       "      <td>6.318554</td>\n",
       "      <td>0.141000</td>\n",
       "      <td>5.057025</td>\n",
       "      <td>0.005407</td>\n",
       "      <td>-0.051605</td>\n",
       "      <td>70</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195</th>\n",
       "      <td>2.090262</td>\n",
       "      <td>9.285171</td>\n",
       "      <td>0.100031</td>\n",
       "      <td>10.267086</td>\n",
       "      <td>0.339982</td>\n",
       "      <td>9.821386</td>\n",
       "      <td>0.188091</td>\n",
       "      <td>9.437234</td>\n",
       "      <td>0.111163</td>\n",
       "      <td>9.831850</td>\n",
       "      <td>0.259734</td>\n",
       "      <td>0.737289</td>\n",
       "      <td>433</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>1.505041</td>\n",
       "      <td>6.092183</td>\n",
       "      <td>0.422231</td>\n",
       "      <td>5.746054</td>\n",
       "      <td>0.116364</td>\n",
       "      <td>5.362013</td>\n",
       "      <td>0.141405</td>\n",
       "      <td>5.086922</td>\n",
       "      <td>0.172231</td>\n",
       "      <td>4.783493</td>\n",
       "      <td>0.146777</td>\n",
       "      <td>0.408820</td>\n",
       "      <td>434</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>197</th>\n",
       "      <td>2.518696</td>\n",
       "      <td>5.246733</td>\n",
       "      <td>0.130821</td>\n",
       "      <td>5.235094</td>\n",
       "      <td>0.071401</td>\n",
       "      <td>5.571376</td>\n",
       "      <td>0.064348</td>\n",
       "      <td>7.271385</td>\n",
       "      <td>0.611691</td>\n",
       "      <td>5.874781</td>\n",
       "      <td>0.120193</td>\n",
       "      <td>0.923741</td>\n",
       "      <td>435</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>2.890822</td>\n",
       "      <td>3.562763</td>\n",
       "      <td>0.027534</td>\n",
       "      <td>4.660997</td>\n",
       "      <td>0.091027</td>\n",
       "      <td>5.500514</td>\n",
       "      <td>0.078425</td>\n",
       "      <td>6.802402</td>\n",
       "      <td>0.565890</td>\n",
       "      <td>6.045024</td>\n",
       "      <td>0.235548</td>\n",
       "      <td>1.061541</td>\n",
       "      <td>436</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>199</th>\n",
       "      <td>2.280925</td>\n",
       "      <td>7.275477</td>\n",
       "      <td>0.210058</td>\n",
       "      <td>7.145018</td>\n",
       "      <td>0.096821</td>\n",
       "      <td>6.848450</td>\n",
       "      <td>0.064335</td>\n",
       "      <td>8.248507</td>\n",
       "      <td>0.458092</td>\n",
       "      <td>7.380515</td>\n",
       "      <td>0.169884</td>\n",
       "      <td>0.824581</td>\n",
       "      <td>437</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200</th>\n",
       "      <td>1.959484</td>\n",
       "      <td>7.639567</td>\n",
       "      <td>0.234217</td>\n",
       "      <td>7.755186</td>\n",
       "      <td>0.212910</td>\n",
       "      <td>6.376130</td>\n",
       "      <td>0.040258</td>\n",
       "      <td>8.074537</td>\n",
       "      <td>0.383260</td>\n",
       "      <td>6.751585</td>\n",
       "      <td>0.128490</td>\n",
       "      <td>0.672681</td>\n",
       "      <td>438</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>201</th>\n",
       "      <td>1.360223</td>\n",
       "      <td>7.424756</td>\n",
       "      <td>0.108953</td>\n",
       "      <td>8.369104</td>\n",
       "      <td>0.550089</td>\n",
       "      <td>7.527496</td>\n",
       "      <td>0.240267</td>\n",
       "      <td>6.297025</td>\n",
       "      <td>0.071693</td>\n",
       "      <td>5.083543</td>\n",
       "      <td>0.027550</td>\n",
       "      <td>0.307648</td>\n",
       "      <td>440</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>202</th>\n",
       "      <td>2.013923</td>\n",
       "      <td>6.710634</td>\n",
       "      <td>0.036147</td>\n",
       "      <td>8.240911</td>\n",
       "      <td>0.322067</td>\n",
       "      <td>8.485223</td>\n",
       "      <td>0.343765</td>\n",
       "      <td>7.946560</td>\n",
       "      <td>0.183695</td>\n",
       "      <td>7.321047</td>\n",
       "      <td>0.112697</td>\n",
       "      <td>0.700085</td>\n",
       "      <td>441</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>203</th>\n",
       "      <td>1.301373</td>\n",
       "      <td>7.286630</td>\n",
       "      <td>0.143235</td>\n",
       "      <td>7.839782</td>\n",
       "      <td>0.517157</td>\n",
       "      <td>7.100898</td>\n",
       "      <td>0.263203</td>\n",
       "      <td>5.888236</td>\n",
       "      <td>0.047157</td>\n",
       "      <td>4.935821</td>\n",
       "      <td>0.028007</td>\n",
       "      <td>0.263420</td>\n",
       "      <td>442</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>204</th>\n",
       "      <td>0.888379</td>\n",
       "      <td>10.342842</td>\n",
       "      <td>0.533790</td>\n",
       "      <td>10.073810</td>\n",
       "      <td>0.245542</td>\n",
       "      <td>8.311780</td>\n",
       "      <td>0.078898</td>\n",
       "      <td>8.551946</td>\n",
       "      <td>0.079971</td>\n",
       "      <td>7.947315</td>\n",
       "      <td>0.060394</td>\n",
       "      <td>-0.118357</td>\n",
       "      <td>443</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>205</th>\n",
       "      <td>2.109250</td>\n",
       "      <td>9.000184</td>\n",
       "      <td>0.088162</td>\n",
       "      <td>9.871528</td>\n",
       "      <td>0.306083</td>\n",
       "      <td>8.888917</td>\n",
       "      <td>0.083196</td>\n",
       "      <td>10.222826</td>\n",
       "      <td>0.450495</td>\n",
       "      <td>8.313079</td>\n",
       "      <td>0.070730</td>\n",
       "      <td>0.746332</td>\n",
       "      <td>444</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>206</th>\n",
       "      <td>1.847811</td>\n",
       "      <td>6.001343</td>\n",
       "      <td>0.230533</td>\n",
       "      <td>6.232854</td>\n",
       "      <td>0.230888</td>\n",
       "      <td>5.215456</td>\n",
       "      <td>0.084024</td>\n",
       "      <td>6.364617</td>\n",
       "      <td>0.366627</td>\n",
       "      <td>4.886858</td>\n",
       "      <td>0.086923</td>\n",
       "      <td>0.614002</td>\n",
       "      <td>445</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>207</th>\n",
       "      <td>2.964043</td>\n",
       "      <td>5.257020</td>\n",
       "      <td>0.098303</td>\n",
       "      <td>5.463536</td>\n",
       "      <td>0.078773</td>\n",
       "      <td>4.758411</td>\n",
       "      <td>0.020794</td>\n",
       "      <td>7.137511</td>\n",
       "      <td>0.362310</td>\n",
       "      <td>7.265979</td>\n",
       "      <td>0.438809</td>\n",
       "      <td>1.086554</td>\n",
       "      <td>446</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208</th>\n",
       "      <td>0.861590</td>\n",
       "      <td>9.833957</td>\n",
       "      <td>0.545566</td>\n",
       "      <td>9.413423</td>\n",
       "      <td>0.179205</td>\n",
       "      <td>8.748819</td>\n",
       "      <td>0.170573</td>\n",
       "      <td>8.251703</td>\n",
       "      <td>0.075352</td>\n",
       "      <td>6.718198</td>\n",
       "      <td>0.027951</td>\n",
       "      <td>-0.148976</td>\n",
       "      <td>448</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>209</th>\n",
       "      <td>1.611011</td>\n",
       "      <td>7.249514</td>\n",
       "      <td>0.042169</td>\n",
       "      <td>8.577903</td>\n",
       "      <td>0.541085</td>\n",
       "      <td>7.922692</td>\n",
       "      <td>0.208474</td>\n",
       "      <td>7.500894</td>\n",
       "      <td>0.177004</td>\n",
       "      <td>5.759305</td>\n",
       "      <td>0.029798</td>\n",
       "      <td>0.476862</td>\n",
       "      <td>449</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>210</th>\n",
       "      <td>1.676016</td>\n",
       "      <td>8.471461</td>\n",
       "      <td>0.020761</td>\n",
       "      <td>10.274089</td>\n",
       "      <td>0.500485</td>\n",
       "      <td>9.907630</td>\n",
       "      <td>0.306834</td>\n",
       "      <td>9.026972</td>\n",
       "      <td>0.122150</td>\n",
       "      <td>8.144387</td>\n",
       "      <td>0.048008</td>\n",
       "      <td>0.516420</td>\n",
       "      <td>450</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>211</th>\n",
       "      <td>1.673405</td>\n",
       "      <td>9.005006</td>\n",
       "      <td>0.085238</td>\n",
       "      <td>10.067114</td>\n",
       "      <td>0.437861</td>\n",
       "      <td>9.741052</td>\n",
       "      <td>0.273496</td>\n",
       "      <td>8.801484</td>\n",
       "      <td>0.123472</td>\n",
       "      <td>8.179657</td>\n",
       "      <td>0.078631</td>\n",
       "      <td>0.514860</td>\n",
       "      <td>451</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>212</th>\n",
       "      <td>2.838242</td>\n",
       "      <td>6.446547</td>\n",
       "      <td>0.029078</td>\n",
       "      <td>7.749261</td>\n",
       "      <td>0.163162</td>\n",
       "      <td>7.424019</td>\n",
       "      <td>0.056495</td>\n",
       "      <td>9.100138</td>\n",
       "      <td>0.439153</td>\n",
       "      <td>8.679939</td>\n",
       "      <td>0.310782</td>\n",
       "      <td>1.043185</td>\n",
       "      <td>452</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>213</th>\n",
       "      <td>1.626563</td>\n",
       "      <td>9.167279</td>\n",
       "      <td>0.175471</td>\n",
       "      <td>9.913418</td>\n",
       "      <td>0.358950</td>\n",
       "      <td>9.476872</td>\n",
       "      <td>0.233437</td>\n",
       "      <td>9.077094</td>\n",
       "      <td>0.124955</td>\n",
       "      <td>8.649745</td>\n",
       "      <td>0.105771</td>\n",
       "      <td>0.486469</td>\n",
       "      <td>453</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214</th>\n",
       "      <td>1.505392</td>\n",
       "      <td>8.994395</td>\n",
       "      <td>0.036615</td>\n",
       "      <td>10.377234</td>\n",
       "      <td>0.551991</td>\n",
       "      <td>9.934749</td>\n",
       "      <td>0.309342</td>\n",
       "      <td>8.620702</td>\n",
       "      <td>0.071465</td>\n",
       "      <td>7.759743</td>\n",
       "      <td>0.029040</td>\n",
       "      <td>0.409053</td>\n",
       "      <td>454</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>215</th>\n",
       "      <td>1.447921</td>\n",
       "      <td>8.297596</td>\n",
       "      <td>0.083002</td>\n",
       "      <td>9.590411</td>\n",
       "      <td>0.547491</td>\n",
       "      <td>8.755449</td>\n",
       "      <td>0.239731</td>\n",
       "      <td>7.753385</td>\n",
       "      <td>0.095000</td>\n",
       "      <td>6.666367</td>\n",
       "      <td>0.033109</td>\n",
       "      <td>0.370129</td>\n",
       "      <td>455</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>216</th>\n",
       "      <td>1.619685</td>\n",
       "      <td>8.971560</td>\n",
       "      <td>0.240291</td>\n",
       "      <td>9.279237</td>\n",
       "      <td>0.196099</td>\n",
       "      <td>9.453570</td>\n",
       "      <td>0.371704</td>\n",
       "      <td>8.323006</td>\n",
       "      <td>0.085372</td>\n",
       "      <td>8.080890</td>\n",
       "      <td>0.105331</td>\n",
       "      <td>0.482232</td>\n",
       "      <td>456</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>217</th>\n",
