7-ML-predict-fire.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "10e6e413-7f90-46ea-b6d3-bd3a1e690d4f",
   "metadata": {},
   "outputs": [],
   "source": [
    "globals().clear()\n",
    "import numpy as np\n",
    "import glob\n",
    "import random\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "import time\n",
    "import pandas as pd\n",
    "from datetime import datetime\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from imblearn.under_sampling import RandomUnderSampler\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn import tree\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import recall_score\n",
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn import metrics\n",
    "from sklearn import svm\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.ensemble import GradientBoostingClassifier\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "\n",
    "\n",
    "import random\n",
    "\n",
    "def modeling(x_train, x_test, y_train, y_test, MT_model, print_opt):\n",
    "    performance_output = {}\n",
    "    if MT_model == 'logistic':\n",
    "        clf = LogisticRegression()\n",
    "    elif MT_model == 'svm':\n",
    "        clf = svm.SVC()\n",
    "    elif MT_model == 'tree':\n",
    "        clf = tree.DecisionTreeClassifier()\n",
    "    elif MT_model == 'XGboost':\n",
    "        clf = GradientBoostingClassifier(n_estimators=50, learning_rate=0.1, max_features=3, max_depth=3, random_state=0)\n",
    "    elif MT_model == 'network':\n",
    "        clf = MLPClassifier(solver='lbfgs', alpha=1e-2, hidden_layer_sizes=(16, 2), max_iter=1e5)\n",
    "    elif MT_model == 'randomforest':\n",
    "        clf = RandomForestClassifier(max_depth=2, random_state=0)\n",
    "\n",
    "    clf = clf.fit(x_train, y_train)\n",
    "    y_pred = clf.predict(x_test)\n",
    "\n",
    "    \n",
    "\n",
    "    cnf_matrix = metrics.confusion_matrix(y_test, y_pred)\n",
    "    if print_opt == 1:\n",
    "        print('************************************************')\n",
    "        print(MT_model)\n",
    "        # print('confusion matrix', cnf_matrix)\n",
    "        # print('precision score', precision_score(y_test, y_pred))\n",
    "        # print('recall score', recall_score(y_test, y_pred))\n",
    "        print('accuracy score', accuracy_score(y_test, y_pred))\n",
    "        # print('f1 score', f1_score(y_test, y_pred))\n",
    "\n",
    "\n",
    "    if MT_model == 'XGboost':\n",
    "        performance_output['feature_importance'] = clf.feature_importances_\n",
    "    else:\n",
    "        performance_output['feature_importance'] = []\n",
    "\n",
    "    performance_output['cnf_matrix'] = cnf_matrix\n",
    "    performance_output['precision'] = precision_score(y_test, y_pred)\n",
    "    performance_output['recall'] = recall_score(y_test, y_pred)\n",
    "    performance_output['accuracy'] = accuracy_score(y_test, y_pred)\n",
    "    performance_output['f1_score'] = f1_score(y_test, y_pred)\n",
    "\n",
    "    auc_score = roc_auc_score(y_test, y_pred)\n",
    "    # if print_opt == 1:\n",
    "    #     print('auc score', auc_score)\n",
    "    performance_output['auc'] = auc_score\n",
    "\n",
    "    return performance_output\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "207e23ba-b31c-496c-a290-0b0aacadecef",
   "metadata": {},
   "outputs": [],
   "source": [
    "def ML_Performance(x_train, x_test, y_train, y_test, MT_model_pool, print_opt):\n",
    "    # Model Development and Evaluation\n",
    "    all_performances = []\n",
    "\n",
    "    \n",
    "    for i_m in MT_model_pool:\n",
    "        if i_m == 'logistic':\n",
    "            clf = LogisticRegression()\n",
    "        elif i_m == 'svm':\n",
    "            clf = svm.SVC()\n",
    "        elif i_m == 'tree':\n",
    "            clf = tree.DecisionTreeClassifier()\n",
    "        elif i_m == 'XGboost':\n",
    "            clf = GradientBoostingClassifier(n_estimators=50, learning_rate=0.1, max_features=3, max_depth=3, random_state=0)\n",
    "        elif i_m == 'network':\n",
    "            clf = MLPClassifier(solver='lbfgs', alpha=1e-2, hidden_layer_sizes=(16, 2), max_iter=1e5)\n",
    "        elif i_m == 'randomforest':\n",
    "            clf = RandomForestClassifier(max_depth=2, random_state=0)\n",
    "        \n",
    "\n",
    "        performances = modeling(x_train, x_test, y_train, y_test, i_m, print_opt)\n",
    "        all_performances.append([performances['precision'], performances['recall'], performances['f1_score'], performances['accuracy'], performances['auc']])\n",
    "        \n",
    "        \n",
    "        if i_m == 'XGboost':\n",
    "            feature_importance = performances['feature_importance']\n",
    "\n",
    "    # create plot\n",
    "    fig, ax = plt.subplots()\n",
    "    fig.set_size_inches(10, 6.18)\n",
    "    index = np.arange(5)  # the number of used metrics\n",
    "    bar_width = 0.15\n",
    "    opacity = 0.8\n",
    "\n",
    "    color_pool = ['r', 'g', 'b', 'c', 'm', 'y']\n",
    "    for i in range(len(MT_model_pool)):\n",
    "        rects = plt.bar(index + i * bar_width, all_performances[i], bar_width,\n",
    "                         alpha=opacity,\n",
    "                         color=color_pool[i],\n",
    "                         label=MT_model_pool[i])\n",
    "\n",
    "    # plt.title('Model comparison')\n",
    "    plt.xticks(index + bar_width, ('precision', 'recall', 'f1_score', 'accuracy', 'auc'), fontsize=15)\n",
    "    plt.yticks(np.arange(0, 1.0, 0.2), fontsize=15)\n",
    "    plt.legend(fontsize=15, bbox_to_anchor=(1.05, 1.0), loc='upper left')\n",
    "    ax.set_ylabel(ylabel='entropy', fontsize=20)\n",
    "    ax.set_xlabel(xlabel='architectures', fontsize=15)\n",
    "    plt.tight_layout()\n",
    "    plt.savefig('comparison.pdf')\n",
    "    \n",
    "def calImportance(x_train, x_test, y_train, y_test, MT_model_pool, used_fields, print_opt):\n",
    "    all_performances = []\n",
    "    for i_m in MT_model_pool:\n",
    "        if i_m == 'logistic':\n",
    "            clf = LogisticRegression()\n",
    "        elif i_m == 'svm':\n",