       "      <td>2.294710</td>\n",
       "      <td>8.539161</td>\n",
       "      <td>0.027746</td>\n",
       "      <td>9.492779</td>\n",
       "      <td>0.325242</td>\n",
       "      <td>9.452364</td>\n",
       "      <td>0.184726</td>\n",
       "      <td>9.445888</td>\n",
       "      <td>0.247488</td>\n",
       "      <td>9.047114</td>\n",
       "      <td>0.214130</td>\n",
       "      <td>0.830607</td>\n",
       "      <td>457</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>218</th>\n",
       "      <td>1.668948</td>\n",
       "      <td>9.227372</td>\n",
       "      <td>0.057657</td>\n",
       "      <td>10.490767</td>\n",
       "      <td>0.461275</td>\n",
       "      <td>10.175851</td>\n",
       "      <td>0.296798</td>\n",
       "      <td>9.197765</td>\n",
       "      <td>0.120210</td>\n",
       "      <td>8.524394</td>\n",
       "      <td>0.062258</td>\n",
       "      <td>0.512194</td>\n",
       "      <td>458</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>219</th>\n",
       "      <td>2.102887</td>\n",
       "      <td>7.554920</td>\n",
       "      <td>0.149278</td>\n",
       "      <td>8.146013</td>\n",
       "      <td>0.148714</td>\n",
       "      <td>8.595156</td>\n",
       "      <td>0.286811</td>\n",
       "      <td>8.476915</td>\n",
       "      <td>0.276535</td>\n",
       "      <td>7.486051</td>\n",
       "      <td>0.137178</td>\n",
       "      <td>0.743311</td>\n",
       "      <td>459</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>220</th>\n",
       "      <td>1.700295</td>\n",
       "      <td>6.978095</td>\n",
       "      <td>0.031609</td>\n",
       "      <td>7.865264</td>\n",
       "      <td>0.456783</td>\n",
       "      <td>7.717421</td>\n",
       "      <td>0.345496</td>\n",
       "      <td>6.604886</td>\n",
       "      <td>0.107426</td>\n",
       "      <td>6.022977</td>\n",
       "      <td>0.056595</td>\n",
       "      <td>0.530802</td>\n",
       "      <td>460</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>221</th>\n",
       "      <td>1.050904</td>\n",
       "      <td>10.134197</td>\n",
       "      <td>0.382277</td>\n",
       "      <td>10.266688</td>\n",
       "      <td>0.372040</td>\n",
       "      <td>8.806283</td>\n",
       "      <td>0.129169</td>\n",
       "      <td>7.902775</td>\n",
       "      <td>0.042825</td>\n",
       "      <td>7.860285</td>\n",
       "      <td>0.072181</td>\n",
       "      <td>0.049651</td>\n",
       "      <td>463</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>222</th>\n",
       "      <td>1.299268</td>\n",
       "      <td>10.346409</td>\n",
       "      <td>0.148433</td>\n",
       "      <td>11.107453</td>\n",
       "      <td>0.551112</td>\n",
       "      <td>10.010084</td>\n",
       "      <td>0.212190</td>\n",
       "      <td>8.599022</td>\n",
       "      <td>0.027276</td>\n",
       "      <td>8.301907</td>\n",
       "      <td>0.059535</td>\n",
       "      <td>0.261801</td>\n",
       "      <td>464</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>223</th>\n",
       "      <td>1.493227</td>\n",
       "      <td>9.565596</td>\n",
       "      <td>0.153262</td>\n",
       "      <td>10.205015</td>\n",
       "      <td>0.457333</td>\n",
       "      <td>9.358499</td>\n",
       "      <td>0.217881</td>\n",
       "      <td>8.527567</td>\n",
       "      <td>0.084145</td>\n",
       "      <td>8.036458</td>\n",
       "      <td>0.085906</td>\n",
       "      <td>0.400940</td>\n",
       "      <td>465</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224</th>\n",
       "      <td>1.042471</td>\n",
       "      <td>10.531901</td>\n",
       "      <td>0.377851</td>\n",
       "      <td>10.622069</td>\n",
       "      <td>0.363683</td>\n",
       "      <td>9.277570</td>\n",
       "      <td>0.154379</td>\n",
       "      <td>8.206935</td>\n",
       "      <td>0.044607</td>\n",
       "      <td>7.955018</td>\n",
       "      <td>0.058277</td>\n",
       "      <td>0.041594</td>\n",
       "      <td>466</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>225 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     average_weighted_class  0_norm_Shannon_entropy  0_norm_mean  \\\n",
       "0                  1.171035                7.899933     0.286520   \n",
       "1                  0.864512                6.838021     0.485793   \n",
       "2                  2.064637                7.306966     0.189055   \n",
       "3                  1.278146                7.320232     0.169000   \n",
       "4                  1.625061                8.846545     0.230677   \n",
       "5                  1.044597                7.739146     0.266650   \n",
       "6                  1.501382                5.940131     0.315366   \n",
       "7                  1.268579                8.265106     0.331241   \n",
       "8                  1.312350                7.890483     0.345092   \n",
       "9                  0.958239                9.323522     0.388478   \n",
       "10                 0.669716                8.220747     0.433721   \n",
       "11                 0.896197                8.850508     0.388632   \n",
       "12                 1.666501                8.263145     0.274544   \n",
       "13                 1.549282                7.510776     0.235223   \n",
       "14                 0.737426                8.547742     0.474317   \n",
       "15                 0.597224                8.647000     0.524134   \n",
       "16                 0.531950                7.902670     0.636454   \n",
       "17                 0.879661                7.698441     0.409864   \n",
       "18                 1.176135                8.029264     0.314214   \n",
       "19                 0.885543                8.210233     0.474525   \n",
       "20                 0.270529                7.633711     0.778365   \n",
       "21                 1.083285                6.495710     0.404672   \n",
       "22                 0.654049                6.969939     0.646319   \n",
       "23                 0.441067                6.083616     0.724400   \n",
       "24                 1.122941                3.505183     0.375294   \n",
       "25                 0.697549                7.286471     0.588725   \n",
       "26                 0.951235                5.671804     0.362716   \n",
       "27                 0.849106                6.306970     0.385366   \n",
       "28                 0.721748                6.307881     0.491748   \n",
       "29                 0.949704                7.430811     0.479593   \n",
       "..                      ...                     ...          ...   \n",
       "195                2.090262                9.285171     0.100031   \n",
       "196                1.505041                6.092183     0.422231   \n",
       "197                2.518696                5.246733     0.130821   \n",
       "198                2.890822                3.562763     0.027534   \n",
       "199                2.280925                7.275477     0.210058   \n",
       "200                1.959484                7.639567     0.234217   \n",
       "201                1.360223                7.424756     0.108953   \n",
       "202                2.013923                6.710634     0.036147   \n",
       "203                1.301373                7.286630     0.143235   \n",
       "204                0.888379               10.342842     0.533790   \n",
       "205                2.109250                9.000184     0.088162   \n",
       "206                1.847811                6.001343     0.230533   \n",
       "207                2.964043                5.257020     0.098303   \n",
       "208                0.861590                9.833957     0.545566   \n",
       "209                1.611011                7.249514     0.042169   \n",
       "210                1.676016                8.471461     0.020761   \n",
       "211                1.673405                9.005006     0.085238   \n",
       "212                2.838242                6.446547     0.029078   \n",
       "213                1.626563                9.167279     0.175471   \n",
       "214                1.505392                8.994395     0.036615   \n",
       "215                1.447921                8.297596     0.083002   \n",
       "216                1.619685                8.971560     0.240291   \n",
       "217                2.294710                8.539161     0.027746   \n",
       "218                1.668948                9.227372     0.057657   \n",
       "219                2.102887                7.554920     0.149278   \n",
       "220                1.700295                6.978095     0.031609   \n",
       "221                1.050904               10.134197     0.382277   \n",
       "222                1.299268               10.346409     0.148433   \n",
       "223                1.493227                9.565596     0.153262   \n",
       "224                1.042471               10.531901     0.377851   \n",
       "\n",
       "     1_norm_Shannon_entropy  1_norm_mean  2_norm_Shannon_entropy  2_norm_mean  \\\n",
       "0                  8.404122     0.422907                7.411710     0.184978   \n",
       "1                  6.611893     0.240549                5.904274     0.227134   \n",
       "2                  7.891524     0.272066                7.206297     0.125626   \n",
       "3                  8.375098     0.618171                6.603034     0.063341   \n",
       "4                  9.026727     0.164896                9.456843     0.410174   \n",
       "5                  8.332433     0.540220                6.788715     0.105477   \n",
       "6                  5.966079     0.235854                5.836206     0.235041   \n",
       "7                  8.507873     0.345432                7.540641     0.143561   \n",
       "8                  8.072578     0.299263                7.271297     0.156544   \n",
       "9                  9.501269     0.386080                8.253264     0.132259   \n",
       "10                 8.353835     0.495685                5.462923     0.042067   \n",
       "11                 9.067475     0.437464                7.256276     0.096567   \n",
       "12                 8.546743     0.300066                7.731857     0.112239   \n",
       "13                 7.975706     0.359356                7.111262     0.168713   \n",
       "14                 8.520286     0.367287                6.362006     0.112040   \n",
       "15                 8.576402     0.405039                5.712645     0.038642   \n",
       "16                 7.571775     0.211241                6.433968     0.141348   \n",
       "17                 7.802700     0.351017                6.992631     0.206102   \n",
       "18                 8.332981     0.391004                7.289096     0.159061   \n",
       "19                 8.136216     0.284502                7.151931     0.171629   \n",
       "20                 7.130560     0.186010                4.871919     0.026298   \n",
       "21                 6.601895     0.230219                6.170636     0.285474   \n",
       "22                 6.259954     0.111043                5.637852     0.196442   \n",
       "23                 5.653626     0.190400                3.064283     0.029067   \n",
       "24                 3.678104     0.328824                2.747662     0.161765   \n",
       "25                 6.891501     0.204804                5.713699     0.144804   \n",