    "            clf = svm.SVC()\n",
    "        elif i_m == 'tree':\n",
    "            clf = tree.DecisionTreeClassifier()\n",
    "        elif i_m == 'XGboost':\n",
    "            clf = GradientBoostingClassifier(n_estimators=50, learning_rate=0.1, max_features=3, max_depth=3, random_state=0)\n",
    "        elif i_m == 'network':\n",
    "            clf = MLPClassifier(solver='lbfgs', alpha=1e-2, hidden_layer_sizes=(16, 2), max_iter=1e5)\n",
    "        elif i_m == 'randomforest':\n",
    "            clf = RandomForestClassifier(max_depth=2, random_state=0)\n",
    "        \n",
    "            \n",
    "        performances = modeling(x_train, x_test, y_train, y_test, i_m, print_opt)\n",
    "        all_performances.append([performances['precision'], performances['recall'], performances['f1_score'], performances['accuracy'], performances['auc']])\n",
    "        \n",
    "        if i_m == 'tree':\n",
    "            model = clf.fit(x_train, y_train)\n",
    "            feature_importance = model.feature_importances_\n",
    "        \n",
    "        if i_m == 'XGboost':\n",
    "            feature_importance = performances['feature_importance']\n",
    "            feature_importance = feature_importance.tolist()\n",
    "\n",
    "    # plot feature importance\n",
    "    fig, ax = plt.subplots()\n",
    "    fig.set_size_inches(10, 6.18)\n",
    "    \n",
    "    feature_importance = [i/sum(feature_importance) for i in feature_importance]\n",
    "    sort_index = np.argsort(feature_importance)\n",
    "    feature_importance = [feature_importance[i] for i in sort_index]\n",
    "    used_fields = [used_fields[i] for i in sort_index]\n",
    "\n",
    "    plt.bar(used_fields, feature_importance)\n",
    "\n",
    "    plt.xlabel(\"fields\")\n",
    "    plt.ylabel('importance score')\n",
    "    ax.set_xlabel(xlabel='Feature', fontsize=15)\n",
    "    plt.xticks(rotation=45)\n",
    "    plt.tight_layout()\n",
    "    plt.savefig('feature_importance.pdf')\n",
    "    return feature_importance\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c29aaa6a-49b5-45b4-87b7-4ceb7635dd98",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "************************************************\n",
      "logistic\n",
      "accuracy score 0.7890085167925438\n",
      "************************************************\n",
      "svm\n",
      "accuracy score 0.848626064599068\n",
      "************************************************\n",
      "tree\n",
      "accuracy score 0.9271251807809738\n",
      "************************************************\n",
      "XGboost\n",
      "accuracy score 0.8630885425036157\n",
      "************************************************\n",
      "network\n",
      "accuracy score 0.8493491884942953\n",
      "************************************************\n",
      "randomforest\n",
      "accuracy score 0.8123091756387595\n",