       "26                 5.906129     0.436667                4.188393     0.119012   \n",
       "27                 6.494167     0.437967                4.872470     0.128618   \n",
       "28                 6.266904     0.375437                4.485850     0.072039   \n",
       "29                 7.230749     0.243741                6.093404     0.128889   \n",
       "..                      ...          ...                     ...          ...   \n",
       "195               10.267086     0.339982                9.821386     0.188091   \n",
       "196                5.746054     0.116364                5.362013     0.141405   \n",
       "197                5.235094     0.071401                5.571376     0.064348   \n",
       "198                4.660997     0.091027                5.500514     0.078425   \n",
       "199                7.145018     0.096821                6.848450     0.064335   \n",
       "200                7.755186     0.212910                6.376130     0.040258   \n",
       "201                8.369104     0.550089                7.527496     0.240267   \n",
       "202                8.240911     0.322067                8.485223     0.343765   \n",
       "203                7.839782     0.517157                7.100898     0.263203   \n",
       "204               10.073810     0.245542                8.311780     0.078898   \n",
       "205                9.871528     0.306083                8.888917     0.083196   \n",
       "206                6.232854     0.230888                5.215456     0.084024   \n",
       "207                5.463536     0.078773                4.758411     0.020794   \n",
       "208                9.413423     0.179205                8.748819     0.170573   \n",
       "209                8.577903     0.541085                7.922692     0.208474   \n",
       "210               10.274089     0.500485                9.907630     0.306834   \n",
       "211               10.067114     0.437861                9.741052     0.273496   \n",
       "212                7.749261     0.163162                7.424019     0.056495   \n",
       "213                9.913418     0.358950                9.476872     0.233437   \n",
       "214               10.377234     0.551991                9.934749     0.309342   \n",
       "215                9.590411     0.547491                8.755449     0.239731   \n",
       "216                9.279237     0.196099                9.453570     0.371704   \n",
       "217                9.492779     0.325242                9.452364     0.184726   \n",
       "218               10.490767     0.461275               10.175851     0.296798   \n",
       "219                8.146013     0.148714                8.595156     0.286811   \n",
       "220                7.865264     0.456783                7.717421     0.345496   \n",
       "221               10.266688     0.372040                8.806283     0.129169   \n",
       "222               11.107453     0.551112               10.010084     0.212190   \n",
       "223               10.205015     0.457333                9.358499     0.217881   \n",
       "224               10.622069     0.363683                9.277570     0.154379   \n",
       "\n",
       "     3_norm_Shannon_entropy  3_norm_mean  4_norm_Shannon_entropy  4_norm_mean  \\\n",
       "0                  6.709523     0.042863                6.362523     0.061784   \n",
       "1                  4.216028     0.013293                3.881828     0.031098   \n",
       "2                  7.357105     0.108176                7.731327     0.303692   \n",
       "3                  6.199967     0.063317                6.349997     0.085146   \n",
       "4                  8.548836     0.135313                7.689018     0.057717   \n",
       "5                  6.164989     0.055648                5.665417     0.030660   \n",
       "6                  5.024997     0.056504                5.278880     0.155691   \n",
       "7                  6.960323     0.081259                6.926848     0.097230   \n",
       "8                  6.868453     0.095207                6.571061     0.103018   \n",
       "9                  7.512133     0.063542                6.801179     0.028239   \n",
       "10                 5.288560     0.022429                4.301908     0.004832   \n",
       "11                 7.006479     0.042650                6.221441     0.033590   \n",
       "12                 7.556725     0.109171                7.767403     0.202902   \n",
       "13                 6.966382     0.091510                6.778723     0.143564   \n",
       "14                 5.753966     0.036832                4.438861     0.008000   \n",
       "15                 5.106374     0.013209                4.306042     0.018091   \n",
       "16                 3.182311     0.003333                4.376511     0.005780   \n",
       "17                 5.112219     0.012814                5.411873     0.018441   \n",
       "18                 6.709758     0.075000                6.126293     0.059716   \n",
       "19                 6.131654     0.018959                5.969771     0.049367   \n",
       "20                 2.841580     0.002404                2.753089     0.005144   \n",
       "21                 5.163017     0.033796                4.927565     0.044088   \n",
       "22                 3.954726     0.032699                3.672541     0.012515   \n",
       "23                 2.673208     0.029200                2.080029     0.025200   \n",
       "24                 2.075510     0.058824                2.083333     0.071765   \n",
       "25                 4.771375     0.040980                3.690693     0.019118   \n",
       "26                 4.147420     0.047901                2.143010     0.031975   \n",
       "27                 4.127193     0.035610                1.579861     0.010894   \n",
       "28                 4.232194     0.036602                3.432663     0.021650   \n",
       "29                 6.318554     0.141000                5.057025     0.005407   \n",
       "..                      ...          ...                     ...          ...   \n",
       "195                9.437234     0.111163                9.831850     0.259734   \n",
       "196                5.086922     0.172231                4.783493     0.146777   \n",
       "197                7.271385     0.611691                5.874781     0.120193   \n",
       "198                6.802402     0.565890                6.045024     0.235548   \n",
       "199                8.248507     0.458092                7.380515     0.169884   \n",
       "200                8.074537     0.383260                6.751585     0.128490   \n",
       "201                6.297025     0.071693                5.083543     0.027550   \n",
       "202                7.946560     0.183695                7.321047     0.112697   \n",
       "203                5.888236     0.047157                4.935821     0.028007   \n",
       "204                8.551946     0.079971                7.947315     0.060394   \n",
       "205               10.222826     0.450495                8.313079     0.070730   \n",
       "206                6.364617     0.366627                4.886858     0.086923   \n",
       "207                7.137511     0.362310                7.265979     0.438809   \n",
       "208                8.251703     0.075352                6.718198     0.027951   \n",
       "209                7.500894     0.177004                5.759305     0.029798   \n",
       "210                9.026972     0.122150                8.144387     0.048008   \n",
       "211                8.801484     0.123472                8.179657     0.078631   \n",
       "212                9.100138     0.439153                8.679939     0.310782   \n",
       "213                9.077094     0.124955                8.649745     0.105771   \n",
       "214                8.620702     0.071465                7.759743     0.029040   \n",
       "215                7.753385     0.095000                6.666367     0.033109   \n",
       "216                8.323006     0.085372                8.080890     0.105331   \n",
       "217                9.445888     0.247488                9.047114     0.214130   \n",
       "218                9.197765     0.120210                8.524394     0.062258   \n",
       "219                8.476915     0.276535                7.486051     0.137178   \n",
       "220                6.604886     0.107426                6.022977     0.056595   \n",
       "221                7.902775     0.042825                7.860285     0.072181   \n",
       "222                8.599022     0.027276                8.301907     0.059535   \n",
       "223                8.527567     0.084145                8.036458     0.085906   \n",
       "224                8.206935     0.044607                7.955018     0.058277   \n",
       "\n",
       "     log_average_weighted_class   id  fibrosis_score  \n",
       "0                      0.157888    2               2  \n",
       "1                     -0.145590    5               0  \n",
       "2                      0.724955    6               3  \n",
       "3                      0.245411    8               3  \n",
       "4                      0.485545   10               1  \n",
       "5                      0.043631   14               2  \n",
       "6                      0.406386   15               3  \n",
       "7                      0.237897   23               2  \n",
       "8                      0.271820   31               3  \n",
       "9                     -0.042658   32               2  \n",
       "10                    -0.400902   33               1  \n",
       "11                    -0.109595   34               2  \n",
       "12                     0.510726   41               3  \n",
       "13                     0.437792   43               3  \n",
       "14                    -0.304590   44               2  \n",
       "15                    -0.515462   50               2  \n",
       "16                    -0.631205   52               0  \n",
       "17                    -0.128219   55               1  \n",
       "18                     0.162234   58               2  \n",
       "19                    -0.121554   60               2  \n",
       "20                    -1.307377   62               0  \n",
       "21                     0.079998  135               2  \n",
       "22                    -0.424573   63               0  \n",
       "23                    -0.818559   64               0  \n",
       "24                     0.115951   65               0  \n",
       "25                    -0.360182   66               0  \n",
       "26                    -0.049995   67               0  \n",
       "27                    -0.163572   68               0  \n",
       "28                    -0.326080   69               0  \n",
       "29                    -0.051605   70               0  \n",
       "..                          ...  ...             ...  \n",
       "195                    0.737289  433               3  \n",
       "196                    0.408820  434               4  \n",
       "197                    0.923741  435               4  \n",
       "198                    1.061541  436               4  \n",
       "199                    0.824581  437               4  \n",
       "200                    0.672681  438               4  \n",
       "201                    0.307648  440               2  \n",
       "202                    0.700085  441               2  \n",
       "203                    0.263420  442               2  \n",
       "204                   -0.118357  443               2  \n",
       "205                    0.746332  444               4  \n",
       "206                    0.614002  445               3  \n",
       "207                    1.086554  446               4  \n",
       "208                   -0.148976  448               0  \n",
       "209                    0.476862  449               2  \n",
       "210                    0.516420  450               2  \n",
       "211                    0.514860  451               2  \n",
       "212                    1.043185  452               4  \n",
       "213                    0.486469  453               2  \n",
       "214                    0.409053  454               2  \n",
       "215                    0.370129  455               2  \n",
       "216                    0.482232  456               2  \n",
       "217                    0.830607  457               2  \n",
       "218                    0.512194  458               2  \n",
       "219                    0.743311  459               4  \n",
       "220                    0.530802  460               3  \n",
       "221                    0.049651  463               0  \n",
       "222                    0.261801  464               2  \n",
       "223                    0.400940  465               2  \n",
       "224                    0.041594  466               0  \n",
       "\n",
       "[225 rows x 14 columns]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set final X and Y\n",
    "\n",
    "# X without slide id\n",
    "X = X.drop(columns = [\"id\"], axis = 1)\n",
    "\n",
    "# Y with target score\n",
    "Y = X_Y[target_score_column]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_ANN(input_dim, max_score_):\n",
    "    \"\"\" Create the scoring ANN\n",
    "\n",
    "    input:\n",
    "        input_dim: number of dimensions for inpit\n",
    "        max_score_: restrict output to 0...max_score\n",
    "\n",
    "    output:\n",
    "        model: ANN model (Keras)\n",
    "    \"\"\"\n",
    "\n",
    "    # Activation function with predefined range of score\n",
    "    global max_score\n",
    "    max_score = max_score_\n",
    "\n",
    "    def mapping_to_target_range(x, target_min=0):\n",
    "        global max_score\n",
    "        target_max = max_score\n",
    "\n",
    "        z = K.backend.tanh(x) + 1\n",
    "        scale = (target_max - target_min) / 2.\n",
    "        return z * scale + target_min\n",
    "\n",
    "    # define MLP network\n",
    "    model = Sequential()\n",
    "    model.add(Dense(100, input_dim = input_dim, activation=\"elu\"))\n",
    "    model.add(Dropout(0.8))\n",
    "    model.add(Dense(100, activation=\"relu\"))\n",
    "    model.add(Dense(1, activation = mapping_to_target_range))\n",
    "\n",
    "    # model\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0: 5.921052631578948, 1: 5.487804878048781, 2: 3.0405405405405403, 3: 4.591836734693878, 4: 9.782608695652174}\n"
     ]
    }
   ],
   "source": [
    "sample_weights_dict = {}\n",
    "\n",
    "for y_unique in np.unique(Y):            \n",
    "    sample_weights_dict[y_unique] = len(Y) / np.sum(Y == y_unique)\n",
    "    \n",
    "print(sample_weights_dict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "scaler = StandardScaler()\n",
    "\n",
    "X_scaled = pd.DataFrame(scaler.fit_transform(X))\n",
    "\n",
    "# Save scaler for later use\n",
    "scaler_filename = os.path.join(model_path, model_name + \".scaler\")\n",
    "joblib.dump(scaler, scaler_filename) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fit_scoring_ANN(X_train, Y_train, X_val, Y_val, model_name, n_epochs, score_max):\n",
    "    # Clean up memory\n",
    "    K.backend.clear_session()\n",
    "    \n",
    "    mlp = create_ANN(X_train.shape[1], score_max)\n",
    "        \n",
    "    optimizer = optimizers.Nadam(lr = 0.0003)        \n",
    "    \n",
    "    if len(X_val) > 0:\n",
    "        callbacks = [ReduceLROnPlateau(monitor='val_loss', factor=0.9, patience=2, cooldown=75, verbose=0, min_lr=1e-8),\n",
    "                     ModelCheckpoint(model_name, monitor='val_loss', verbose=0, save_best_only=True)] \n",
    "        \n",
    "        val_data = (X_val, Y_val)\n",
    "    else:\n",
    "        callbacks = [ReduceLROnPlateau(monitor='loss', factor=0.9, patience=2, cooldown=75, verbose=0, min_lr=1e-8)]         \n",
    "        \n",
    "        val_data = None\n",
    "    \n",
    "    mlp.compile(loss = \"mean_squared_error\", metrics = [\"accuracy\", \"mean_squared_error\"], optimizer = optimizer)\n",
    "    history = mlp.fit(X_train, Y_train, validation_data= val_data, epochs = n_epochs, verbose = 0, class_weight = sample_weights_dict, callbacks = callbacks)\n",
    "    \n",
    "    return mlp, history"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_history(history, save_file_name = \"\"):\n",
    "    fig, ax1 = plt.subplots()    \n",
    "    \n",
    "    if (\"val_loss\" in history.history):\n",
    "        ax2 = ax1.twinx()\n",
    "        l2 = ax2.plot(history.history['val_loss'], color = \"blue\", alpha = 0.8, label = \"Validation\")\n",
    "        ax2.set_ylabel('Val MSE loss')\n",
    "        \n",
    "    l1 = ax1.plot(history.history['loss'], color = \"red\", alpha = 0.8, label = \"Training\") \n",
    "    ax1.set_ylabel('Training MSE loss')\n",
    "        \n",
    "    plt.title('%s - ANN' % (target_score))\n",
    "    plt.xlabel('epoch')\n",
    "    \n",
    "    # https://stackoverflow.com/questions/5484922/secondary-axis-with-twinx-how-to-add-to-legend\n",
    "    fig.legend(loc = [0.55, 0.7])        \n",
    "    \n",
    "    if save_file_name:    \n",
    "        plt.savefig(save_file_name, dpi = 300, bbox_inches='tight')\n",
    "        \n",
    "    plt.show()    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_cont_score_vs_path_score(AI_score, ground_truth, target_score, save_plot = True):\n",
    "    if target_score == \"fibrosis_score_pathologist\":\n",
    "        key = \"Continuous_AI_fibrosis_score\"\n",
    "        score_str = \"Fibrosis_score\"\n",
    "        title_str = \"Fibrosis\"    \n",
    "    elif target_score == \"ballooning_score_pathologist\":\n",
    "        key = \"Continuous_AI_ballooning_score\"\n",
    "        score_str = \"Ballooning_score\"\n",
    "        title_str = \"Ballooning\"\n",
    "    elif target_score == \"inflammation_score_pathologist\":\n",
    "        key = \"Continuous_AI_inflammation_score\"\n",
    "        score_str = \"Inflammation_score\"\n",
    "        title_str = \"Inflammation\"\n",
    "    elif target_score == \"steatosis_score_cv_based\":\n",
    "        key = \"Continuous_AI_steatosis_score\"\n",
    "        score_str = \"Steatosis_score\"\n",
    "        title_str = \"Steatosis\"    \n",
    "\n",
    "    col_str = {\"Inflammation_score\": \"#ff1f20\", \"Ballooning_score\" : \"#32cd32\", \"Steatosis_score\" : \"b\", \"Fibrosis_score\" : \"#111111\"}\n",
    "\n",
    "    score_label_str = score_str.replace(\"_\", \" \").capitalize()\n",
    "    score_label_str = score_label_str.replace(\"pathologist\", \"\")\n",
    "\n",
    "    if (score_str != \"Fibrosis_score\"):\n",
    "        score_label_str = score_label_str.replace(\"score\", \"sub-score / sample\")\n",
    "    else:\n",
    "        score_label_str = score_label_str.replace(\"score\", \"score / sample\")\n",
    "\n",
    "    if target_score != \"steatosis_score_cv_based\":\n",
    "        score_label_str = score_label_str + \"\\n(Human expert)\"\n",
    "\n",
    "    matplotlib.rcParams.update({'font.size': 14})\n",
    "    fig = plt.figure()\n",
    "    ax = plt.subplot(111) \n",
    "    jitter = 0.06\n",
    "\n",
    "    ax.scatter(AI_score, ground_truth + np.random.normal(0, jitter, ground_truth.shape[0]), c = col_str[score_str], alpha=0.4, s=20)\n",
    "\n",
    "    plt.ylabel(score_label_str)        \n",
    "    plt.xlabel(key)   \n",
    "    plt.title(title_str)\n",
    "\n",
    "    if save_plot:\n",
    "        fig_file_name_str = os.path.join(model_path, full_results_file_name[:-4]) + \".png\"\n",
    "        fig.savefig(fig_file_name_str, bbox_inches='tight', dpi=600)\n",
    "        print(\"Figure saved\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "def save_confusion_matrix(y_true, y_pred, save_file_name, label_list = []):\n",
    "    cm = confusion_matrix(y_true, y_pred)\n",
    "\n",
    "    matplotlib.rcParams.update({'font.size': 20})\n",
    "\n",
    "    cm_norm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "    plt.imshow(cm_norm, interpolation='nearest', cmap=plt.cm.Blues)\n",
    "    \n",
    "    if len(label_list):\n",
    "        tick_marks = np.arange(len(label_list))\n",
    "        plt.xticks(tick_marks, label_list, rotation=0)\n",
    "        plt.yticks(tick_marks, label_list)\n",
    "    else:\n",
    "        tick_marks = np.arange(len(np.unique(y_true)))\n",
    "        plt.xticks(tick_marks, sorted(np.unique(y_true)), rotation=0)\n",
    "        plt.yticks(tick_marks, sorted(np.unique(y_true)))            \n",
    "\n",
    "    plt.gca().invert_yaxis()\n",
    "    fmt = 'd'\n",
    "    thresh = cm_norm.max() / 2.\n",
    "\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt),\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm_norm[i, j] > thresh else \"black\")\n",
    "    \n",
    "    plt.ylabel('Pathologist score')\n",
    "    plt.xlabel('Rounded AI-score')\n",
    "    plt.gcf().subplots_adjust(bottom=0.15)\n",
    "    plt.tight_layout()\n",
    "\n",
    "    plt.savefig(save_file_name, dpi=600)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac695fda90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac623627f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaMAAAEICAYAAADyTpvZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3Xd8VFX6+PHPM5MGoYSA9CqiNBEQUawUC7gqoq6KXVFWv7a1rKKLa/mtu6y9i6iIuIplFXEV2yqKrCsICIiA9F4CkR4gmczz++PcSSZhMhlCJpPyvF+vYWZuPTMT7nPPuc89R1QVY4wxJpF8iS6AMcYYY8HIGGNMwlkwMsYYk3AWjIwxxiScBSNjjDEJZ8HIGGNMwlkwquJE5AgRmSMiO0XkFhEZLSL3efP6isjaCirHLhE5tCL2VZWJyEoRObWM66qIHFYOZfhURK482O0YU56SEl0Ac9DuAqaoavdEFkJV6yRy/5WRiIwD1qrqyESXJZyqDoplORFRoIOqLj2Y/YlIHWAj8F3xfYvISqA20E5Vd3vTrgUuU9W+YeWYDxylqkFv2l+Blqp61cGUzVQeVjOq+toAv5THhkTEXx7bqWxExE66Eut8YB9wmog0jTDfD9xayjaaAxeXd8FM5WHBqAoTka+BfsBzXjPZ4SIyzjtrDF/uXhHZ4jURXRo2fZyIvCgik0VkN9BPROqLyHgR2Swiq0RkpIj4vOUPE5FvRWS7t713wrZV0IQkImeKyAKv6XCdiNxZyudoJCIfi8g2EflNRL4L22crEfnAK0+2iDznTfd5ZVslIllemet789p65RkmIquBr73px4nI995+5opI3xi+429E5O8iMkNEdojIJBHJDJv/nohs9L6TqSLSxZs+HLgUuMv7bf4dttnuIjLPW+cdEUkL2951IrLU+x4+EpHmJZQr2u/kF5HHvd9ohYjc5H0fSWGf6dpov6mITPV2Ndcr/0WlfVdRXAmMBuYBl0WY/yhwp4hkRNnGI8CDdmJRjamqParwA/gGuDbs/Tjgr97rvkAAeAJIBU4BdgNHhC27HTgBd2KSBowHJgF1gbbAYmCYt/wE4M9hy54Ytl8FDvNebwBO8l43AHqW8hn+jjtYJXuPkwDBnTHPBZ4E0sP3CVwDLAUOBeoAHwBvePPaeuUZ761XC2gBZANneuU/zXt/SAzf7zqgq7et94F/hs2/xvuuUoGngDmRfouwaSuBGbgz/UxgIXC9N68/sAXo6W3vWWBqCd9xtN/pemAB0NL7/v/jrZtU/G8m1t/0IP4+2wBBoDNwBzAvwvdxqvf7hf5urwW+KVaODsCssHL/FRiX6P9/9ii/h9WMaob7VHWfqn4LfAJcGDZvkqr+V11bfB6uKeQeVd2pqiuBx4HLvWXzcAeX5qq6V1WnlbC/PKCziNRT1a2qOruU8uUBzYA2qpqnqt+pqgK9cQftP6nq7mL7vBR4QlWXq+ou4B7g4mJnzg946+3BnZFPVtXJqhpU1S+BmbjgVJo3VHW+umsa9wEXhpo0VXWs913tAx4AjgrV0KJ4RlXXq+pvwL+B0PW+S4Gxqjrb2949QB8RaRu+srfvaL/ThcDTqrpWVbcCo6KUJdbftKwuxwWgBcDbQBcR6RFhub8AN4vIISVsR3Hf/X0iklLOZTSVgAWj6m+rdxANWYU7wIesCXvdCFczWVVs+Rbe67twNZYZIvKLiFxTwj7Pxx3kV3lNQH1KKeOjuFrOFyKyXERGeNNbAatUNRBhneYRypkENCnhs7UBfu810W0TkW3AibggWJrw7azCfUeNvOawUSKyTER24M7ywX2P0WwMe52Dq9nt95m8IJtN4fcfUtrv1LxYmcNfFxfrb7ofb/ld3uOkEha7AngTQFXXAd/imu2KUNX5wMfAiOLzwpaZDKwF/hBrGU3VYcGo+msgIulh71sD68Peh3fbvoXCM+Xw5dcBqOpGVb1OVZvjDggvSIRUY1X9UVUHA42BD4F3oxXQO7u/Q1UPBc4BbheRAbiDaOsSrhOsj1DOALCphM+2BlfDyQh7pKtqtFpDSKti+8nDfVeXAINxzUz1cc1l4A7uxfcfiyKfyfvdGuJ9/2Gi/k64ZtKWJZS/iFh/0xLW7aKqdbzHd8Xni8jxuOa1e7zrahuBY4FLSvhN7weuY//gG+7PwL24DDxTjVgwqhkeFJEU7+z1LOC9SAupaj4ucDwsInVFpA1wO/BPABH5vYiEDnJbcQfbYPg2vP1cKiL1VTUP2FF8meJE5CzvQrrgrmHle+vMwB1YR4lIuoikicgJ3moTgNtEpJ241OG/Ae+UUIvC+wxni8gZXo0mTdx9WC1LWD7cZSLSWURqAw8B//K+q7q4LLFs3MHxb8XW24S7phWrCcDVItJdRFK97U33muEKlPY7efNuFZEWXlLA3SXtsJTf9EDLX9yVwJe460XdvUdX3DW8/dLL1aWQvwPcUtIGVfUbXJq33SdVzVgwqv424g4y63HNJder6qIoy9+MS3JYDkwD3gLGevOOAaaLyC7gI+BWVV0eYRuXAyu9pqvrcddCoumAu8i+C/gf8IKqTvEOumcDhwGrcU00oayuscAbwFRgBbDXK3tEqroGV4u5F9iMqyn9idj+D7yBS0bYiLvIHzpYjsc1j63DJQz8UGy9V3HXzraJyIel7URV/4O7LvI+Lgi3p+R05mi/08vAF7jstZ+AybhaY36E7UT7TR8AXvfKf2GEdUvkZQheCDzr1b5CjxW477OkYPIQLlEkmpG45A9TjYi7TmyMiUREvsFlz72S6LKUlYgMAkaraptSFzYmQaxmZEw1IyK1xN3rlSQiLXDXYiYmulzGRGPByFQIcTfe7orw+LQSlC1SuaJliFV2AjyIa579CXcv018SWiJjSmHNdMYYYxLOakbGGGMSrkr08+Tz+bRWrVqJLoYxxlQpOTk5qqpVotJRJYJRrVq12L17d+kLGmOMKSAie0qZ3wp3i0IT3D1mY1T16WLLXIq7V02AncANqjq3vMtaJYKRMcaYuAgAd6jqbBGpC8wSkS+9vgRDVgCnqOpW7zaBMbieNMqVBSNjjKmhVHUD7gZrVHWniCzEdce0IGyZ78NW+YGiXU2VGwtGxhhTfSWJyMyw92NUdUykBb3e4XsA06NsbxgQl9sxLBgZY0z1FVDVXqUt5PXv+D7wR1XdUcIy/XDB6MTyLaJjwcgYY2owEUnGBaI3VfWDEpbpBrwCDFLV7HiUo0qk/BljjCl/Xk/5rwILVfWJEpZpjRuJ93JVXRy3slSFHhjS09PVUruNMebAiEiOqpbYC7qInAh8B/xM4dAh9+LGx0JVR4vIK7gBM0ODOcbU9HfAZa3Owei7KQGWTZzHVU91B59VAo0xNUtpwagyqdZH6O+fmsEb4wLw7beJLooxxpgoqnUw8m1cT1B9kJaW6KIYY4yJIm7BSERaicgUEVkgIr+IyK3e9AdEZJ2IzPEeZ8arDP5kH/n4wO+P1y6MMcaUg3imdkfsZsKb96SqPhbHfQMgfY5D526CQCDeuzLGGHMQ4haMonQzUWH8yT7y1WfByBhjKrkKuWYUoZuJm0RknoiMFZEGJawzXERmisjMQBmDiS/JRxALRsYYU9nFPRhF6GbiRaA90B1Xc3o80nqqOkZVe6lqr6SkslXgfEk+l8CQn1+2whtjjKkQcQ1GkbqZUNVNqpqvqkHgZaB3vPbvT/YRRNA8qxkZY0xlFs9suojdTIhIs7DFhgDz41UGX5L7eBaMjDGmcotnNt0JwOXAzyIyx5t2LzBURLrjRhVcCfwhXgXwJ7tglJ8XrN43VBljTBUXz2y6abhhaoubHK99Fid+qxkZY0xVUK0rDOE1I2OMMZVXtQ5GPi8YBfMsm84YYyqz6h2M/FYzMsaYqqBaj/TqT3F90lnNyFRFeXl5rF27lr179ya6KNVGWloaLVu2JDk5OdFFMcVU62AUSu0OBqxmZKqetWvXUrduXdq2bYu7U8IcDFUlOzubtWvX0q5du0QXxxRTrZvp/EkCCPm5VjMyVc/evXtp2LChBaJyIiI0bNjQapphShpdodgyIiLPiMhSrxu3nvEoS/WuGfkAAQ1YMDJVkwWi8mXf534ijq6gqgvClhkEdPAex+K6dDu2vAtSrWtGLhiJJTAYUwbZ2dl0796d7t2707RpU1q0aFHwPjc3N6ZtXH311fz6669Rl3n++ed58803y6PI5gCp6gZVne293glEGl1hMDBenR+AjGI96ZSLGlAzEktgMKYMGjZsyJw5rvOUBx54gDp16nDnnXcWWUZVUVV8vsjnta+99lqp+7nxxhsPvrCmJEkiMjPs/RhVHRNpwQijK4S0ANaEvV/rTdtQfsWsCTUjrGZkTHlaunQpnTt35tJLL6VLly5s2LCB4cOH06tXL7p06cJDDz1UsOyJJ57InDlzCAQCZGRkMGLECI466ij69OlDVlYWACNHjuSpp54qWH7EiBH07t2bI444gu+//x6A3bt3c/7559O5c2cuuOACevXqVRAoTVSB0OgH3qOkQFR8dIUKV61rRn4/VjMy1cPjj0MpzV0H7Igj4I47yrTqokWLGD9+PL169QJg1KhRZGZmEggE6NevHxdccAGdO3cuss727ds55ZRTGDVqFLfffjtjx45lxIgR+21bVZkxYwYfffQRDz30EJ999hnPPvssTZs25f3332fu3Ln07BmXa+g1UqTRFYpZB7QKe9/Sm1auqn/NSCCYr4kuijHVSvv27QsCEcCECRPo2bMnPXv2ZOHChSxYsGC/dWrVqsWgQYMAOProo1m5cmXEbZ933nn7LTNt2jQuvvhiAI466ii6dOlSjp+m5ippdIViPgKu8LLqjgO2eyN5l6saUTOyZjpT5ZWxBhMv6enpBa+XLFnC008/zYwZM8jIyOCyyy6LmD6dkpJS8Nrv91PSCM6pqamlLmPKTUmjK7QGUNXRuM6tzwSWAjnA1fEoSLUORqEEBkvtNiZ+duzYQd26dalXrx4bNmzg888/Z+DAgeW6jxNOOIF3332Xk046iZ9//jlizcscuCijK4Qvo0Dcs0yqfzBCyA9YM50x8dKzZ086d+5Mx44dadOmDSeccEK57+Pmm2/miiuuoHPnzgWP+vXrl/t+TOKIC3qVW3p6uu7evfuA15s2Df44eDmvn/kuXd7Y/0KpMZXZwoUL6dSpU6KLUSkEAgECgQBpaWksWbKE008/nSVLlpCUdODn0zXpexWRHFVNL33JxKv+NSO7ZmRMlbdr1y4GDBhAIBBAVXnppZfKFIhM5VWtf82C1G7LpjOmSsvIyGDWrFmJLoaJo+qf2o1Yr93GGFPJVetg5GpGWAKDMcZUctU6GFlqtzHGVA3VPxgh5FssMsaYSq36ByOxa0bGlEW/fv34/PPPi0x76qmnuOGGG0pcp06dOgCsX7+eCy64IOIyffv2ZebMmRHnhe8nJyen4P2ZZ57Jtm3bYi26qYJqQDCya0bGlMXQoUN5++23i0x7++23GTp0aKnrNm/enH/9619l3nfxYDR58mQyMjLKvD1T+VXrYFSQ2m01I2MO2AUXXMAnn3xSMJDeypUrWb9+PT169GDAgAH07NmTI488kkmTJu237sqVK+natSsAe/bs4eKLL6ZTp04MGTKEPXv2FCx3ww03FAw9cf/99wPwzDPPsH79evr160e/fv0AaNu2LVu2bAHgiSeeoGvXrnTt2rVg6ImVK1fSqVMnrrvuOrp06cLpp59eZD+m8qvW9xkVNNPZfUamikvECBKZmZn07t2bTz/9lMGDB/P2229z4YUXUqtWLSZOnEi9evXYsmULxx13HOecc06JQ3q/+OKL1K5dm4ULFzJv3rwiwz88/PDDZGZmkp+fz4ABA5g3bx633HILTzzxBFOmTKFRo0ZFtjVr1ixee+01pk+fjqpy7LHHcsopp9CgQQOWLFnChAkTePnll7nwwgt5//33ueyyy8rluzLxV/1rRtY3nTFlFt5UF2qiU1XuvfdeunXrxqmnnsq6devYtGlTiduYOnVqQVDo1q0b3bp1K5j37rvv0rNnT3r06MEvv/xSageo06ZNY8iQIaSnp1OnTh3OO+88vvvuOwDatWtH9+7dgehDVJjKqUbUjDTfmulM1ZaoESQGDx7MbbfdxuzZs8nJyeHoo49m3LhxbN68mVmzZpGcnEzbtm0jDhlRmhUrVvDYY4/x448/0qBBA6666qoybSckNPQEuOEnrJmuaqnWNaOCBAZL7TamTOrUqUO/fv245pprChIXtm/fTuPGjUlOTmbKlCmsWrUq6jZOPvlk3nrrLQDmz5/PvHnzADf0RHp6OvXr12fTpk18+umnBevUrVuXnTt37retk046iQ8//JCcnBx2797NxIkTOemkk8rr45oEqhE1I0tgMKbshg4dypAhQwqa6y699FLOPvtsjjzySHr16kXHjh2jrn/DDTdw9dVX06lTJzp16sTRRx8NuBFbe/ToQceOHWnVqlWRoSeGDx/OwIEDad68OVOmTCmY3rNnT6666ip69+4NwLXXXkuPHj2sSa4aqNZDSKxfD+ccl8Vfaj3KOb8+GoeSGRM/NWmog4pUk77X0oaQEJGxwFlAlqp2jTC/PvBP3MivScBjqvpaPMparZvpQn3TBfODUAWCrjHGVLBxQLRheW8EFqjqUUBf4HERSYmyfJnFLRiJSCsRmSIiC0TkFxG51ZueKSJfisgS77lBvMpQ0EynPghaU50xxoRT1anAb9EWAeqKy9uv4y0biEdZ4lkzCgB3qGpn4DjgRhHpDIwAvlLVDsBX3vu4CN30mo/PshiMMTVRkojMDHsMP8D1nwM6AeuBn4FbVTUuZ/ZxS2BQ1Q3ABu/1ThFZCLQABuOqewCvA98Ad8ejDKGOUlUFAgFIiUvt0pi4UdUSbyY1B64qXCMvZwFV7XUQ658BzAH6A+2BL0XkO1XdUS6lC1Mh14xEpC3QA5gONPECFcBGoEm89luQ2o3fBSNjqpC0tDSys7Nr4gE0LlSV7Oxs0tLSEl2UquRq4AN1lgIrgOjpk2UU99RuEakDvA/8UVV3hJ/lqaqKSMT/aV51cjhAShlrNAXXjIJiwchUOS1btmTt2rVs3rw50UWpNtLS0mjZsmWii1GVrAYGAN+JSBPgCGB5PHYU12AkIsm4QPSmqn7gTd4kIs1UdYOINAOyIq2rqmOAMeBSu8uy/1AwspqRqYqSk5Np165doothqjERmYC7bNJIRNYC9wPJAKo6Gvh/wDgR+RkQ4G5V3RKPssQtGHnZF68CC1X1ibBZHwFXAqO85/27/C0nob7pgmo1I2OMKU5Vo44HoqrrgdMroizxrBmdAFwO/Cwic7xp9+KC0LsiMgxYBVwYrwIUNNPhs2BkjDGVWDyz6abhqnWRDIjXfsOFbnrNV0vtNsaYyqxa98Ag4v5RrJnOGGMqs2ofjCR006sFI2OMqbSqdTAC8Plx3QFZM50xxlRa1T4Y+f1YzcgYYyq5ah+MfD6vo1QLRsYYU2mVGoxEpJZ3zxAi0l5EzhSRKjMon8+PS+22ZjpjjKm0YqkZfQfU8npL+Bq4Dhgb11KVI7/fS+22mpExxlRasQQjn6rmAOcDL6rqEKBbfItVfnx+S+02xpjKLqZgJCLHAJcCH3vT/PErUvny+cRqRsYYU8nFEoxuBx4EPlbV+SJyKK7prkqwa0bGGFP5lZqIoKpf464VhTo/3aSq/xfvgpUXf5LVjIwxprKLJZtuvIjUE5HauGFnl4rI7fEvWvnw+a2jVGOMqexiaabr5g0xey7wJdAGuCqehSpPPp8QtAQGY4yp1GIJRsnefUWDgUmqmgsE41us8uOa6WxwPWOMqcxiCUav4IaebQB8KyKtgV1xLVU5cqndWDAyxphiRGSsiGSJyPwoy/QVkTki8ouIfBuvspQajFT1SVVtrqqnq6oCa4D+8SpQefMl+VzNKC8v0UUxxpjKZhwwsKSZIpIBvACco6pdgN/HqyCxJDDUFZFHROQHEfkBN1JrarwKVN58ST53zSg3N9FFMcaYSkVVpwK/RVnkEuADVV3tLZ8Vr7LE0kw3FsgDrvAeucBr8SpQefMnCfmSbMHIGFMTJYnIzLDH8ANc/3CggYh8IyKzROSKWFcUEZ+I1Iu5oDEs00FVw6tm94nInFh3kGg+HwR9SbBvX6KLYowxFS2gqr0OYv0k4GhgAFAL+J+I/KCqiyMtLCJvAdcD+cCPQD0ReVpVHy1tR7HUjPaKyHFhOzsO2BvDepWCzwdBf5LVjIwx5sCtBT5X1d2qugWYChwVZfnOYbcCfQq0Ay6PZUex1Iz+D3hDRFIBAXJwzXVVgt+PNdMZY0zZTAKe827vSQGOBZ6MsnyyiCTjgtFzqponIhrLjmLpDmg20EVEMr330S52VTo+H6jfb810xhhTjIhMAPoCjURkLXA/kAygqqNVdaGIfAbMw91f+oqqlpgGDrwErATmAlNFpA2wI5aylBiMROSWEqbjFfSZWHaQaD6f1YyMMSYSVR0awzKPAqVe8/GWfQYIjw2rRKRfLOtGu2Z0SCmPKsESGIwxpmKIyK1eX6YiIq+KyGxivC+1xJqRqt5XbiVMIL8f8sQSGIwxpgJco6pPi8gZuF57LgfeAL4obcVYsumqNJ/PG88oWGW60zPGmKpKvOczgTdU9ZewaVFZMDLGGFNeZonIF7hg9LmI1CXGjrVjSe2u0vx+yLdgZIwxFWEY0B1Yrqo5ItIQuDqWFUusGYnI42Gvbyo279UyFrTC+Xyg+EBjSnU3xhhTRqoaBFoCI0XkMeB4VZ0Xy7rRmunC0/GuKTavx4EVMXF8Ptyw41YzMsaYuBKRUcCtwALvcYuI/C2WdaM100kJr6sUd81IrGZkjDHxdybQ3ashISKvAz8B95a2YrRg5PMuPvnCXoeCkv/gyltx/H7ceEZWMzLGmIqQQeGwFPVjXSlaMGoIhKflLQibV2WqGQU1IwtGxhgTb38HfhKRKbjYcTIwIpYVo9302vJgSiQiY4GzgCxV7epNewC4DtjsLXavqk4+mP2UxueDoFoCgzHGxJuqThCRb4BjvEl3q+rGWNaN1jddK2C71x04InIyMBjXCd5oVS1tHO9xwHPA+GLTn1TVx2IpXHkoSO3Oz6+oXRpjTI0iIj2LTVrrPTcXkeZeh9tRRWumew+4ANghIkcBE4FHcF2IHwlEHTFQVaeKSNvSChBvLrXbEhiMMSaOHo8yT4mhf7powai2qoai22XAWFX9h4j4cN2Dl9VN3tC1M4E7VHVrpIW84XGHA6SkpJR5Z5babYwx8aWqMfXMHU20+4zC07n7A195Ow1S9gSGF4H2uDt0NxAlmqrqGFXtpaq9kpLK3lGEXTMyxpjKL9pR/ltvPPMNuMy6rwFEpClQ2vWiiFR1U+i1iLwMfFyW7RwIl9ptNSNjjKnMotWMbgEmAxuBk1Q1NAZDc6BMw0uISLOwt0OAaCMGlgtL7TbGmMhEZKyIZIlI1GOxiBwjIgERuSBeZSkxGKlqUFX/qaqPquqasOmzY0nH9oaz/R9whIisFZFhwCMi8rOIzMN1N3RbOXyGqKyZzhhjSjQOGBhtARHxA/8gyphEInJZ2OsTis27af819hcttXsrRa8NifdeAFXVzGgbLmE42wrvYNWa6YwxJrIYs55vBt6n8N6hSG4H/um9fhYIT/W+BnebT1TRrhn9F2gM/At4G1hX2sYqI5fajQUjY0xNlCQiM8Pej1HVMbGuLCItcJdU+hE9GEXryzSmvk2j9cBwlog0wN1rNBbXpPc28K6qbotl45WBpXYbY2qwgKr2Ooj1n8L1ohAUiRpTtITXkd5HFDVn2rsH6GUReQW4GFf9qkv0G5wqFbtmZIwxZdYLeNsLRI2AM0UkoKofFluuo5cLIEB77zXe+0Nj2VHUYCQivYGhuCraD8DvgW9i/BCVgt/vwrLmB6vuOBjGGJMAqtou9FpExgEfRwhEAJ0Odl/REhiWATtxTXPDKLy36EgRIdbR+xLN5wMQgsEqNO6FMcZUAC/ruS/QSETWAvcDyQCqOjrW7ajqqmLbbYjrsXu1qs6KZRvRakYbcJWK3+EGTAqvWKi3o0rP5wME8oNiwcgYY8KUkPVc0rJXlTRPRD4GRqjqfO9+0tm4Lt/ai8gYVX2qtO1HS2A4MdZCVmZ+LwJZ/oIxxsRNO1UN3Th7NfClql7hDcr6X1wiRFTRemCoFsKb6YwxxsRFeBdxA3C996CqO4GYjr5l74G0igg10wXzLZvOGGPiZI2I3Iwbx6gn8BmAiNTCuwZVmhpSM7JmOmOMiaNhQBfgKuCisHtRjwNei2UDpdaMRKRbhMnbgTXecBKVmrtmJOQHLbHbGGPiQVWzgOsjTJ8CTIllG7E0072KG3/oF1xGXSdgAVBXRIar6lcxlzgBCprpKn3YNMaYqklEPoo2X1XPKW0bsQSjlcCw0H1FInIkbgiJe3H91nWPYRsJE2qms5qRMcbETR9gDTABmE6M/dGFiyUYdQq/wVVVfxaRzqq6tJS+iiqFUDNdUMV1CVQFymyMMVVMU+A0XI89lwCfABNU9ZdYNxBLAsMiEXlWRE7wHs9401KBQFlKXZEKmulCwcgYY0y5UtV8Vf1MVa/EJS0sBb6JdSwjiK1mdAVuPIsR3vv/AvfgAtGAAytyxSvIpsM6SzXGmHjxKii/w9WO2gLPABNjXb/UYKSqObhR/v4RYfb2WHeUKAU3vSKQn1/YJYMxxphyISLjga64m10fDOuNIWaxpHYfh+s8r0348qp6+IHuLBH8flzfdOq3mpExxsTHZcBu4FbglrB8gtDI4PVK20AszXSvAXcBs4D8spUzcQqb6Sy/2xhj4kFVD7oDhVg2sENV/62q61V1U+hxsDuuKC6BQVzN6L33El0cY4wxEcQSjL4Wkb+LyDEi0i30iHvJyolrpvOuGT3zTKKLY4wxJoJYmulOLPYMVW48I58betwYY0ylFEs23UkVUZB48fkAn7jUbmOMMQVEZCxwFpClql0jzL8UuBuXiLATuEFV58ajLNGGHR+qqhNE5JZI81W1SrR5FdaMrOcFY4wpZhzwHDC+hPmw+2JUAAAgAElEQVQrgFNUdauIDALGAMcWX0hEduJazPabRTlk0zXwng8pbSOVWeiaUb4NOm6MMUWo6lQRaRtl/vdhb38AWpawXN2DLUu0Ycdf8J7vO9idJFJBarfVjIwx5mAMAz6NZUERaQykhd6r6urS1onlptdGwDW47h3Cb3odHkuhEq2g126rGRljap4kEZkZ9n6Mqo450I2ISD9cMDqxlOXOAR4HmgNZuM4SFuIG3ote0BjKMQlXPZtGFbzp1fX+o1YzMsbURAFV7XUwG/Bu5XkFGKSq2aUs/v9wHaX+R1V7eEHsslj2E0swSlfVO2LZWGXk8wHJKQTzLJvOGGMOhIi0Bj4ALlfVxTGskqeq2SLiExGfqk4Rkadi2VcswehTETldVb+IZYOVjc8HpKYSTM+Enr0TXRxjjKk0RGQC0BdoJCJrcf2QJgOo6mjgL0BD4AWvv7nSalrbRKQOMBV4U0SycH3WlSqWYHQ9cLeI5AC5FKbqZcayg0QrSGBoeAiwKqFlMcaYykRVh5Yy/1rg2gPY5GBgL3AbcClQH3golhVjCUaNDqAglU5oxIh8/NZRqjHGxIGIPA+8par/DZv8+oFso8QLKSLSwXvZpYRHaYUbKyJZIjI/bFqmiHwpIku85wbRtlEeigyuZ8HIGGPiYTHwmIisFJFHRKTHgW4g2lX90Miuz0d4PBfDtscBAyNs8ytV7QB8FbaPuCmS2m3jGRljTLlT1adVtQ9wCpANjBWRRSJyv4jENPZdtJteh3nPZeqbroQ7ewfjLpaBq8J9g+v3KG5CzXRBsWY6Y4yJJ1VdhTcyuFc7GotLgij1Rs9YrhkhIh2BzhS9o/atMpS1iapu8F5vBJqUYRsHpEgPDFYzMsaYuBGRJGAQcDEwAFfheCCWdWPpgWEkcDrQEfgcOAN3A2xZglEBVVURKTE6iMhwYDhASkpKmfdTEIzEmumMMSYeROQ0YChwJjADeBsYrqoxpXVDbIPrXQT0Azao6uXAUUD6gRcXgE0i0gzAe84qaUFVHaOqvVS1V1JSTBW4iCyBwRhj4u4e4Hugk6qeo6pvHUgggtiC0R5VzQcCIlIX17zW5sDLCsBHwJXe6ytxXQ3FVZHUbqsZGWNMuVPV/qr6iqpuLes2Yqly/CQiGbgLUTOBHbhqWFQl3Nk7CnhXRIbh7kC9sIzljpnVjIwxpvKLGozE9f/wgKpuA54Xkc+Beqo6u7QNR7mzd8CBF7PsClK71YKRMcZUVlGDkZdk8CXQ1Xu/tEJKVY6KpHYHAoktjDHGmIhiuWY0pyx301YWBc10a9bB8uUwo9QWRmOMMRUsWndAoVpTD+BHEflVRGaLyE8iUmozXWURSsQL7MhxL374IXGFMcYYE1G0ZroZQE/gnAoqS1wkJ7vnPC17ergxxpj4inaEFgBVXVZBZYmL0DWjgNqw48YYU1lFC0aHiMjtJc1U1SfiUJ5yJ+JqRwU1I7vXyBhjKp1oCQx+oA5Qt4RHlVEkGBljjAEiD/VTbL6IyDMislRE5olIz3iVJdoReoOqxjRCX2WXlGQ1I2OMiWAcbkig8SXMHwR08B7HAi96z+UuWs1I4rHDREhOtmtGxhhTnKpOBX6LsshgYLw6PwAZof5Fy1u0YFShPSXEk10zMsaYMmkBrAl7v9abVu6iDa4XLVpWKXbNyBhTQyWJyMyw92NUdUzCShNFjThCF7lmJNWm9dEYY0oTUNVeB7H+OqBV2PuW3rRyF0t3QFWeNdMZY0yZfARc4WXVHQdsDxutu1zVmJqRJTAYY0xRJQz1kwygqqOBybjRW5cCOcDV8SpLjQhGVjMyxpj9RRnqJzRfgRsroiw1opkuNRX2aUqii2GMMaYENSIYpaXB3qAFI2OMqaxqTjCympExxlRaNScYBVPdG7tmZIwxlU6NCUb71BvYyIYeN8aYSqdGBKPUVNgrtdyb3NzEFsYYY8x+akQwSkuDfW2PIKgCeXmJLo4xxphiakwwIjmFfa07WM3IGGMqoRoRjOrXd8/byLBgZIwxlVCNCEaNGrnnLfkNrJnOGGMqoRoVjLLzrWZkjDGVUY0IRg0buuctgQyrGRljTCVUI4JRZqYbxig7UB+2bEl0cYwxxhRTI4KR3w8NGkDWomxYvx6WLEl0kYwxxoSpEcEIoFMnmLTtFHbkp8PChYkujjHGmDA1JhidcQaQmcl5yx6xYGSMMZVMjRhcD6BvX6BJU7bl5pI7ez7Wh7cxxlQeNaZmVLs2/OMfgM/P0vl73bUjY4ypwURkoIj8KiJLRWREhPmtRWSKiPwkIvNE5Mx4lSUhwUhEVorIzyIyR0RmVtR+O3UCfD4W7D0UzjmnonZrjDGVjoj4geeBQUBnYKiIdC622EjgXVXtAVwMvBCv8iSyZtRPVburaq+K2mGzZtC60W4mbz8BgE2bYOZM+O034K67YNKkiiqKMcYkWm9gqaouV9Vc4G1gcLFlFKjnva4PxK1JqcZcMwJ3r9Ggozbw0tLDeHTj5UwcArmrN1Bn+3o+Omw69b7+GgYX/y2MMabKSirW+jRGVcd4r1sAa8LmrQWOLbb+A8AXInIzkA6cGq+CJqpmpLgPOEtEhkdaQESGi8hMEZkZKMcB8c4/cRMNk7bzztbTyM2F0/I/Y1ewFqM2XukWWLMG8vPLbX/GGJNAAVXtFfYYU/oqRQwFxqlqS+BM4A0RiUvcSFTN6ERVXScijYEvRWSRqk4NX8D70sYApKenl9tY4ZnNUnmz3Uje/e00Op6ZRf/Jz5MsASZvP4HBGd9y7JAhMHQo3HFHee3SGGMqo3VAq7D3Lb1p4YYBAwFU9X8ikgY0ArLKuzAJqRmp6jrvOQuYiGu7rBj169MoaTv/1/hf9J98JwD3Nn2NTP8Oblx9N70WjueLibuLrrNsGUydGmFjxhhTZf0IdBCRdiKSgktQ+KjYMquBAQAi0glIAzbHozAVHoxEJF1E6oZeA6cD8yusALVr7zcpzZfLH5tMKHg/ZvkAPvwQLrkEfvxuL1x0Edx+OzzzDKwrfuJgjDFVj6oGgJuAz4GFuKy5X0TkIREJpRvfAVwnInOBCcBVqlpuLVXhJE7bLXmHIofiakPgmgnfUtWHo62Tnp6uu3fvjrZI7JYtc8GlBP/aOsBdP+rUCRAGNviBv269qXCBww+Ht94qn7IYY0wciUiOqqYnuhyxqPBrRqq6HDiqovdboH176NcPpkyJOLt3uldJy83j2HoL+PqXpuQ0TaW2b5+bvngxBAKu91WRCiq0McZUbzWmB4YiRo0qcVbrlE28c+g9jGv7AJdueYbcXbks2tu26EKTJ8Mxx8DatfEtpzHG1BA1Mxj5/VFnt09dR9c9P9IpbYXrsWHPoUUXmDzZPQ8fDr16gSp89BF89lmcCmyMMdVbzQxGAP/3f9Hnb91Kg6SdNEvezOycjkXnzfTuIcvyshvz8uChh2DkyPIvpzHG1AA1Nxj161f0fffuERc7JnkuU3f1YG5Oh5K3tWtX0fdr1sDy5SUv+/XXB1BQY4yp/mpuMEpNLXx9/fVw7bURF7umkUu7v3/9H9gaqBt5W+GZfn/8IwwZAhde6HV6V8xf/+r6wVu1qqwlL7RuHWRnH/x2jDEmwWpuMEoJG9Ho2mvhuOPgnnv2W6xlShYPN3+BtXmNOW3J87y4+XyeybqIS5b/lcV7WzM3pwOTXtlMbtBLTJw2rXDl008vGnQmTYL//Me93rbNdTu0YAFs3Bh7uefPLxz+YvBgb9RA3DY2x+VeNGOMibsKv8+oLMr1PqOQnTsLm+pmhvUjmJsLxx+/3+JjNg9hzJYhUTf57qH3cGiquylWFdbmNSGgftr96QK4/HKXgRdSvz5s3174fupUGD0a2rWDRx+Fs86Ce+91NZ/MzMI08l5eJ+ePPQZ33llY/tD00GfJzYXnn4frroM6dWL6Sowx1UtVus+o5taMwpvpwoXXmJ55puDl8EMm8la7kYxs9irn1J9K/7r7D8P0dNbF5Kv7Sp/dfBFDlj3K75eP4t0HFsDEieRpWBZfeCACmDXL3Uz78MMukHzwgRse/Ywz4JNP9i9nKBBBYXZfuI8+gjffhFdfjfw5jTGmEqlRQ0gUkZxc8rw//MHdHFushnR42moOT1vNuRnfkhNM5dQHL6LPX04jiPDqlnN567cz+GrnMSSRz/js3xWs98imK3jkchDGcm/T1xjS4Jv993nbbftPC92Yu3w5zJhR9NpUuL/8pfD1Aw9AkyaFQai0Hs8vuMAF4Gi9SqjCihVw6KElLxMSqlnedZe7bhZu3jzXlDhgQOnbMcbUKDW3mQ5c01bfvq7JqyTLlkGtWjBnDqSnF+3Ne+ZMV3saP5589XHCr68SCKv9vN72ATqkrmb0lvMLglOyBPi8w83U88fh80Ryxhlw443QvLl7HwzC2LHuc+/eDcOGFX6WSIJBuPJKV0s7/HA3bcgQ+P3vIy+/ZQsMHOheP/009OkDPq8CXrwpEeCWW9z3+ve/R95efr5LnU9Li+njAq6meNxxrnmzNDk57ubl0GczphqpSs10NTsY7djhOk5NOoAK4qJFcNll7vXMmfDEEwW1in8c+wHvjXNp3n++ZRdDmk+HI4+EO+9kS6A+WXmZXLHyQf7cdCx7NZXl+1pwXsbXdKq1smDzucEkkiVQ/j0N1avn+uQ79liXsNGmTdHkikjBaMwYl3SxadP+80LL//STC3YPPOASNlavhvPO2387PXtGDkaRpoVs2wYjRhR+z127FgaYP/3JnSQ89FDRdTZtgt/9zm139OgSv44CN90EP/wA//tf9NpyeQkG3bOv5raQm4pTlYJRzf4fUa/egQUigI4doUGDwvd5ee75iiu4+/nW/DjiA6b1v58hf+8NN9/saiBAo6TtdPz2JQ5LXcvDG6/h8U2XMnFbX4avHsmyfi6tfHNeBn0Xv8QFyx8puPZUbnbsgJdfLkxhD782BvDNN4X3S+XkuOXHjIkciEJUXYJEbq5Ltli71nV1Xty77+6/HhQemAFmz3a1qlmzXJPl+vVwzTWFQer2213TX26uez9lSuRrZaFrcVu2FE7LzYW//S1yqv0PP7jnPXvc87x58MILkT9rrCdu0ZpGzz4bzj03tu2E5Oe7fc+f736TmuzDD+HXXxNdChMHNfea0cGYOBH27nWvhw51B0yvtiT3jaSkBiVfo0yGj+7BiKuz6Ji2kls/OZ0/3OjnoindGX95ez5ccRS5u5NZtSmV3+W8y7C+y7nglubI5ZeVXJaWLd0jdFCN1ZIlRd+HJ0TE4p579r8m9MYbhd9LuAULigaCvXthwoSiB/3hxQb8/e67/bczZ467HvX004XTVF1z44wZ0L9/YRBKSXHzJk1yB/EPP3QH9fvuc4FnyRJ4++3C7eTkuHING+bWu/pqV/MKueoql4H5wQduH+PHQ9u27mSjTp3C4P7FFy4wP/SQq8m1bl30M0QL7pEEg642e+mlLiEl9NkaNHDNm6ee6sqtWrS8B2DjRpfvMmgQtGpV+vKA29+0aXDEEdC4cZn2W+DXX2HlysLbFIrLyXHXNkeOdPfpQcnNysV98YV7Pv1097xnjztp6hB2E3tWFjzyiPvNIgwxYypGzW6mqyjFmqL06F6uGW7mTMaOLXpMPv54l68QuvVo5Eg496+9yMprwK5gbSZtO4U9wVSOOTpI3cWzyLj3/+jYr5m7jgMuhfyNN4ruv1MnaNq0xJ7KK8QllxQmSZx/Prz/ftm3deKJRe/n6tvX1exSUgprTt26uQP1E08UXfeHH+DPf4avvio6PbSNSNPGjXPBCODTT91RO9xxx8Fzz7nXDz3kjuzF/fOfLiiGOumdOtUd+ObNg48/dsF92zbXlPrYY678+fnw4IPoJ5P5cV83cgNCn/Sf8UtYjXLmTHeLwq5dLsA3a8bW3HTyV62l1rqlpB/dkY2BRox/00+dtYvotuR9OnVUvjr+Pv77xnLmLUsnp3Yj8jdtpkX3Qxg7zs+mTW5z2dmuwhoIQKvG++g/7ynaDu0DJ59M1ne/svn6++hySqMif8CqbvmkJO9uBFUQYc8e9/G2bXO5NTt3uhh79tlwyCDv/8fw4fuflIR+s5tugg4d0MVLyApksu/fX1C7duF5QImtnsWbgW+/3X3306bBI48wd0Njlm9rwO9+fZKUv4xgasY5LFgA2Vt9/Pqri4Gnn154yXLmTHfe07WrK39qKq4V4dprXaA82GuPS5e65KlyaqevSs10FowqQvH/EGHvVd0lljVr3KQXX3THoexsdxL/889wYtIPTJ9Xi311MmHfPkhKhsaHeGfCQpcu8OANG2lzdCMkOcmd/aWmuia0Nm0Ks+3Crm+Vu+TkwibLSNq2dWe/VdERRxQ2DQ0YsH8gAxeEGjWC6dPh9dcBCKifd7eeigAXN/h8/+NLr16FfxM9e7pgFQqmn3xC/nXX89KcYxmbfU7BKrV8+zgvYwq3Np6AT9Q1ZY4dC8CSva14dt9wvq8/CBb/ij8/j8yk7Wyu3RZ2FLuVoH17WLaMI9JW0ezIQzh13es8yP3k1W3gmkgbHeKOwKEbrH1C8rbNnFznJ7KadePndZmwezfPH/oEHe6/hP/mHsOn0+oy69tdkJ9P5/b7OFcnMnuOsGHA5Sz6JUDO0vWA4EtNpk56kB3pzWjcSHl8y5V0qrWSXfm1+G343WRcehb16oWVdeRIciZP4esG5/PizN5sCmS632TTJmh0CI0yg5xzfjK9evvo3Rt04yb0m29ZO2M9C/69jCTJp+HDt7Nge3OWPPAWefk+6g/uy6J/zefnPe3R+hlk7FpLA/9OVgRaQu10ah3alAZZi1mf14jMjk3o33wRKzanM2tTS/d3npxCw4bKgAFC0rJf8U3+mHTfHpo+9H8c078uTdN3kvXJj3wy6md2Dr6MtA3LaTyoF8GkFDrm/8JhhwZJWb7Iddrcv7/73X/8Ee6/n9mn3MaHe86ga99GrFvnKutFvo8DYMGonFX5YJSd7c50QhffiwWnrCz3/+qww4q2tCxZ4hLZcnOhTdO9tGibzEVD/Rx2mFv188/diVSov9a77y6W5OadlRbx3nvw1FMuqA0b5ppG6tVzTU+XhTUHnniiq1G9/LJrBgwfLuPII902tm51TUWHHALnnFN44Dr8cDfu00UXuTT5wYPdqXAkt99eWHu54grX/FVO1uQ2YWugLkfWWlrwNewJpvLdru5M3n4CzZO38MfGb5HiC7A2tzGvZ5/F2fWn0jFtJXsP70a9pbPZkZ/OnWtvZU8wjSbJ2WwJZHBSnTks29eCM+r9wMl1Zhdse01uExbtbcvCvW1ZurcVW/PrsdAbfmRQve+5ufE7+CVIw6TCwBDquSNJ8vGJElRhr6bwTdebeezD9uzIT6e2by9n1Z/GkbWW8lr22Szf14Jjai/gry1eBEBQvtl5NE9mXUKyBDj34lps/fi/ZAfqsyavCYenrubaRh/SKGkbb/x2JtsCdTk34xv2aQq9ai8oKP/iva2ZuK0fbVI20C51HbVefJLDbj+bNMll8b7WvJD1e77f3Y1M/w4aJm1n2b6WBCn8+2qevIU+6fMQUSZu7Uc+PjL8O2ncvi5N1s+hS61l1PPt5rg6P9Mk6Tdm5nTmrxuuYXOgAa1TNrIlkEFOMI16SXu484os0gacwI5/TmL59C38Z0dvsgINODx1NQPr/486vhyC+NgSyGDStpPZ7GsCfj+ZnZvC0qX8tj1yz/yNk7ayO5jG7mAtGidt5ZS6s+lWawnvb+1PPf9ueqUv4LyMKfglHx/KzJxOvF7nRuYsTKVtygZOqTubSzI/49sLn+fFUdvZknEYqWlCYN1GctVLgKmdTv19m9ie7242T5E88jQJTUmD3H0F046vM49+dWfSInkzW/PrsjM/nf/s6M33u7vhQwl26oLf7+6BP/nksv0fKC0YichA4GnAD7yiqvuNryMiFwIPAArMVdUIF4YPngWjRIiWQVbM+vUuTnTrVvIyK1YUBqFWreCll/Zvxv/5Z8jIcNNTRz0I//63uy4Tin7Z2XDGGWzOyyD10guod9f1Ltvu/PPhhhtcU9SPP7rmqPPOc9dFwoWC0d/+5mpld9zhrinddVfh5wXX3BJq0gL48UcCvY5jep0BtHrjb/w0eQNt139P2z7NuPmyrVw/eAPHzwvLijvvPPjgA7Z1OIbVS3PppnP3+z72BFN5Juti3tvq7mfqV3cmV9+eyTfPzS9SywD4fYOvGFTvvwxf/efCtPz0dFIOa0PfnE+Yv7Iu6/Ma4atbh+DOXcV3ReuUjdzS+B3GZ/+OeXsOK5he+5A6HLJ9KedmfMPK3GZM2naK27RvD8MbTSTdv4ffAvV5dctg9mkyaZLL8c1X8PW6I4psf1SL5xhQd0ZBwAiq8N7WU3l802VFAgHAEWmreKLlkzRJjpCoUU6CKq5GBuzIT+eT7SeSq0m0T13LCelzC8q5aG8b1uU25uS6s0mW/BK3tyM/nVe8e/R6p/9C91qL+WxHH1bnNt3vs93W+C161l5UsP/wMu3VFP6ZfSaL97UmVVyNsEXyZnrWXsTeYAqrcpvRNDmbY9IXsC+YTJ4mUce/56C/C6CgPJvzMtiWX5dvd/VkY14jmiVvYWC972mRspmcYCpbAhn4CTJ3z+HM39Oer3f2Yksgo8g2myT9xuCMb7ms4WSWyuG0eug6Gvxu/x5hYhUtGImIH1gMnAasBX4EhqrqgrBlOgDvAv1VdauINFbVrDIXKFpZLRglwMcfu9pIWU93IsjOdsf5JUvgqKNchcbnc61Ll15auFxmJrwzfh+/rdjOm182Ztky6NEDTumTy6tnf8iMvO40PPYwRv3Dx1FHwfof17F4V3NOPElImjcbhg9H7x6B/P6CogUYPRpeecX1SF6nDlmjP2Bu80H0Pzsd/7dfu6AEMHMmuas3su6+0bRplos+/DduuHovs+cngy/y2ez/u2UzZzSZg2/fHsjN5ds/f8Ed8jjUqUvnFZ/QMGk7x6bPJ199vLf1VLIadiQ3tR7tutSmT8ffePvTDIL4XBVz2VKOT5/HNUf+yAcd72XyF0mwfh3UqsWf6o3h9eyzyEptSYs+bVi3Dli4gCdaPknXZ//A4qU+mr34F77ddTRn1PsfH5w1llef2gGAD+XGxu9yeOpq9gRT6ffraAR1tb4JE/jfriP5dMfxzK7fn41Z4pIODutA13Wf0a3WUuZ0v5LFmzPpWHs1PXN/oGPWt/StM4sUX8BVedPS3LWon36CHj1Y1PJUPvypNfWzl7NrxRZOrjub3rV/ObBLDSef7K6fxIPXDBiV3++uiwG78msVBIf8m25l5j/+Q1B91PHn0CltJUkDTjmw3u4zMyNnT1YiQRUW72vN2twmHJK0lQz/TlqlbNov2PL99/tnv8aolGDUB3hAVc/w3t8DoKp/D1vmEWCxqr5SpgIcSFktGFUf+fnwzjvu+NeggbtWPnKk6/SgTx+XmDV9epQN7NvLyf2SmDU3id27C1vbQo49FpomZ7NocyZvvCFFLhovWhBkxnf76HlCLebPd8Fw+3Z3yerQQ2Hf3IVk5TekZY/GrFjhKl1durjj0bx50Lu3u55/+unuevWOHW7/X3/tmiIBOneG7kcpb7+2B3/d2px2Gmycs4Ety3eyOtDM1fKyNlPvsMb8aYSfgQNdK+WyZe6SztFHw439FuC/+goYPZpgz168/TZs/CWbi4b6aOHfSM6f7mfbnx+leZ82LF8O/k8/ps2M91wSg4hr3pw+3VVVMzJYNC+X2bOCHNk9iSNbb4dbb3XZfRMnukKrumt43olHcPqPzJ8XJLB9Nzuox0l39HYJCd71QxHcD3XKKW69UKIDFN7j9sgj7jpD6EcPBt2Zx/PPuyDXu7fLjjzySPeHkJXl1gVXc+3VyzW7hpIkxo1z2WVHH+22v2aNu3L/r3+5m5IbNoT773fr33ab+2x5eS7RYuhQN/3mm+HZZ93rMWPcLRArV7o/oI0b3YlKuDvvdD92ZqZr8v3nP10ZOnd2yRwffeTKCu5s6pprXIJAs2aFGYSjRrn72p591mVannuuuzn7rrvcMv37F9bgX37ZXUMNfQf33VcQCPH5XIJPqJm5WzdXu1d1fUiuWFF0SJjjj3cZHj16uNaFkoaLAffbDRrksjomTdp//uDBbnrfvq6lJHR7Rb167j8BuOu8ZUyMEJFc4OewSWNUdYw37wJgoKpe672/HDhWVW8KW/9DXO3pBFxT3gOqGpdRRC0YVUPDhsHcsNar555zrWzgspl/+skFiAED3N/+jh3ulp3zznPHsS1bXMvcqlWu44ZIo6sff7w7Pk2bBr/84o5P4X9KPXu6jKoZM/Zft1kzOOkkl6iWleWOY1deGfmzBAKuzE89VXT63/5WmK0L7tj1ySfuWHnEEeWWjHTgSrqpdfJkF2WL36S7fLk7623Zsuj07dtdTe6QQ4pO373bXaeLRtVl/Q0Y4A6qO3a4iL5+PZx5ZvQbbnfudD9K+/ZFp5fUtDx9ulv+7LMLlwlv/gWXmj1qlPtxtmxxfxjt2hX9TK++6oZyCa8BXHaZ69rq8ccjlzUQcIEzfFuq7g/mjDPc5zz1VBek7rnHnU2Fmo7BfR+vvOL2e8ghLpi3bu2anMPt2+eSRF591f3RhYIkuEzCsWNd0MnJcW3hKSkuoWfdOndz9kUXFWZZXnih+0yBgCtP164uxbB2bXdmNnq0+1sYPBhee82V6YUX3H/MMiilZhRLMPoYyAMuBFoCU4EjVXVbmQoUrawWjKqfHTvcydx997mTwz/+sWzbCZ1wg/s//u9/u5PJNWv2zzM45xyX/bxggTs2XXKJO3l95x13YhgMuv/ve/e6kz4Rt63s7NhvU1F1wXPJEndSmrCAUxN9/bX7Y/Bu4o7oAK6FVpiVK90ZVUqKCxZpaWXr/WLBApdgcyScdccAAAV9SURBVNttRdu98/JcFmSPHkWX37nTBZFbbnGBef16V7MbNarsqXFlUA7NdKOB6ar6mvf+K2CEqv5Y7mW1YGTK4pNPXO2re3dXEyl+Im1qoDlz3FlQ6J636mb5cneLQhXqyqmUYJSEa4IbAKzDJTBcoqq/hC0zEJfUcKWINAJ+ArqrarmP6mnByBhjqqkYUrvPBJ7CXQ8aq6oPi8hDwExV/UhEBHgcGAjkAw+r6tslbe+gymrByBhjqqeqdNNr1alvGmOMqbYsGBljjEk4C0bGGGMSzoKRMcaYhLNgZIwxJuEsGBljjEk4C0bGGGMSrkrcZyQiQaCs/b0nAYFyLE5VYJ+5ZrDPXDMczGeupapVotJRJYLRwRCRmaraq/Qlqw/7zDWDfeaaoaZ85ioRMY0xxlRvFoyMMcYkXE0IRmMSXYAEsM9cM9hnrhlqxGeu9teMjDHGVH41oWZkjDGmkrNgZIwxJuGqdTASkYEi8quILBWREYkuT3kQkVYiMkVEFojILyJyqzc9U0S+FJEl3nMDb7qIyDPedzBPRHom9hOUnYj4ReQnEfnYe99ORKZ7n+0dEUnxpqd675d689smstxlJSIZIvIvEVkkIgtFpE91/51F5Dbv73q+iEwQkbTq9juLyFgRyRKR+WHTDvh3FZErveWXiMiVifgs5anaBiMR8QPPA4OAzsBQEemc2FKViwBwh6p2Bo4DbvQ+1wjgK1XtAHzlvQf3+Tt4j+HAixVf5HJzK7Aw7P0/gCdV9TBgKzDMmz4M2OpNf9Jbrip6GvhMVTsCR+E+e7X9nUWkBXAL0EtVu+JGH72Y6vc7j8ONnBrugH5XEckE7geOBXoD94cCWJWlqtXyAfQBPg97fw9wT6LLFYfPOQk4DfgVaOZNawb86r1+CTeGfWj5guWq0gNoiftP2h/4GBBgC5BU/PcGPgf6eK+TvOUk0Z/hAD9vfWBF8XJX598ZaAGsATK93+1j4Izq+DsDbYH5Zf1dgaHAS2HTiyxXFR/VtmZE4R92yFpvWrXhNUv0AKYDTVR1gzdrI9DEe11dvoengLuAoPe+IbBNVUPdpPz/du7eNYooCuPw70A0YgRdu0gEDYitWgW0EJQUQbRJJyjqP2ArVvYiWNkoFiIKagjBRvCj9iMgKiq6QdEENcEiglWKY3HPxDE2bjbmMpf3gYHMvVPM2Tdwdu5ctl7XYs0xPx/XN8l2YA64GkuTl82sj4JzdvcZ4DzwCfhCym2SsnOudJpr4/NequRmVDQz2wDcAU67+4/6nKevSsXs2TezQ8Csu0/mvpdV1APsAS65+27gJ7+XboAic24BR0iNeAvQx9/LWcUrLdd/VXIzmgG21s4HYqzxzGwNqRFdd/exGP5mZv0x3w/MxngJn8Ne4LCZfQRukpbqLgKbzKwnrqnXtVhzzG8Evq/mDa+AaWDa3R/H+W1Scyo554PAB3efc/cFYIyUfck5VzrNtYS8/1ByM3oK7IidOGtJL0InMt9T18zMgCvAG3e/UJuaAKodNcdJ75Kq8WOxK2cImK8tBzSCu59x9wF330bK8aG7HwUeAaNx2dKaq89iNK5v1DdNd/8KfDaznTF0AHhNwTmTlueGzGx9/J9XNRebc02nud4Dhs2sFU+UwzHWXLlfWv3PAxgB3gFTwNnc97NCNe0jPcK/AJ7HMUJaK38AvAfuA5vjeiPtKpwCXpJ2KmWvo4v69wN34+9B4AnQBm4BvTG+Ls7bMT+Y+76XWesu4FlkPQ60Ss8ZOAe8BV4B14De0nIGbpDeiS2QnoBPLSdX4GTU3gZO5K6r20M/ByQiItmVvEwnIiINoWYkIiLZqRmJiEh2akYiIpKdmpGIiGSnZiQiItmpGYmISHa/APmVCzEBT/oxAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac61da1438>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac506c7358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# K-fold cross validation\n",
    "np.random.seed(10)\n",
    "\n",
    "kf = KFold(n_splits = 4, shuffle = True)\n",
    "cross_val_model_name = os.path.join(model_path, \"cv_tmp_model.h5\")\n",
    "\n",
    "# Dataframe to hold predictions on val and ground truth\n",
    "pred_vs_gt = pd.DataFrame(X_Y[[\"id\", target_score_column]])\n",
    "pred_vs_gt[\"ANN_score\"] = np.NaN\n",
    "\n",
    "i = 0\n",
    "\n",
    "for train_index, val_index in kf.split(X_scaled, Y):            \n",
    "    X_train = X_scaled.iloc[train_index]\n",
    "    X_val = X_scaled.iloc[val_index]\n",
    "    \n",
    "    Y_train = Y.iloc[train_index]\n",
    "    Y_val = Y.iloc[val_index]    \n",
    "    \n",
    "    # Fit ANN    \n",
    "    mlp, history = fit_scoring_ANN(X_train, Y_train, \n",
    "                                   X_val, Y_val,\n",
    "                                   cross_val_model_name, \n",
    "                                   n_epochs = 1000,\n",
    "                                   score_max = score_max[target_score])\n",
    "    \n",
    "    # Plot history\n",
    "    plot_history(history)\n",
    "\n",
    "    # Load best model\n",
    "    # predict held back validation samples\n",
    "    # add predictions on val to merged summary \n",
    "    mlp.load_weights(cross_val_model_name)           \n",
    "    Y_val_pred = mlp.predict(X_val)\n",
    "    \n",
    "    pred_vs_gt.loc[val_index, \"ANN_score\"] = Y_val_pred\n",
    "        \n",
    "    i = i + 1\n",
    "    \n",
    "os.remove(cross_val_model_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Make plot of cross-validated data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Figure saved\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac691d84e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_cont_score_vs_path_score(pred_vs_gt[\"ANN_score\"], pred_vs_gt[target_score_column], target_score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save ANN predictions (4xCV) with ground truth\n",
    "summary_results_file_name = full_results_file_name[:-4] + \"_ANN_summary.csv\"\n",
    "\n",
    "pred_vs_gt.rename(columns = {\"ANN_score\" : \"ANN_\" + target_score_column}, inplace = True)\n",
    "pred_vs_gt.to_csv(os.path.join(base_path, \"ANN//result//\", summary_results_file_name), sep = \";\", decimal=\",\", float_format='%.3f', index = False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train model on all data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZgAAAEiCAYAAADEasRGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzt3XecFPX9x/HXh6IgKoqAKAYRe4kFUNGAooRExR5/9ggx2HsNKDFYsMUC9pIYJGqIvSB2RaVIU4MEECtEkGYBC9Lu8/vjO+vO7W292729g/fz8djHTvnOzHfm9uYz3zIz5u6IiIgUW4NyZ0BERFZPCjAiIlISCjAiIlISCjAiIlISCjAiIlISCjAiIlISCjB1hJntZmZvm9n3ZuZmtquZDTQzT0n3uZm9WK58pjKz7lF+u5c7L1KZmbWP/jb9irjOPtE62xdrnRm242Y2sJTbkNJTgKkDzKwB8G9gE+Ai4PfArLJmSuoNMzvezM4vdz7KzcwOqo2gZGbrmtkPURDcK0OagdH8hWa2bpr5U81sVMo0z3RBYGZH1ccLOQWYumFTYGvgdne/190fcvdvgGuApuXNWk5vEfL4VrkzsgY7HljdAkxTwu+/EAcBfylBXlIdCTQG5gAn5kjbEji7wPVflC4o1UcKMHVD6+h7cXyiu69095+KtREzW8vMGhVrfQDuXuHuP7l7RTHXWxeYWbNy52FNFf2mVpY7HxmcCLwCPAgcbWaNs6R9n8ICxvtULyjVSQowZWZmQ4HJ0eg/omLwqGhelTaY2HL7m9lkM/vJzGaa2Ukp8xNtIydE65kNLAU2i+a3NLP7zGxetI6pZnZKmu0cbWYTzWyJmX1nZtPN7M9pttM9Nm1jM/ubmf3PzJaZ2Xwze8HMdizguDQyswHRvi01s6/NbLyZHZmSbhsz+5eZLYj24yMzG5ySZhczGxntww9mNsrMuqWkSbQt7G9mt5nZfOD72PzmZnaLmc02s+Vm9qmZ/dnMGua7T9F6hkb53NzMno/a3Oab2fWpwT/K06vR32hZtG/9oyrVRJpRQC9g81gVS5XfjJmdYmafROuZaGa7p0mzr5m9GR2jxWY2wsx2ynO/fmdmk6K/1VfR36RdmnT/Z2bTYr+5I6Nj8nlKukptMLl+D9H/0VmxZROf9vnkP19mtgmwPzAc+BchGPw2yyJXRmnOyXMT44GXgIsLCEp1VlGvZqVa7gU+Aa4C7gPeBubnWKYD8ARwPzAMOAF40MyWufu/U9JeBlQAQwADvjezJsAbwHbAndH2DwfuM7ON3P16ADP7NeEf6XWgP7AK2BbomiN/jwO/BO4APgNaAfsC2wD/zbFswl+Ay4G/AxOAZsBuwB7Ak1H+dgTGRPt3H/Ap0B44hqjKyMy2JxzTH4C/Aj8BpwCvmllPd0+t2rsd+AYYBDSP1tGUcLzaA/cAn0f5GAhsDvTNc58SGgAvEq5W/wR0j76bA2fE0p0FTAdGRvnuAVwbpUvU0yfyuRlwQYbtHQOsS/itOXAp8KSZdXD3FdE+7ge8TPh7DQSaRNsfY2a7u/vMTDtjZicC/yRcKPUn/L3PBbqa2W7uvihK14vQ1jiV8LvcAPgboaopl1y/h3sJVc09CW2YCQvzWHchjgNWAE+7+3dmNpVQohmRIf07hL/1RWZ2u7t/nyFd3EBgHCEoXVfzLJeRu+tT5g/QmfCP3ydl+sDwJ6o07fMo7XGxaU2BGcBsoEE0rXuUbjbQLGUd50bzesemNQReJZzINoqm3UqotmuYJe+J7XSPxjeIxi+u4TF5DxiRI80oQilji5TpFht+ElgObB2b1hJYBEyKTesT5Xs80ChlfZcBPwLbpUy/PFpm2wL2a2i0zL0p04cRAuU2sWnrpFn+vmif145NGwF8niZt+2hbi4ANY9MPjaYfHJv2bpRuo9i0raNj93ia49Q+Gm8MzAOmAU3T/C5uik2bQggm68Wm7Rul+zwl7w4MLPD3cAcp/y/F/kTHKX48Er+N9VLSDYz2oQ2wZzTcPzZ/KjAqzT7fEw2/EP091ovGjyL2f1ZfPqoiq58WEK4EAXD3pYQrwV8AO6ekHebuP6RM60W4snsoto5VwGBgbeDX0eTFhCvF3xSQt6WEk1J3M2tRwHKpFgM7mtk26WaaWaJUNNTdP4vP8+g/Mqq++i3wnLt/FJu/iHCi72RmG6es+n6vWvd/NDAaWGSharGlmbUkBGQIJ9NC3ZZm3AgN1Yl8/pjYDzPbMNrmm4S/yXYFbOsJD51GEt6OvjtE69+EUBp40N2/im3/I+BZ4IAsVYGdgY2Bu6PfYWLZUYQSTa9oG5sSSrUPuft3sXRvAh/ksQ9Zfw+1ISoN70aoGksYTrjAOzLtQoC7jydZilkvz80NBDYi/6q1OkkBpn76xKs2qieqMNqnpk2z/ObAx1FQiZueso67gA+BkWY2x8weNLNDzcwyZczdlxGqew4A5pvZaDO7zMx+kXWPqrqCUPXzoZn9N2r/6Byb3yH6npplHa2AdaJ9SJW6rwnpjtc2hKqXhSmfd6L5rdMsk40DH6dMq/L3M7OuZvYW4Qr5aypfFDQvYHuzK208GWw2jL43j74zHadmhFJfOrmWbZ+SLnW/M01Llev3UBALHV7apHxytaf9HlgW5WErM9uKcA6dRu7eZAMpIGBEQekF4MICglKdowCz+luaO0l67r6AcMXWi1DVtDfwDPBsjiAzmFC9cgnhyvPPwHQroA+/h7aRLYHehOqRk4AJZnZptXYmf+mOVwNCO1TPDJ+Hi50JM+tAKCFtQGhbOSTa1p9iecpX6oXEz5updgZrWQl+D3sDX6Z8Ml4ERb/34wkl/A+Aj2KfHYD9o5JgpvxXJ2AMpJ6XYtTIXz9taWYNUkoxiaqDz/NYfhawm5k1TCnFJKpdfl6Huy8nNDKPjP7JriOc5PYmNLCnFVVbDQYGm9lmhAbtywntJnmJrrSHAcOihvaRwJVmdjPJkka2Xk4LCVf/26aZV2Vfs/iEUBf+as6U+TFgKyp3eEj9+x1KOJkd4u4/33RrZlukWV9N3xqYWH+m4/QDoT0g17Ivp1n285R0W6VZR7ppVWT7PUS/40KOw38IATtuXpb03QilsCupWmpuEuXrOOCWLOsYSGjjOzefDLr7BDN7gXDz9cX5LFPXqARTP7Um9AwCfu7l1Bf4gtCQmssIQvXR8bF1NADOI1QBvBpN2yi+UNS28V40ukG6FZvZOlF+4st9QWg3SrtMhvWkbnspoSNDE0Jj8iJCe0Sf1JNuonQVnXReBA4xsy1j81sQroQnuXuuHnsQ2rt2N7ODUmeY2Xpmtna++xWTepI5h3CCHBmNJwL/z6WMaDvp7o/4AdggW6kyG3f/ktB4fVK83Sw6ZocCL6SpTk2YROj1eFrUOzGxbDdC+8yIaBtzCSfmE+NX8Ga2L6FtJqtcv4do8g9R2g3Jwd2/cfdXUz7Z7jk7kVC6vdHdH0/5PETo2Za1mszdJxD+vhcSevXlYyDQgnpailEJpn76CLjLzHYjBJUTCVeQJ6Rpm0nnfuBU4O/ROj4ldFPuQejpkmjo/VvUsPwa8D+gLeEE9yWZ79zfBnjdzB4jXKEvIzRcb09hV2HTo/aHiYSr510IQXSEJ7t6nkNofJ9sZvdG+9EOOJZQRQcwgNBJYbSZ3Umym/IGhJ45+fgroYrqGTN7kNB43ZRQevo/wgny8wL2bQWwj5k9EuV/vygv93myO/BLhM4SI6J9W5vQBpDu7zuJcMEx2MzGAxXuPryA/ED427wMjDOz+0l2U/6JUPJMy91XmNklhCv4t83sIZLdlOcAN8SSX0aoYh1jZv8g/A3OJgSeXCfcfH4Pk6LvO6Ir/5WEDh6pnVwKEgX2o4DXEh0v0ngWGGRm27v79AxpIASMCYSg8XmubUelmJHEOn/UK+XuxqZPtbopv0i42Wsy4QTwUZplu0frPDbDNlsSAs18QhD4L3BKSprfEeqN50VpZhN6X22RZjvdo/GNCPeSTAO+A5YQ/vFPLvCYXEZoRP+acOU4E7gaWDcl3faE+27i6W5JSbML4crxO8JV7pvAPilp+kT70SVDfpoRHl0yMzoWiwj3KlwKNClgv4ZGf7PNozz9QCjd3Qg0Tkl7IKHEuJQQ4AcRqnUqdVcldGR4EPiKEIA8mt4+StsvTT4qdQOO/S0TnQqWAM8Dv8xwnNqn+a1Mivbta0Lvqs3TbPcYQuP/T4S2jEOjv9/0bPnL5/dAqJG5lfB7rUiXz2r+fx4ZrevULGl2jNJcG//fBdqkSTsimjcqzT7fkyb97tG8etdN2aIdEJFaYOGO82PdvUmutGsKM3sfWOjuqW0iUs+pDUZEaoWZNbaqj8PpTihhjipHnqS01AYjtcrC85Vy1bd/7aH3Wr0RNY6vlS2Nu2frpbQmaEt4RM9DwFxCL7PTCVVa95QzY1IaCjBS2y4m9yPV96P+XdE+SXiyQDb15r6TEvmG0G7Yl9AR4AdCO08/jz1BQFYfaoORWhXdQNghR7LJXvnRJnWemXUieWd8Wl68+2hE6oU1OsC0bNnS27dvX+5siIjUK5MnT17k7q1ypVujq8jat2/PpEmTcicUEZGfmVler3RXLzIRESkJBRgRESkJBRgRESkJBRgRESkJBRgRESkJBRgRESkJBRgRESkJBZjqWLAA7rkHZs/OnVZEZA2lAFMdCxfC3/6mACMikoUCTHUk3ky7Bj9mR0QkFwWYmlCAERHJSAGmOmxNf+q6iEhuCjDVoSoyEZGcFGCqQwFGRCQnBZjqUIAREclJAaYmFGBERDKq1QBjZvuY2bNmNsfM3Mz6pMwfGk2Pf97JY737mtlkM/vJzD41s9NLthNhgyVdvYjI6qC2SzDrAlOB84ClGdK8CmwS+xyUbYVmtgUwEhgL7AZcB9xuZr8rUp7TbTR8qwQjIpJRrb4y2d1HEoIBZjY0Q7Jl7j6vgNWeDsx193Oi8elmtidwMfBEdfOaVYMoLivAiIhkVBfbYLqa2QIzm2lm95tZ6xzp9wJeTpn2EtDZzBqnJjazU81skplNWrhwYc1yWlFRs+VFRFZjdS3AvAicBPQALgL2AF43s7WzLNMGmJ8ybT6hdNYyNbG73+fund29c6tWraqXS7XBiIjkVKtVZLm4+/DY6AdmNhmYBfQCnixPrtJQG4yISE51rQRTibvPBb4Ats6SbB6wccq0jYGVwKKSZEwBRkQkpzodYMysJdAW+DJLsnFAz5RpPYFJ7r6iRBkL3wowIiIZ1fZ9MOua2a5mtmu07XbReLto3k1mtpeZtTez7sBzwALgqdg6hpnZsNhq7wHamtlgM9vezPoCfYCbSr5DCjAiIhnVdgmmM/Be9GkKXBkNXwWsAn4JPAPMBB4EPgT2cvfvYutoF30AcPfPCPfK7AO8D1wOnOvupemiDGrkFxHJQ23fBzMKyHZ2/m0e6+ieZtqbQMdqZ6xQqiITEcmpTrfB1FkKMCIiOSnAVIcCjIhITgowNaEAIyKSkQJMdaiRX0QkJwWY6lAVmYhITgow1aEAIyKSU516Flm9oQAjRbZkyRIWLFjAihWlefiE1Fzjxo1p3bo166+/frmzUm8owNSEAowUwZIlS5g/fz5t27aladOmmNr46hx3Z+nSpcyZMwdAQSZPqiKrDp0ApIgWLFhA27ZtWWeddRRc6igzY5111qFt27YsWLCg3NmpNxRgqkNVZFJEK1asoGnTpuXOhuShadOmqsYsgAJMdSjASJGp5FI/6O9UGAWY6lCAERHJSQGmJhRgREQyUoCpDhWTRUrq2GOP5aijjipomS5dunDxxReXKEdSHeqmXB0NorisEoysoXK1RfTu3ZuhQ4dWe/333nsvXuD/18iRI2ncuHG1tynFpwBTHYl/roqK8uZDpEy+/DL5FvMRI0ZwyimnVJqWqVfcihUr8goCzZs3LzhPLVq0KHgZKS1VkdWESjCyhmrTps3Pnw022KDKtObNmzNjxgzMjMcee4x9992XJk2a8OCDDzJ//nyOOeaYn+/92WmnnXj44YcrrT+1iqxLly5ccMEFXHLJJbRo0YI2bdrQv3//SqWc1CqyNm3acMMNN3DyySez3nrr8Ytf/ILbbrut0namTZvGr371K5o0acIOO+zAK6+8QqNGjRg+fHgpDtsaRyWY6lAbjJTSzTfDhx/W7ja33RYuuqgkq+7Xrx8333wzu+yyC2uvvTZLly6lS5cu9O/fn/XXX58XXniB3r17s/nmm9O1a9eM63nggQe45JJLGD9+PBMmTOCkk05ijz324Igjjsi4zE033cTVV1/NZZddxtNPP815551H165d6dixIytXruSwww5jyy23ZMKECSxZsoQLLriACtVMFI0CTHWom7JI3i688EIOP/zwStMuuOCCn4fPOussXnnlFYYPH541wHTs2JEBAwYAsPXWW3PPPffw2muvZQ0wBx98MKeffjoAF198MUOGDOH111+nY8eOPP/888yaNYsxY8bQunVrAG644QZ69OhR7X2VyhRgqkMBRkqpRCWJcuncuXOl8ZUrVzJo0CAef/xx5syZw/Lly1m2bBkHHnhg1vXsvPPOlcY33XTTnI9tybbMjBkzaN++/c/BBWDPPffMuT+SPwWY6lCAEclbs2bNKo0PGjSIO++8k8GDB7PjjjvSrFkzLrroIpYtW5Z1PamdA8yMVatWFbyMqsBqT16N/GZ2uJntHxu/1Mw+NrNnzKxV6bJXxynAiBRs9OjRHHHEERx//PHssssudOjQgZkzZ9Z6PrbbbjtmzZrFwoULf542YcKEWs/H6izfXmTXAGsBmNkuwNXAMKAFcHNpslaHqZFfpNq22WYbXnrpJcaNG8f06dM57bTTmDt3bq3no1evXrRr147evXszZcoUxowZQ79+/TAzPXOsSPINMO2BGdHwkcAz7n4VcB7wmxLkq25TFZlItV155ZXsvPPO9OzZk+7du9O6deuC79ovhkaNGvHMM8/w7bffsvvuu9O3b1+uuOIKAJo0aVLr+VktuXvOD/A1sEM0PBo4NRpuD/yYzzqi9PsAzwJzAAf6xOY1Bm4ApgA/AF8CjwDtcqyze7Su1M92ufLTqVMnr5affnLv1Mn9H/+o3vIiMdOmTSt3FiTyzjvvOOBTp07NmEZ/L3dgkudxzs+3kX8McIOZvQXsARwbTd86Chb5WheYSqheG5Yybx2gIzAIeB9oTqh+e9HMdnb3lTnWvSMhECYszJSwxlSCEVktPPbYY2y44YZstdVWfPLJJ5x//vnsscce7LjjjuXO2moh3wBzDnA/0Bc4z92/iKYfCryW78bcfSQwEsDMhqbMWwz0jE8zs9OA/wLbAx/kWP0Cd1+Ub16KQgFGpF5bvHgx/fv354svvmCjjTaiR48e3HLLLeXO1mojrwDj7p+TcvKPpp9T7AylSLz4+ps80k4ys7WBacA17v5GyXKlBkCR1ULfvn3p27dvubOx2sq3m/KGZrZhbHxbMxtgZplvoa0hM1uLUEX2XKzElM6XwBnA7wgdED4EXjOzbhnWe6qZTTKzSfHuiQVmLnyrBCMiklG+VWSPA8OB+82sBTCWUKpoY2aXufttWZcukJk1Ah4CNiBUw2Xk7h8SgkrCODNrD1wCvJ0m/X3AfQCdO3euXoRQgJEiq6iooEEDPXu2rtNNmoXJ9xe9KyGoQCgpzCI08PcGzixmhqLg8i9gZ6CHu39VjdWMJ+SvNBRgpIiaNWv28yNTXL+pOsndWb58OXPmzKnyZALJLN8STFPgu2i4J+E+GDezCUC7YmXGzBoTSko7Ad3dfV41V7UroeqstHQykCLYbLPNWLRoEbNmzWLlylydJaVcGjVqRPPmzWnZsmW5s1Jv5BtgPgF6mdkThBsrb42mtwaW5LsxM1sX2CoabQC0M7NdCd2L5wKPAbsDhwBuZm2itIvdfWm0jmEA7n5SNH4+8Dmht9lawInA4YSSVmmokV+KqEGDBrRu3brSQxdFVgeFPCrmNkIQeM/dx0XTexLuWclXZ+C96NMUuDIavgrYDDgM2BSYTCiBJD7HxNbRjsqlprWAvxJu0Hwb6Ar0cvcnC8hXYVRFJiKSU77dlP9tZuOAtsDE2KzRwIh8N+buo4Bsl/85iwbu3j1l/EbgxnzzUFQKMCIiGeX9uH53nw3MNrPGZtbY3Ve4++gS5q1uM1OAERHJIu9+kWb2RzObCfwI/GhmH5rZyaXLWh2nACMiklVeJRgzuwQYSGiHSZRaugG3m9mG7q5H9ouISCX5VpGdBZzh7vEHVD5vZtMJDfVrZoDRTVciIhnlW0W2CckbLePGAG3STF/9qYpMRCSrfAPMx0C6NwL9XzRvzaMAIyKSVb5VZFcBj5hZV0KpBeBXwAHAcaXImIiI1G+F3AfzP+BCwp3yANOBru7+TqkyV6epBCMiklUh98GMJX07zJpJAUZEJKuMAcbM1sl3Je7+Y3GyU48owIiIZJWtBPM9kO8ZtGER8lK/KMCIiGSVLcAcWGu5EBGR1U7GAOPuL9VmRuodlWBERLLSO1qrSwFGRCQrBZjqUoAREclKAaa6FGBERLJSgBERkZJQgKkulWBERLLKGmDM7FMz2yg2/oCZtYqNtzaz5aXMYJ2lACMiklWuEkx7Kt9EeRSwXmzcKOBxM6sVBRgRkawKrSJL9xrHNfMsqwAjIpKV2mBERKQkcgUYp2oJRZftoBKMiEgOudpPDHjJzFZG402Bp81sWZ7Lr74UYEREssoVIG5IGX8tTZo185llCjAiIlllDTDu3r+YGzOzfYCLgU7ApsAf3H1obL4BfwFOBTYExgNnuft/c6z3d8DVwJbAJ8Dl7v5UMfOeZqMKMCIiWVSrkd/M9jCz/c1s3QIXXReYCpwHLE0z/1LgIuAcYHdgAfCKma2XJm0iL3sB/wYeBnaNvh8zsz0LzFthFGBERLLKdaPlaWb2p5RpTwLjgFeBaWa2db4bc/eR7n6Zuz8OVKSs14Dzgevd/Ql3nwr0Jtx3c3yW1Z4PvOHug9x9ursPAkZF00tHAUZEJKtcJZiTgXmJETM7FDiUUIXVFZgP/LlIedkCaAO8nJjg7kuBt4C9syy3V3yZyEs5lqk5BRgRkaxyBZitgMmx8V7ACHf/u7uPBfoB+xQpL22i7/kp0+fH5mVaLu9lzOxUM5tkZpMWLlxYrYxGK1KAERHJIleAaQp8Fxvfi1D9lPARsHGR81RS7n6fu3d2986tWrXKvYCIiFRLrgAzi9DjCzNrCewAjInN3xj4tkh5SVTFpQasjWPzMi1X6DI1pxKMiEhWuQLMQ8AdZnYZMBz42N0nxuZ3AbJ2IS7AZ4Sg0DMxwcyaAN2AsVmWGxdfJtIzxzI1pwAjIpJVrhstrweaA78nnPyPTpnfA3g8341F3Zq3ikYbAO3MbFfga3efbWaDgcvMbAYwExgAfA88ElvHa8CE2D06Q4C3zKwf8DRwBLAfoRNC6SjAiIhkletGy1WEe1MuzTD/8AK31xl4IzZ+ZfR5EOgD3Eho97mT5I2Wv3H3eDvQlsD/YnkYa2bHAtcAVxFutDzG3ccXmLfCKMCIiGRVq88Sc/dRpH/kf2K+AwOjT6Y07dNMe5wCSlIiIlJ6WQOMmU3IZyXuvkdxslOPqAQjIpJVrhJMZ0JPsqeBn0qfnXpEAUZEJKtcAeYiwt38vQkN7Q+4+7slz1V9oAAjIpJV1m7K7n6ru/8SOIAQjF43s/fM7Bwz27BWclhXKcCIiGSV19OU3X2Cu58ObALcCpwIzDGz9UuZORERqb8KfVx/R6A74Y7+KcCKYmeo3lAJRkQkq5wBxszamFk/M/sQeAr4Buji7l2ipx2vmRRgRESyytVN+Tng14RXJfcDnnP3lbWRsTpPAUZEJKtcvch6AXMJD4/sD/QP7wWrTPfBiIhIqlwB5oZayYWIiKx2cj2LrH+2+Ws0lWBERLIqtBeZJCjAiIhkpQBTXQ0aKMCIiGShAFMTCjAiIhkpwFSXqshERLJSgKkuBRgRkazyeuGYmaW+KjnBCY/x/9jdpxctV/WBAoyISFb5vtHyYcKbKFNLPBXRdMxsPHCwu39dvOzVYQowIiJZ5VtFdiDwLtADWDf69AAmAQcDe0fTbi5BHkVEpB7KtwQzGDjV3cfGpr1hZhcB97n7jmZ2ATC02Bmss1SCERHJKt8SzJbA4jTTlwAdouGPgY2Kkal6QQFGRCSrfEsw7wI3mtlJ7v4VgJltBFwPTI7SbAXMKX4W66j33y93DkRE6rR8A0xf4FnCWyw/j6ZtDvwPODQa35AQcNYsFRXhrn4REakkrwDj7tPNbDtCg/620eQZwPPuvipK83hpslhHdegAn34K994LZ5xR7tyIiNQ5+ZZgiALJMyXMS/2y9trhe9q08uZDRKSOyjvAmNmuwP5Aa1I6B7j7pcXITFT9tnmaWSPdvVeGZdK1tJ/h7vcUI08ZNWkSvpcvL+lmRETqq3zv5D8HGAJ8QXjDZfykXsyuVLsDDWPjmxA6ETyaY7lTgBGx8XQ93oprrbXCtwKMiEha+ZZgLgEucfeS3kjp7gvj42b2R0JX6FwB5lt3n1eyjKXTMIqDq1bV6mZFROqLfLs/bQA8VcqMpDIzA/4IPOTuS3MkH2Jmi8xsopmdbmYZ98vMTjWzSWY2aeHChZmS5dYois0KMCIiaeUbYB4Dfl3KjKTRE9gCuD9HuiuAYwj5G054XM1lmRK7+33u3tndO7dq1ar6uVMJRkQkq3yryD4ErjazPYAPgBXxme5+V7EzRmhXmeju/8mWyN2vjo2+b2YNgcuBa0qQp6TEvS8KMCIiaeUbYC4gPDm5V/SJc6CoAcbMWgOHAWdVY/HxwPpmtrG7zy9mvipRCUZEJKt8b7TcpNQZSdEHWAb8qxrL7kp4R823xcxQFYk2mJUrS7oZEZH6Ku/7YGpL1LjfFxju7t+nzDsbONvdt4vGDwHaAOOApcB+wFWEJzwvK2lG1cgvIpJVxgBjZjcCV7r7D9FwRsW60TLSHdgaODHNvJYkH1UDoS3oTOAWQoeFTwmN/ncWMT/pqYpMRCSrbCWYbkDj2HAmRX1mvbu/QfSWzDTzBgIDY+MvAi8Wc/t5syh9JwbZAAAc+klEQVSLCjAiImllDDDuvle6YYn06gVPPgmdOpU7JyIidZKeM19du+wSHhfTtm25cyIiUicV8rDLw4AepH/Y5dFFzlf90KCBqshERDLIqwRjZtcCTxC6AAOsSvmsmRo0CC8cExGRKvItwfwBOMndHyllZuqdhg1VghERySDfNpi1CHfIS5xKMCIiGeUbYP5OeKCkxKkEIyKSUb5VZI2AS8ysBzCFqg+7LOaNlvVHw4YqwYiIZJBvgNkLmAGsA3RJmVfUGy3rFfUiExHJKN+HXepGy3RURSYikpFutKwJNfKLiGSU7WGXjwJ93X1JNJzRGn2jpQKMiEha2arIVpFsX1E9UDqqIhMRySjbwy6PSzcsMSrBiIhkpDaYmlAJRkQko0Iedrk3cCzQjnBn/8/c/aAi56t+UIAREcko34ddHg+MAn4BHAgsB9oDewNzSpS3uk9VZCIiGeVbRdYfOM/djyAElwuBHYFHgXklylvdpxKMiEhG+QaYDsAL0fByoJm7O3Ar0LcUGasXVIIREcko3wDzDbBeNDwH2CEabg40K3am6o0GDWDFChivB02LiKTKN8CMBvaPhp8AhpjZ3cAjwOulyFi90KgR/Oc/cNZZMHZsuXMjIlKn5NuL7FygaTR8TfT9K2Ak8JdiZ6remDkzOfzll+XLh4hIHZQzwJhZI+BgQjDB3VcBV5Y4X/XDt98mh5ctK18+RETqoJxVZO6+ErgDWLv02anHli8vdw5EROqUfNtgJgC7lDIj9d4338DKleXOhYhInZFvgLkDuNnMTjez3c1sh/inWJkxs4Fm5imfrPfZmNkvzexNM1tqZnPM7Aozs2LlKavzzksOP/wwXHJJrWxWRKQ+yLeRP/G4/rui78RTli0abljEPH0IdI+NZ7yT0czWB14B3gJ2B7YD/gH8ANxcxDylt/76lcfffrvkmxQRqS/yDTDblzQXla1093yfDnAC4TXOvd19KTDVzLYDLjSzW6KbQUunUd6PchMRWeNkrSIzswfMbD13/zDbp8h56mBmc83sMzMbbmYdsqTdC3g7Ci4JLwGbEp6VVoWZnWpmk8xs0sKFC2uWUwUYEZGMcrXB9CZ5/0ttGA/0AQ4ATgHaAGPNbKMM6dsA81OmzY/Nq8Ld73P3zu7euVWrVjXLrQKMiEhGuc6QtdNYHnH3F+LjZvYO8Ckh0N1Sm3nJS8NiNj2JiKxe8ulFVtp2jGwbdv8e+C+wdYYk84CNU6ZtHJtXWirBiIhklE+AmWdmq7J9SpU5M2tC6BmW6Tks44BuUbqEnsBc4PNS5etnCjAiIhnlc4Y8Ffg2Z6oiMLObgOeA2UBr4M+EpzU/GM2/DtjD3XtEizxCeBbaUDO7BtgG6AdcWfIeZKAqMhGRLPIJMM+5+4KS5yTYDPgX0BJYCLwDdHH3WdH8TYAtE4ndfbGZ9QTuBCYRXitwM7XVXjN3bq1sRkSkPsoVYGq1/cXdj80xv0+aaR8A+5QqT1l161Z12ooV0Lhx7edFRKSOydUGU6u9yOqdZmnetTZtGnTuDOefX/v5ERGpQ7IGGHdvUIvVY/VPujaYP/4xfI8enZw2c2blR/uLiKwB8n3YpaSTq5H/nXfC9/HHwzHHQJ8+oYQjIrIGUICpiVwPbX74YfjXv8LwV1/B1Klw002lz5eISB2gGzlKady48BERWQOpBCMiIiWhAFMMLVoUvsy0afDTT8XPi4hIHaEqspp69FHYaCPo0SN32oSvv4aTTgrDxx8PFRWw7bZwyCGlyaOISBkowNRUh+h1NRdcALfemjv9lCnwyCPJ8fiwAoyIrEZURVYsJ5wAb72VX9qhQ9NP//BDmD49Of7RRzBmTI2zJiJSDirBFNM668C//x3ueamOE04I308+CRtsAMcdF8YnTUqmWbgwdI9u2bJmeRURKTEFmGLbcsvcaXI58kjYOPaam7lzYdNNw/CBB4bveNAREamDVEVWV82PvQl6wgTo3x8WLSpffkRECqQSTH1wzTXhu2nTytOfeAKefx4eeKD28yQikoNKMPXZddeFXmmqLhOROkgBppT22qu463v33fTTTz8d7roLli9PTquogFp4qaeISCYKMKV0++1w7bXFW98XX2Se98ADsPfeYbiiAvbYI/nqAPfivH3z5ptDTze9yVNE8qAAUwpduiSH27ev/e0nHkEzZQo89xycey4ceih89lnldLNnw6hRYXjGjOwBDMKToT/8MKxLRCQHNfKXwm23wcqVYXibbeC+++B//4Orrw7T2raFOXNqto0jj0w//dBDK7+n5sork8NffhkCzo8/Qr9+yXVMmgQnnhiGH30Ujj4ahg+HzTYL62rUSE+FFpGCqQRTCg0awFprJcc7doTDDkuON29e823Mnp1++ty5IZhlMmwYPP54CCTpvPJK+H7ttVDyOewweP31MBw3dSr84x/QtSucd15y+uLF8Oyz+e+HiKy2VIIph1xvwiyVBrHriRtvTA7HOwPcf3/4btQo2alg4cKq6+rTJzk8Zkyolnv33VCNNm4c7LxzCLKbbJL9xWxLloSguN12he3Lf/4Dm28ennhQXV99BXfcEUpza69d/fWISFoqwZRDaoDZcstQTZXa3XjffStXcdVUaikkYeLEqtPuvjs5nM9bOG+6Kaw/UZV27bWhuu7WW+Hyy5OPwUl11lmhei5dj7eKinCvz4oVlae7hw4Mp56aO18JM2fC2LGVp91+e6gyTJTaRKSoFGDK4YgjKo//9a/J4d/9Ljl8882hDadYKirSTz/zzJqv++mnK48nSj+PPAIvvRQ6ByxfHqrQPvoolEAg+XDP3XeHBQuSy7/zDlx1VbjX56GHKq97xIjw/emn6fPSr1/VB48ef3wIgJ98At9/H6YlSnSrVuW/n5nMnZs+SC5YAMuW1Xz9q6vFiyt3r8+XOzz1lN6pVMcpwJRDr16htHL33eFE2q5dct6f/hSCzPDhYTz+TLL67s9/Du/NOe64UAL56qvKbVV33AEPPwwvvwxnn50MJHfeWXk98VLd/Pkh3eLFYdwdXn0VLrwwfR6OOSa5fKIk+cgjoRdddb3/fiitjRgB8+aFNq6Egw6Cc86p/rpr6plnMrfX1QU9elRuw8tm8eLkRcjYsTBoUOhQUxNTpuT3mo3asGxZ+A2tRvev1akAY2b9zWyimS0xs4Vm9pyZ7ZRjmfZm5mk+B9RWvvP2298mH1oJ4aq9UUozWIMG4bljW20VxtdfH371q+T8nTIcjr/8Bbp1K25+i+211yqP//BD5avXkSPDP/tll1VddsqU9EGgVy8YODAcM6hcnZb4R039h33jjXDSfeqpMP7JJ8ledAlLlsABB8DkyTl3i08+Cd8ffBBKT9dfHwJfosSYKM0lehbGzZgBAwZkLl1m8u230Llz1WMa5x56Lsb37aefQsDOt9SQCNwQjknv3qE7+8cfh3UUku+KivTbnTgxrDth7twQtOM+/jgEo4MOgtGjk6XQb7/Nf/vpnHxyuKgp1Ul98ODwe8vHkCHhtzx+fGnyUgZ1KsAA3YG7gL2B/YGVwKtmls87iQ8ANol9Xi9RHqtv0KDq9bAaMiQ5nKgya9WqcppDDql8JdarV+HbqW2Zulqnc/LJ4UTZuXP6+RMmhKq3eJXJySfDGWeEQJ7PtidODOtwDye4RYvgtNPCsbz//rDtVauSHRMeeSQ5DcK2p04Nw199FW52TZg2LdwfdemlldujLr0UXnwxpF2wILyq4b33wrpuugm++y6UuOKdMk4+GQ4/PAz/+98hv19/Hbqfvx772SdO/j/+mJw2bFjo/XfvvVX3f9EiOOqoUAqD0DbVowf07Rt6Jr7yCvz3v6Ht6thjw429e+yR/0NYr78+eTMwwNtvJ4cvvjh8T5kSSoN9+1ZuMzv22OTw+edXPyCsWpU+yL30UvGr2yZODNW7l1wSAmJq0EyVOI4//FDcfJRRnepF5u6/jY+b2e+BxcCvgOdyLP6Vu88rVd7qnL59w5VXt26hOikh3lHgvPPCySVxMlrdHXdc5aqxDz4obPkzzkg/ff785Al5zz2rzv/44/A9cmRyWuKV2Akvvhi+EwHgrbfCyTbe/jNkSDjRxSWqSiGkP//8yvPffTecwEaNgl13DSex668PpaJhw6rmNRFsHnwwrG/DDcN9T6eeGk68n38OBx8cngyRKBW+/34IhIkee6kn9y++CO8nmjw5BORnnw0l9f/8J3wOPxyuuCKUPCDs8403hg4cCTNnhu/77ktOO/fcEGTi1agJU6Yk1zVhQjKYP/BA2O+nnw7/I9tuG4LTdttB48bh+I0bV7VDzYAB4SJtwIDQ67FBgdfeS5bA/vuHv0XifVDx39OFF4a/1ZgxyR6L8+eHi5c776z6u5o9O9zO0Lx5ODZTpoTgn868eaEqPVtvzbgpU2DHHWulN6t5Ha7vM7NNgLlAN3cfnSFNe+Az4H9AE+Aj4FZ3fzxd+rjOnTv7pPryoMgPPwz/TO+9F0oqgweHe1DyMXdu1bvvr7giNKInNGwYnmd22mnFy7Nkd8ABycBTCscfn3wl92WXhWDy7LOVSw5xJ55YtUNFPjp2DCfCeFXQI4+E7UMIPvFSzi23VG0jW2ut9CWLLbYITxGfNi13PsaOrVxCivv978OFWOJEPmFCCCKpJWKzsM1HHw3Vg88/DzvsEEo3I0eG49ikSSiptmoVgta664YLhwEDwjomTQrVofEneiS89VaYt/76IQhec00I6AMHhiD++uthH8aOhdatYb/9QikVQslzxx1DvgcNghYtQkn56adDT80jjgjrmDMnXHxOmRIC9i23JIP022+H17sn8llNZjbZ3TNUJ8TS1fEA8yiwNdDZ3dN29TGzlkBvYAyhSu1Q4HKgt7tX+W8xs1OBUwHatWvXadasWSXKfYlUVIQroUxVRZk880z4QV5wQWjHGTo0uY4rrkgGoHTBKKFr1+RVaEKbNskqFZFyyvaEjObNK7cnjRoVeiGefHL69JMmpf8fO+eccNLPVr376KOhRLHvvtnz26tXCGAdO4aLjXyeW/iXv4QSUGo75U47hfwmXsf+3HOhRAah2nXmzFCN+tBDyWcJPvNMOGbVUO8DjJndAhwLdHX3DP1RMy57V7TcztnS1asSTDG4h7aEgw8OVRhXXBGuiNK94rlPn3CV1qVL6OkGoXj/+9+HDgi77gpLl4bqh65d86sT/9e/kq+Bjst09SpSLjvskLnU9OCDobNDNptuWvcfCrvPPqF0Uw31OsCY2a2E4LKfuxfcf9TMegP3uHvTbOnWuABTiKVLQ8Nu+/bJaoeJE9PX8y5aFK7AUu20UyjKjx8frq422yxZV3700cnH1TzwQOjCnLj6PPzwcGWV2j1ZRIqrmue/fANMXetFhpkNAY4D9q9OcInsCnxZvFytgZo2DT3W1lor1NvfcEPmRsSWLUPpJOGVV0Jd8j/+EQLFkUeGe30aNAhXhv37h2q4LbYI9do77xwafN94IxT9L7882QsuUWfeqVPlbR50EPztb2HbqS66qOq0dA2kBx5YtZt4osdZaqmumlUJVXr7iaxB6lSAMbM7gT8AxwPfmFmb6LNuLM11ZvZabLy3mR1vZtub2bZmdjFwFnB7re/A6mrTTUN31Wy23jp8b7RRaExef/30AWnYsHAj6XbbwWOPhYZMCCf69dYLwcgsTJ8wIdzJ37591cfcNGkSqunOOy80sl52WUg/fnwIXqkNrP36JXt2nXRS6NBw9dXJ6r+EP/whVAVecknl6enqx9dZJzn86KOVxxN5fPjhqgEvXQC8667KbV9XXRUaeh94ADp0qJoeQjXmoYeGao577kmf5he/CDeupu4PhOfEZRK/+be6/vnPmq8jnd/8JneadPtbLjvsUO4cpFfIbQLVVKcCDHAmsB7wGqEEkvhcHEuzCbBlynIDgEnARELV2snuXkduz12DjB0bGheLpUGDEKgefzy0FT3+eDhZ7rcfnHJKSHPggaHB9sgjQ/qGDaFZs/BUgITECfrss8Py555b+R6VhGefDdMT3UgTpZlhw8L24y69NJS+mjYNAatDh8qPp7nwwtAhokWL0O501VXhZA8huB50UDLtVVeF7V5xRTixn3VWmL/WWqF0l+nJ12efHZbZZ59QBXnXXZWfG9eiReiW26IF7LZbcvpBB4UgGn+8T+vW4e+X2M9dd626vauvDstBCJ5/+1vVpxQ0aZIc3n778EBSCA3NTz5ZdZ2NGlVuaG/SJBzvRMnyj39Mngh/85tQbZoa7H/72/D3iNt226rbKpe77w5VwvnKVDV8113J4RtuCDUANdGvX82Wz4e7r7GfTp06uazGOnUKn++/z57uwQfdn3qq6vRly9y//jo5/tJLyXVmMnq0+wMPpJ9XUeE+a1YYXrHCffjw8J2PUaPCdv/5z/B9882Z037ySchr3IIFybwPGJCcft117i++WDntuHHuS5aEtN27u/ft6/7EE9nz9/33If3JJ1c+RvFjGM/DwIGV83LllWF81aqq63766TDvyiuT0959t+q+JNY9dmzl8eXL3c88Mwwn9ivTZ9askP7ss92HDHGfMCFMv/Za9xNOCMN77hnWP3Gi+/jxYR9HjXL/8cfK6xo0yP3uu5PHYfDg5Lxhw6pu+7PP3CdPDulXrnT/6afkvJEjw/SKiqrHKHU9zz8fvt94w/3OO93//nf3GTPcZ89Opnnwwex/zxyASZ7HObbsJ/lyfhRgVnP33uvepUtx1/n99+EEXg5ffFGz5RMnl7lz80u/YkU4oeVr7Fj3xYszB+FVq9wvuCAEB/cQeBIBdsUK92+/Tb/ed94J63v44crTFy0KwSDhkEPcDz88Ob5smfsPP4ThxAnbPeTxxx/dx4xxnzcvBI9Ondzfey/7/q1a5f7MM2FdmfTsGdbVpUvVY1dREfLYqVMYnj8/mb9MEscy29/hllvc33yz8nH/6qvs662hfANMnexFVlvUi0zWKIsXh2rE9dYr7XYSj0RJV81WXVOmhF6Jue6wd8//jvb6YMmSsD/5/M1efjmk22uvkmcr315kdepRMSJSQsV4k2o+ihlYEnbOektb0uoUXCC0QeYrn84PtayuNfKLiMhqQgFGRERKQgFGRERKQgFGRERKQgFGRERKQgFGRERKQgFGRERKQgFGRERKYo2+k9/MFgI1eaVlS2BRzlSSoONVGB2vwuh4FaYmx2tzd8/5Loo1OsDUlJlNyudxCRLoeBVGx6swOl6FqY3jpSoyEREpCQUYEREpCQWYmrmv3BmoZ3S8CqPjVRgdr8KU/HipDUZEREpCJRgRESkJBRgRESkJBRgRESkJBZhqMLMzzewzM/vJzCabWbdy56kczGygmXnKZ15svkVp5prZUjMbZWY7pqxjQzP7p5ktjj7/NLMNan9vis/M9jGzZ81sTnRs+qTML8rxMbNfmtmb0TrmmNkVZvXv1Y55HK+haX5v76SkWdvMbjezRWb2Q7S+zVLStDOz56L5i8zsNjNbqxZ2sWjMrL+ZTTSzJWa2MNqfnVLSlP33pQBTIDM7BhgCXAvsBowFXjCzdmXNWPl8CGwS+/wyNu9S4CLgHGB3YAHwipnFXzD+CNAROCD6dAT+Wfps14p1ganAecDSNPNrfHzMbH3gFWB+tI7zgEuAC4u8L7Uh1/ECeJXKv7eDUuYPBn4HHAd0A9YHRphZQ4Do+3lgvWj+ccBRwM3F3JFa0B24C9gb2B9YCbxqZi1iacr/+3J3fQr4AOOB+1OmfQRcV+68leFYDASmZphnwJfA5bFpTYHvgNOi8e0BB34VS9M1mrZtufevyMfqe6BPsY8PcAawBGgaSzMAmEPUS7Q+flKPVzRtKDAiyzLNgeXACbFpvwAqgN9G4wdG47+IpTkR+AlYv9z7XYPjtS6wCjikLv2+VIIpQFSM7gS8nDLrZcKVxJqoQ1QE/8zMhptZh2j6FkAbYsfK3ZcCb5E8VnsRTiRjY+sbA/zA6n88i3V89gLejpZNeAnYFGhfioyXWVczW2BmM83sfjNrHZvXCWhM5WP6P2A6lY/X9Gh6wkvA2tHy9dV6hBqpb6LxOvH7UoApTEugIaG4GDef8Mdc04wH+hCK1qcQjsFYM9uI5PHIdqzaAAs9uiwCiIYXsPofz2IdnzYZ1hHfxuriReAkoAeh6mcP4HUzWzua34ZwFZ/6AMfUY5p6vBZFy9Xn4zUEeB8YF43Xid9Xo3xyLpKOu78QH48aXD8FegPvpF1IpJrcfXhs9AMzm0x4Gnov4Mny5Kr8zOwWQtVWV3dfVe78xKkEU5jElc7GKdM3BuZVTb5mcffvgf8CW5M8HtmO1TygVbxHSjTcmtX/eBbr+MzLsI74NlZL7j4X+ILwe4Owvw0JNQ1xqcc09Xglaibq3fEys1sJHRX2d/dPY7PqxO9LAaYA7r4cmAz0TJnVk8r1mGskM2sCbEdoXPyM8APsmTK/G8ljNY7QOLlXbDV7Ac1Y/Y9nsY7POKBbtGxCT2Au8HkpMl5XmFlLoC3h9wbhf3MFlY/pZoTG7Pjx2j6l63JPYFm0fL1hZkNIBpcZKbPrxu+r3L0f6tsHOIbQU6Uv4Yc7hNBQtnm581aGY3ETsC+hQXFPYAShx8nm0fw/AYuBI4GdgOHRD3O92DpeAD6Ifth7RcPPlXvfinR81gV2jT4/AldEw+2KdXwIPafmRcvuFK1rCXBRufe/mMcrmndTdAzaE7rpjiOUYOLH6+5o2q8JtxG8QWibaBjNbxgdw9ej+b8m9Ii6vdz7X+CxujP6O+9PaAtJfNaNpSn776vsB6o+foAzCdE7cdWzT7nzVKbjkPjBLo/+SZ8AdojNN0JX5i8J3UDfBHZKWceGwEPRj3ZJNLxBufetSMenO6HLZ+pnaDGPD+Heo7eidXwJ/IV62EU52/EidLF9idAAvZzQ9jKUWHfjaB1rA7cDXxGC1HNp0rQjXAz9GKW7DVi73Ptf4LFKd5wcGBhLU/bfl56mLCIiJaE2GBERKQkFGBERKQkFGBERKQkFGBERKQkFGBERKQkFGBERKQkFGJHVhJm1j17C1bnceREBBRgRESkRBRgRESkJBRiRIonegX6pmX0Svb/8AzM7MZqXqL463sxGm9lPZjbDzH6Tso59zGx8NH++md0af198tI2LzOwjM1tmZl+Y2XUpWdnczF4xsx/NbJqZpT6cVaRWKMCIFM81wB+Bs4AdgOuAe82sVyzNjYRnX+1KeNf5M2bWFiD6fgF4j/Agxj8SnpYbDyDXAn+Opu0I/B8QfzsjwKBoG7sAE4HhZrZu0fZSJE96FplIEZhZM8L7gn7j7m/Hpg8GtiE8IPUzYIC7D4rmNQBmAI+6+wAzGwQcTXgfekWUpg9wL+GhhA2ibZzv7vekyUP7aBunu/u90bS2hKcLd3P30cXfc5HM9EZLkeLYAWgCvGhm8au2xlR+b0bilba4e4WZjY+WhfD6h3cSwSUyGlgL2Cpa/9rAaznyMiU2PDf6bp0uoUgpKcCIFEeiuvkQYHbKvBWER6fXRCFVDSt+XsjdoxcWqjpcap1+dCLFMY3wfqDN3f3jlM+sWLouiYHo9bR7ANOjSdOBLlHVWUJXwvtPPonmLwN6lHA/RIpGJRiRInD378zsJuCmKHC8RXgLYxegAng5SnqGmc0kvDnwTGBzwlsYAe4Czgfuil6H2wG4HrjD3X+En1+Te52ZLYu2sRHQyd0T6xCpMxRgRIrnz8B84GJC0FhCeF3vjbE0/YALgY6EtzIe4e5fALj7HDM7EPhrtNy3wCPAZbHl+wPfRNvaLNresNLtkkj1qReZSC2I9fDa3d0nlTc3IrVDbTAiIlISCjAiIlISqiITEZGSUAlGRERKQgFGRERKQgFGRERKQgFGRERKQgFGRERK4v8BZTf2nwsXBFcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fac696d3fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Fit ANN on all data for application to new data\n",
    "mlp, history = fit_scoring_ANN(X_scaled, Y, [], [], os.path.join(model_path, model_name + \"test_ANN.csv\"), n_epochs = 2000, score_max = score_max[target_score])\n",
    "plot_history(history, save_file_name = os.path.join(model_path, model_name + \"_ANN.png\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save weights\n",
    "mlp.save(os.path.join(model_path, model_name + \"_ANN.h5\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# To verify the model predictions, run on all data again\n",
    "Y_pred = mlp.predict(X_scaled)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_pred = Y_pred.flatten()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plot_cont_score_vs_path_score(Y_pred, X_Y[target_score_column], target_score, save_plot = False)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}