      "-2.970337261116947\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAsMAAAG1CAYAAAAcH/kDAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAABSJ0lEQVR4nO3deXhU1f3H8c83G0lYAyQkASEYBNlElqZoFRBQxB2hVWlVrIo7ClaroghuKFTAFRVaUStaBbHuFkTw54YCLYoKirImYQ8EyEKW8/vjzuAQEshAwgTm/XqeeZI599x7z70Ths+cOfdcc84JAAAACEcRoW4AAAAAECqEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLAVFeoGhELjxo1dWlpaqJsBAEDILVq0aLNzLjHU7QBCJSzDcFpamhYuXBjqZgAAEHJmtjrUbQBCiWESAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYCsup1QAAQOgtXry4X1RU1L3OuWTRQYeqV2pm64uLi8d06dLlw4oqEYYBAMBht3jx4n61atV6Mi0tbXdcXFxORESEC3WbcHQpLS21/Pz8+qtWrXpy8eLFN1YUiPkUBgAADruoqKh709LSdteuXTufIIzqEBER4WrXrp2flpa2Oyoq6t4K6x3ORgEAAEiScy45Li6uINTtwNEvLi6uwDcUp1yEYQAAEAoR9AjjcPD9nVWYeQnDAAAACFuEYQAAAIQtZpMAAAA1R4cObUOy36VLfziY1QYOHJi2fPnyuKUHuX5FRowYkfr8888n5uTkLKlM/Y8//jj+7bffbjBhwoSsQ9lOOKJnGAAAoIa54YYbNr311ls/Vbb+F198UXvixIkph7qdcETPMAAAQA2Tnp5elJ6eXlRTtnM0o2cYAACginz++edxJ510Uuu4uLjO9erVO/G8885ruXbt2r06H3/66aeYHj16HBcbG9uladOmHR9//PFGZ5555rEZGRlt/HVGjBiRmpCQ0Mn/vLCw0IYOHdosJSWlY0xMTJekpKQTTj/99PSCggJ7/PHHG40cObK5JJlZVzPr6t9W2e1I0vr16yMHDx7cIjEx8YRatWp1SUtL63DfffclVe+ZqbnoGUbY6tYtuPoLF1ZPOwAAR4esrKyofv36tUlPTy947rnnVu7YsSNizJgxzfr06dP6m2+++SE2NtaVlpbq3HPPbbVjx47Ixx57bFVcXFzpww8/nLJ169bo5s2bF1a07ZEjRybPmjWr4ahRozJbtWpVmJmZGf3+++/XLy4u1qBBg7YvXbp0w5QpU5rMmTNnmSQ1aNCgpLzt7Ny503r06NFmy5Yt0bfddltWu3btCn788cdaK1asiK2u81LTEYYBAACqwAMPPNBEkubOnftjw4YNSyWpbdu2Bb179277wgsvJFxzzTVbX3vttfrLly+Pmzt37g+nnXZaniSdeuqpu4477riO+wvDixYtqn3BBRdsvemmm7b4y6666qocSapTp05xWlpaoST16dNn1/7aOHny5EYrVqyI+/TTT78/+eST833FOw7pwI9wDJMAAACoAosXL6596qmn5vqDsCSddtppeampqbs//fTTOpL01Vdf1W7cuHGRPwhLUsuWLYvat2+fV942/Tp27Jj3+uuvN7777rubLFiwIK60tHR/1Ss0b968em3bts0LCMJhjzAMAABQBTZu3BidmJi4z8VqjRs3Ltq2bVukJK1fvz4qISGhuGydRo0a7VMW6OGHH84eMmTIxueffz6pe/fu7VJSUk64//77gx7nm5OTE5WUlMQFdQEIwwAAAFUgKSmpaPPmzdFlyzdv3hztH8ObnJxcnJOTs88w1S1btux36Gp8fLybNGlSVmZm5rfffPPN0vPOO2/rqFGjjpkxY0a9YNqYkJBQvHHjxn3aGM4IwwCqRLduwT8A4GjStWvXXZ988km9nJycPflq/vz58VlZWTGnnHLKTknKyMjYtXnz5uiPP/443l9n5cqV0d999118edssT8eOHQufffbZdTExMW7p0qVxkhQTE+MkKS8vz/a37mmnnZb7ww8/xC9YsCAu2OM7WnEBHQDUQAfzYYEZTyrG+cThMHLkyA0vvvhiYu/evVvfeuut6/2zSRx33HH5l19+eY4k/eEPf9g+evTo/D/96U/po0aNyoyPjy996KGHUhs1alQcERHhKtr26aefnt65c+e8Ll265MXHx5e+9tprCSUlJda7d+8dktS+ffsCSXrwwQebnHHGGbkNGjQo6dSp0z4X5F1//fVbpkyZknT22We3vv3227Patm1b8PPPP8f8+OOPsU8//XRmdZ2bmowwDAAAao4qvq3x4ZSamlr8wQcfLL/11luPufrqq1tGRUW5Xr16bX/66afXxsbGOkmKiIjQW2+9teLKK69sMWzYsLRGjRoVjRgxInvWrFkJcXFxFV4V1717952zZs1qOHny5CalpaWWnp6eP23atJ979OiRJ0n9+vXbec0112yYMmVK0tixY5t269Zt51dffbW87Hbi4+Pd/Pnzl99yyy3NHn744dRdu3ZFpqamFl555ZWbqu/M1GzmXIUfQo5a3bp1cwv5yB/2mGe4atHzVrU4n1WL81kxM1vknDvsA5eWLFmyqlOnTpsP935roi1btkSmp6d3vOKKKzZOnDgxK9TtORotWbKkcadOndLKW0bPMAAAwGE0bty4xIiICNemTZvCDRs2RD3xxBNNdu/ebddddx0fDkKAMAwAAHAY1apVq3TSpEnJ69evj5GkE044Ydfbb7/9Y+vWrXeHum3hiDB8BOFrfQAAjnw333zzlptvvnnLgWvicGBqNQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELaYZxgAANQYHZ7u0DYU+116/dIfQrFfhB49wwAAAAhb9AwDldTtIG7ptzDY2wYCAI4oCxcujB0+fPgx33zzTe3du3dbSkrK7quvvnrjxo0bo1966aXEDRs2LImMjNxT/5VXXqk/ePDgVt9+++3SDh06FDZt2rTj2WefndO4cePiKVOmJBUUFERcfPHFm5999tl1r7/+ev2RI0c2y87OjjnppJNyX3nllVWJiYklITzcoxJhGAAA4CANGDCgVXp6esFzzz23MjY2tvSHH36Izc3Njbz00ku3Tpo0KeW9996re+655+7w13/ttdcS2rdvn9ehQ4dCf9m///3vhp06ddo1efLkVV9//XX8uHHjmpaWltqXX35Z55577snMy8uL+Otf/9r85ptvbjp9+vQ1oTnSoxdhGAAA4CBkZ2dHrVu3rtbMmTN/zsjIyJek888/f0/wbd26df4rr7yS4A/D+fn5NmfOnAbDhw/PDtxOrVq1St97772fo6KiNGjQoNz33nsv4YUXXkhaunTpt8cff/xuSVqyZEn866+/3kgSYbiKEYZxVOj23MEMRwh+2AMAAH5JSUnFycnJu6+99toW11133YazzjprR9OmTYv9ywcMGLD12WefbVJUVLQmOjpaM2bMqL9r167ISy+9NCdwOyeddNKOqKhfI1laWlrB9u3bI/1BWJJatWpVkJOTE1VQUGCxsbHusBxgmOACOgAAgIMQGRmp995776fExMSiYcOGpR1zzDGdunbt2uazzz6Lk6TLLrts67Zt26LefvvtepI3ROLEE0/cddxxx+0O3E79+vX3GgccExPj6tWrt0+Zc06FhYVW3ccVbgjDAAAAB6lz584FH3744c/btm3735tvvvljYWFhxIABA44rKSlRu3btdrdv3z7v1VdfTdixY0fERx991ODCCy/cGuo2Y28MkwCq0cJuwQ/F6LaQGSgA4EhTq1Ytd9555+3Izs7ecO2117bcvHlzZJMmTUoGDhy4ddKkSSn//Oc/dxQWFkaUHSKB0KNnGAAA4CAsWLAg7ne/+91xEydObPz222/XfeGFFxpMnDgxuU2bNvlNmjQpkaRLL710644dOyLvvffeZt26ddvRokWLolC3G3ujZxhAyAQ7dzPzNgNHvyPpTnDNmjUrSkxMLH700UdTNm3aFF23bt2Sk046KXfChAmZ/jqtWrUq6ty5887FixfXue2227JC2V6UjzAMAEcJPlxULc4nDqRp06bFb7755soD1Vu0aNHyipZlZmZ+W7Zs5syZq8qWDRs2bMuwYcO2BN1IHBDDJAAAABC2CMMAAAAIW4RhAAAAhC3GDAPYR029ox9T1QEAqho9wwAAAAhb9AwDNczCIHsyux1EbykOL3raqxbnE0BVIgwfxYKdFkhiaiAAABBeGCYBAACAsEUYBgAAQNgiDAMAACBsMWYYe+ECEABAKHXooLah2O/SpfrhYNabOnVqQl5eXgS3Sj5y0TMMAABwkGbMmJHwz3/+s3Go24GDRxgGAACoRoWFhVZcXBzqZqACDJMIkZo6TyYAAKicgQMHpn344YcJkmRmXSVp+PDh2Z9++mndhg0bFvXt2zf3scceS87Kyqq1fPnyb1q1alU0YcKExk8//XSTNWvW1GrcuHHRn//8540PPPDAhsDtfvjhh3XuvvvupkuXLo2vVauWO/PMM3MmT568NiEhoTQUx3m0IwwDAAAchPvuuy87MzMzJjc3N/KJJ55YI0lpaWm7P/3007qLFi2qs3r16tj77rsvMz4+vrRhw4Yl99xzT5OxY8c2ve666zb07t17x9dffx0/bty4pvHx8aV33XXXJkn6z3/+U/vcc89tffrpp2+bNm3aL5s3b44cM2ZMs0suuSTygw8++CW0R3x0IgwDAAAchPbt2xc2aNCguLS01Pr06bMrcNmOHTuiFi1a9H3z5s2LJWnr1q0Rjz76aOrNN9+c/eijj2ZL0oABA3Lz8vIiJkyYkHr77bdvioqK0p133tmsc+fOO9999909wfeYY44pOv/881t//fXXsb/5zW8KDu9RHv0IwwCOatzeGkAotG/ffpc/CEvSxx9/XCc/Pz9i8ODBOUVFRXvq9e3bd8djjz2W8ssvv8SkpKQUL1mypM6DDz64JrDOGWecsTMqKsp9+eWXtQnDVY8wDAAAUMUaN2681xVzGzdujJKkbt26tS+v/sqVK2Oio6NdSUmJ7rjjjuZ33HFH87J11q5dG1M9rQ1vIQ/DZtZO0hOSTpK0TdJUSWOccyUHWK+bpIckdZVkkhZLGumcW1CtDQYAADgAM3OBzxs1alQsSa+88sqK1NTUorL1O3bsWBAZGSkz04gRI7LOO++87WXrNG/efJ/1cOhCGobNLEHSHEnfSzpfUrqkR+VN+Xb3ftY7xrfeYkmX+Ypvk/QfMzvBObe6OtsNAOGKYSdVi/N55IuOjnaFhYV2oHq9e/feFRsbW5qZmRl98cUX7xN0/Tp16rTrxx9/jO3Ro0d21bYUFQl1z/C1kuIkXeicy5U028zqSRptZuN8ZeU5W1Jd33rbJMnMPpe0WdJZkiZXe8sBAEDYa9OmTcGcOXMavPTSSw1atGixu6Le28aNG5fceuutWXffffcxq1evjunVq9fO0tJSLVu2LHb+/Pl1Z8+e/bMkPfzww+vOPffc1hdccIEGDRqUU69evZJVq1bFvP/++/UfeeSRzBNOOKHw8B7h0S/UYbi/pA/LhN5XJT0iqaektytYL1pSsaSdAWU7fWUH/HQGAABqpoO9LXKo3HrrrZuWLFkSf+ONN6bl5uZGDh8+vMIe3QceeGBDampq0VNPPdXkueeeS46JiSlNS0sruPDCC3P8dfr167fzgw8+WH7vvfemXnvttS1LSkqUmpq6+7TTTtvetGlT7txRDUIdho+XNDewwDm3xszyfMsqCsMzJd0n6VEze9BXNkpSjqTXq6mtAAAAe0lJSSn29+pWxvXXX7/1+uuv37q/Or17997Vu3fvnw69daiMUN+OOUHeRXNl5fiWlcs5lyXpNEkDJW3wPS6U1M85t6nqmwkAAICjUajDsCS5csqsgnJvoVmKpBmSFskbatHf9/u7ZrbPVCS+dYaa2UIzW7hpE3kZAAAAoQ/DOZIalFNeX+X3GPvdJm+IxyDn3AfOuQ/k9RKXSPpLeSs4555zznVzznVLTEw8pEYDAADg6BDqMLxM3tjgPXzTptX2LavI8ZK+c87tuWLTObdb0nfypmcDAAAADijUF9C9L+k2M6vrnNvhK7tIUr6k+ftZb7Wks8wsxheCZWa1JHVQxRfdVa9uwc0VqaHV04xQYJ5MAABwpAp1GH5G0jBJb5jZI5KOlTRa0oTA6dbMbIWk+c65K31FUyVdJWmWmT0tb4zxDZJSJD13+JoPHCHC+MNateB8Vi3OJ4AQCmkYds7lmFkfSU/K69HdJmmivEAcKEpSZMB6i8zsTEn3SnrJV/ytpNOdc0uqudk4HPjPEQAAHAah7hmWc+57Sb0PUCetnLKPJH1UTc0CAABAGAj1BXQAAABAyBCGAQAADlL37t1bt2nTpl1RUdFe5dOmTWtgZl1nzZpVz1/2xRdfxJ177rktk5KSToiOju5Sv379E08++eTWU6ZMSQhcPyMjo82ZZ5557OE7iv3729/+1vill15qEOp2VJeQD5MAAADw6/DVV21Dsd+lGRk/HMx6kydPXpORkdHuwQcfbDJ69OgNkrR9+/aIO+64o/lZZ52VM2DAgFxJev755xOGDh3askuXLjtHjRqVmZ6eXrh58+ao9957r/4NN9zQcufOnZHDhw/fXJXHVFWmTZuW2KZNm/xLL710W6jbUh0IwwAAAAepc+fOBUOHDt0wfvz41Msvv3xry5Yti/7yl7+k7ty5M/Kpp55aK0krV66MvuGGG9LOP//8ra+99tqqiIhfv5i/9NJLty1ZsmT92rVrY0J2EGGOYRIAAACH4OGHH85u0KBB8Q033HDMggUL4p5//vmk2267LTMtLa1Ikp588snGJSUl9tRTT60NDMJ+nTp1KjznnHN2lC3/29/+1rhp06YdY2Nju/Tq1avVypUrowOXZ2dnR1144YVpDRo0ODEuLq5zRkZGm08++SQ+sE5xcbFGjBiRmpKS0jEmJqZLq1at2j/zzDMNA+ssXLgw9tRTTz2ufv36J8bFxXU+9thj248dOzZR8oZsfPfdd/FvvPFGIzPramZdH3/88UZVcNpqDHqGAQAADkHdunVLH3nkkbWXXnpp+pIlS2qnp6cX3HXXXRv9yz///PO6HTp02NWkSZOSym5z8eLFdX755ZfYhx56aG1+fn7E6NGjm5177rmtli5dumc4xznnnJO+evXq2DFjxqxNTEwsnjBhQnL//v3bLFiw4PsOHToUStLw4cObTp48ucmIESOyf/vb3+6aMWNGwnXXXdfSzHTNNddslaQBAwa0Sk9PL3juuedWxsbGlv7www+xubm5kZI0efLk1b///e/TmzdvXnjPPfdkS1Lbtm0Lq+rc1QSEYQAAgEP0pz/9advDDz+c991338W/9dZbP0ZH/9qJu2HDhuj27dvnlV0n8KK5iIgIRUbuuaWCtm7dGvXpp58ua9269W5JOvbYYwv79et3/IwZM+oNGjQod8aMGfUWL15c55133ll+9tln75Skc845Z0eLFi06PvTQQ8nTp09fvWHDhsipU6cm3Xzzzdnjxo3LlqSBAwfmZmVlRY8dOzb1mmuu2ZqdnR21bt26WjNnzvw5IyMjX5LOP//8Pb3UXbt2LYiPjy9t1KhRcZ8+fXZV+YmrARgmAQAAcIg++eST+B9++CHezDR37ty6gcucc+XWj4mJ6ep/nH322XvNHtGuXbs8fxCWpDPOOGNXw4YNi7/88svakvTll1/WTkhIKPYHYUmqV69eaZ8+fbZ/9dVXdSRp8eLFcQUFBRGDBw/OCdz2oEGDclavXl0rMzMzKikpqTg5OXn3tdde22LKlCkJmZmZYddRShgGAAA4BCUlJbr++utbnHjiiTuHDx+eNXny5OTvv/9+zwVxycnJRdnZ2XtdINe5c+eC+fPn/zB//vwf2rVrt0+vcePGjYvLljVq1KgoOzs7WpKys7OjGzduXFS2TlJSUtH27dsjJWndunXRktS0adO96qWkpBRJ0ubNmyMjIyP13nvv/ZSYmFg0bNiwtGOOOaZT165d23z22WdxB3c2jjyEYQAAgEMwfvz4xGXLlsU//fTTax544IH1SUlJu2+44Ybm/uUnn3zyjqVLl9betGnTnnEQdevWLe3Ro0dejx498mrXrl1adpubN2/ep4d2y5Yt0f4gm5KSUrRly5bosnU2btwYXb9+/RJJatasWZEkZWVllb3wLlqSEhMTSyQvmH/44Yc/b9u27X9vvvnmj4WFhREDBgw4rqSk0kOcj2iEYQAAgIOUmZkZ9dBDDzW9/PLLN/72t7/Nj4uLc+PHj187b968+v/85z8bSNKNN964OSIiwt14443NKrvd77//Pv6nn37a05v8n//8p/bWrVujunfvvkuSTjrppF1bt26Nev/99+v46+zYsSNi7ty59TMyMnZKUpcuXfJjY2NLp0+fnhC47ZkzZya0aNGiMDU1da/e51q1arnzzjtvx0033bRh06ZN0Zs3b46UpOjoaFdYWHjUZsawGxcCAABQVW688cZmsbGxpePHj8/0l1100UXbp0yZsu2vf/3rMRdccMH2li1bFj355JOrrrnmmpbdu3evdemll25OT0/fnZubG/H111/XXr58eVzfvn33mqGhYcOGxWeffXarkSNHZvlnk2jXrl3eoEGDciXvQrguXbrsvOyyy44dNWpUZuPGjYsnTpyYXFhYGHHXXXetl6QmTZqUXHXVVRsfe+yxlKioKJeRkZE3Y8aMBvPnz6//7LPP/iJJCxYsiBsxYkSzQYMG5bRq1apw69atkRMnTkxu06ZNvn/2i1atWhXMnz+/3syZM+slJiYWt27dujA5Ofmo6TYmDAMAgBrjYO8EFwrvv/9+nTfffLPRM888s7Jhw4Z7DXV4+umn13bq1Kn9yJEjUx577LGsP//5zznHH398wdixY5PHjBnTbOvWrVG1a9cubdu2bd7dd9+9btiwYVsC1+/SpcvO3r1777jjjjuOycnJic7IyNjxj3/8Y1VgnbfffvvnG264odnIkSOP2b17d0THjh13vffee8v906pJ0sSJEzOjoqLctGnTkh599NGo5s2bFz799NMrhw4dmiN5QykSExOLH3300ZRNmzZF161bt+Skk07KnTBhwp5wP2bMmKyrrroqZsiQIcfu3Lkz8rHHHltVtr1HMsIwAADAQejfv/9O59yi8pa1bt16d35+/n8Dy04++eT8t99+e+WBtvvVV18t9/9+++23b6qoXmpqavGsWbNW7W9bUVFRmjhxYtbEiROzylvetGnT4jfffHO/bWrXrt3uzz///McDNPuIddSO/wAAAAAOhDAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLDF7ZgBAECN8VWHr9qGYr8ZSzN+CMV+/f72t781btKkSfGll166LVRtWL58eczxxx/fcfr06SsuueSS7aFqx+FGzzAAAECITZs2LfHNN99sEOp2hCPCMAAAQJjLy8uzULchVAjDAAAAB2ngwIFpHTp0aDtr1qx6rVu3bhcXF9e5a9eubRYuXBjrr1NSUqK77roruXnz5h1iYmK6pKWldXjiiSca+ZdnZGS0+e677+LfeOONRmbW1cy6Pv74441uueWW1LS0tA7+erm5uRFRUVFd2rVrt2coSXZ2dlRERETXWbNm1fOXTZ06NaF169btYmJiuiQnJ59w0003NS0qKtrT5scff7yRmXX9+OOP4zMyMtrExsZ2uffee5PLO7533nmnbu3atTvfeOONTavwtNUohGEAAIBDkJWVFXPnnXc2u+2227KnTp36y5YtW6Ivvvji9NLSUknSkCFDmk+aNCnlsssu2/Taa6/91L9//5xbbrkl7ZVXXqkvSZMnT17dsmXLgp49e26fM2fOsjlz5iwbNGjQ9l69eu1YvXp1rbVr10ZJ0kcffVQnKipKy5cvj9+6dWuEJM2ePbtORESEevfuvVOS3njjjXpXX331sR06dMibPn36iquuumrDs88+2+Tyyy9vXrbdl1122bFnnnnmttdff/2nCy64YFvZ5TNnzqz3+9//vtV11123/sknn8ysvjMYWlxABwAAcAhyc3Oj5s+fv6xjx46FklRaWmqXXXZZ+pIlS2Kjo6Pdyy+/nPjYY4+tuummm7ZI0gUXXLBj/fr10Q8++GDqJZdcsr1r164F8fHxpY0aNSru06fPLv92e/fuvSsyMtLNmTOn7hVXXJEzf/78Oj179tz+3//+t/bcuXPrDBo0KPeTTz6pc/zxx+fVr1+/VJLGjBmTmpGRseONN95YJUmDBg3KlaSxY8c2u//++7PT09P3dBEPHTp04z333LPR/3z58uUx/t9ffvnl+n/+85/T//rXv2bed999G6r5FIYUPcMAAACHIDU1tdAfhCWpU6dO+ZK0evXq6Pfff79uRESE/vjHP+YUFRXJ/+jdu/eOZcuWxRUXF1e43Xr16pW2bds2/5NPPqkjSZ9//nndU045ZcdvfvObnfPnz68rSV9++WXd7t2775Ck4uJiff/99/EDBgzICdzOZZddllNaWqr58+fXCSwfMGBAuTNGzJw5M+GKK65IHz169NqjPQhL9AwDAAAcknr16pUEPq9Vq5aTpPz8/IjNmzdHlZSUqFGjRp3LW3f16tXRgb21ZZ100kk7Pv3007oFBQX27bffxk+YMGFtZGSke+ONNxpu3749YtmyZfG33357tuSNHy4uLrbk5OS9ttesWbMiSdqyZUtkYHnTpk3L3e+cOXMa1K9fv/iSSy7ZVonDP+IRhgEAAKpJw4YNSyIjI93s2bOXRUZG7rO8adOmFXcNS+rRo8fOv//9703eeuutulFRUa579+55kZGRbsyYMce88847dUtKStS3b9+dkpSSklIcFRXlNmzYEB24jXXr1kVLUqNGjfYK7RER5Q8QePjhh9c8+eSTTfr27dv6008/XZacnFxSbsWjBMMkAAAAqsmZZ56ZW1paatu2bYvq0aNHXtlHbGysk6To6GhXWFi4Ty7r27fvDuecxo8fn9KlS5edUVFRysjIyK9Vq1bppEmTklu2bFmQmppaLElRUVFq37593qxZsxICt/HSSy8lREREqGfPnjsr0+b69euXzJkz50czU9++fVv7L9Y7WtEzDAAAUE06depUOHjw4E1XXHHFsQsXLlyfkZGxKz8/P2Lp0qWxP/30U+y//vWv1ZLUqlWrgvnz59ebOXNmvcTExOLWrVsXJicnlyQnJ5ekp6cXLFy4sM6dd96ZKUmRkZHq2rXrznnz5tW/+OKLNwfub9SoUVkDBw48btCgQWmXXHLJ1iVLlsSNGzeu6UUXXbRpf8MxykpOTi6ZM2fOj6eeemqbM84447iPP/74p7p165ZW7dmpGQjDAACgxgj1bZGrw4svvrjmgQceKHjxxRcTx48fn1q7du2SVq1aFVx++eV7guyYMWOyrrrqqpghQ4Ycu3PnzsjHHnts1bBhw7ZIUkZGxo4VK1bEnnbaaTv89U8++eSd8+bNq/+73/1ur97eCy+8MPe55577Zfz48Sn//ve/GzZs2LB46NChGyZMmBD01GgtWrQomjNnzo89e/Zsc9ZZZ6XPnj17hb8n+2hCGAYAADhIM2fOXFW2rE2bNrudc4v8zyMiIjRq1KiNo0aN2li2rl+7du12f/755z+Wt+zll19e8/LLL68JLHvwwQfXP/jgg+vLq3/11VfnXH311TnlLZOkYcOGbfEH7f21W5Jat269Ozs7+9uKtnU0OKrHgAAAAAD7QxgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAOIJs3749wsy6Pv74440O1z5Xr14d3bNnz1Z169Y90cy6vvPOO3UP174r8vHHH8ePGDEi9VC3w+2YAQBAjfHVVx3ahmK/GRlLfwjFfo8U99xzT8oPP/wQP3Xq1JWNGzcu7ty5c36o2/TFF1/UnjhxYsqECROyDmU7hGEAAIAqUlxcrOLiYouNjXWhbktVWrFiReyJJ56466KLLtp+qNvKy8uz+Pj4GnN+GCYBAABwkAYOHJjWoUOHti+99FKDVq1atY+Nje0yb9682r///e/TmjVr1jE2NrZLWlpah2HDhqUWFBSYf73ly5fHmFnXqVOnJgwePLhF3bp1T2zSpMkJw4cPTy0pKdlrH9OmTWuQlpbWITY2tku3bt3aLFmyJLZsO4qLizVixIjUlJSUjjExMV1atWrV/plnnmlYXltfffXV+unp6e3j4uI69+rVq9WGDRsily5dWuu3v/1t67i4uM4dOnRou2DBgjj/embW9Ysvvqg7e/bsBmbWtWnTph39y6ZOnZrQunXrdjExMV2Sk5NPuOmmm5oWFRXt2efjjz/eyMy6fvzxx/EZGRltYmNju9x7773JkvT111/H9urVq1Xt2rU7165du3P//v2PXbNmzZ6O2sLCQhs6dGgz/zElJSWdcPrpp6cXFBTY448/3mjkyJHN/e0zs64ZGRltDuY1pGcYAADgEGRmZsbcc889zW6//faslJSUYklKSEgoHjt27NqGDRsWL1u2LPaRRx5J3bx5c/T06dNXB6577733NjvrrLNyXnzxxV9mz55dd9KkSSnt27fPv+qqq3Ik6dNPP42/6qqr0k8//fSccePGrVm6dGnc4MGD08u2Yfjw4U0nT57cZMSIEdm//e1vd82YMSPhuuuua2lmuuaaa7b662VlZcXcf//9qaNGjcrctWtXxB133NH88ssvb7Fu3bpal19++aZbb711/ahRo5oNHjz42J9++um7iIgIzZkzZ9lNN93UvF69eiUPPvhgZmxsbKkkvfHGG/WuvvrqYwcMGLDloYceWve///0vbty4cU23bNkSOX369DWB7bvsssuOveKKKzbdc889WQ0bNixeunRprd69ex/foUOHvGeeeWZlSUmJ7r///qZnnXXWcd98880PERERGjlyZPKsWbMajho1KrNVq1aFmZmZ0e+//3794uJiDRo0aPvSpUs3TJkypcmcOXOWSVKDBg32/hRRSYRhAACAQ7Bt27aod99998eTTz55zzjaM888c6f/9zPOOGNn7dq1S2+++ea0goKCNYFDKDIyMnZMmTJlnSQNGDAgd+7cufXffPPNBH8Yfuihh5JbtGhR8O677/4SERGhP/zhD7mFhYU2bty4pv5tbNiwIXLq1KlJN998c/a4ceOyJWngwIG5WVlZ0WPHjk0NDMO5ublR//d//7esffv2hZL0zTffxD/77LNNnnjiiVU33njjFklyzmVefPHFrf73v//FdunSpaBPnz676tSpU9qgQYPiPn367PJva8yYMakZGRk73njjjVWSNGjQoFxJGjt2bLP7778/Oz09fU8X8dChQzfec889G/3PL7jggpaNGjUq/vjjj3/yn4+uXbvmn3jiiR1ee+21+hdffPH2RYsW1b7gggu23nTTTVv86/nPS506dYrT0tIKJSmwTQeDYRIAAACHICkpqSgwCJeWluq+++5LSk9Pbx8bG9slJiam63XXXddy9+7dtmLFipjAdU8//fTcwOfHHXdcfnZ2drT/+ZIlS2r369dvW0TEr5Htoosu2ha4zuLFi+MKCgoiBg8enBNYPmjQoJzVq1fXyszM3NP5mZqaWugPwpLUqlWrAknq37//nna0bdu2QJLWrFkTrQoUFxfr+++/jx8wYMBe+7zssstySktLNX/+/DqB5QMGDNhrrPFnn31W96yzzsqJjIx0RUVFKioq0vHHH1+Ymppa+PXXX8dLUseOHfNef/31xnfffXeTBQsWxJWWllbUnENCGAYAADgEjRs3Lgp8fv/99yeNGTPmmLPOOmvb9OnTV8ybN++HsWPHrpGk/Px8C6ybkJCw11f7MTExrrCwcE8+27x5c3RSUlJxYJ3U1NS99rdu3bpoSWratOle5SkpKUW+bUT6y+rVq7fP/nzHsKe8Vq1aztfWCnNidnZ2VHFxsSUnJ++1z2bNmhVJ0pYtWyIDy8u2LScnJ2ry5MnJMTExXQMf69atq7Vu3boYSXr44YezhwwZsvH5559P6t69e7uUlJQT7r///qSK2nSwGCYBAABwCMz2yrd68803G/bv3z/niSeeyPSXffPNN3H7rFgJjRs3Ltq4ceNeeS0rK2uvHlt/AM3KyopOTk7eE2r9PcyJiYkHNZZ2f1JSUoqjoqLchg0b9mqLP5g3atRor30G9mxLUv369Uv69euXc+21124uu+0mTZoUS1J8fLybNGlS1qRJk7K+/fbbWo8//njiqFGjjmnbtm2Bf0hGVaBnGAAAoAoVFBRExMTE7PWd/quvvtqwovr7c8IJJ+z68MMPGwQOEfjXv/7VILBOly5d8mNjY0unT5+eEFg+c+bMhBYtWhSmpqbu1bNcFaKiotS+ffu8WbNm7bXPl156KSEiIkI9e/bcWdG6knTyySfnLl++PO6UU07J69Gjx16PNm3a7C5bv2PHjoXPPvvsupiYGLd06dI46dde7by8PCtbP6hjOZSVAQAAsLeePXvmPv/880kPP/zwruOOO67wn//8Z8PVq1fvMx1aZdx5553rTzvttLZnn332sVdeeeXmb775Ju7ll19ODKzTpEmTkquuumrjY489lhIVFeUyMjLyZsyY0WD+/Pn1n3322V+q5qj2NWrUqKyBAwceN2jQoLRLLrlk65IlS+LGjRvX9KKLLtoUePFceR588MGs3/3ud21PO+20VkOGDNmSmJhYvHbt2ug5c+bUu+KKK7acc845O04//fT0zp0753Xp0iUvPj6+9LXXXksoKSmx3r1775Ck9u3bF/i21eSMM87IbdCgQUmnTp0K97ff8hCGAQBAjXE03AnukUceydq8eXPU2LFjm0rSmWeemTN+/Pg1gwcPbhXstnr06JE3ZcqUX0aPHt30j3/8Y6sOHTrsevnll3/u1avXXnfqmzhxYmZUVJSbNm1a0qOPPhrVvHnzwqeffnrl0KFDcyra9qG68MILc5977rlfxo8fn/Lvf/+7YcOGDYuHDh26YcKECZkHWveEE04o/OSTT5bdeeedqbfcckuLwsLCiKSkpN2nnHLKDv8FfN27d985a9ashpMnT25SWlpq6enp+dOmTfu5R48eeZLUr1+/nddcc82GKVOmJI0dO7Zpt27ddn711VfLgz0Oc67G3ADksOnWrZtbuHBhVW80uOpDD2IfzwXZ5meCP8Znrg16laBX6tatEu3ifFbaEXs+paDPKefzADifFQun8xkkM1vknAvuRagCS5YsWdWpU6d9xosC1WHJkiWNO3XqlFbeMsYMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAACIXS0tLSQ5ofFqgM399ZhfdyDnkYNrN2ZvaRmeWZWZaZ3WdmkQdeUzKzC83sazPLN7MtZvaBmdWu7jYDAIBDY2br8/PzD2ruXSAY+fn5sWa2vqLlIQ3DZpYgaY4kJ+l8SfdJulXSmEqse5Wk6ZLel9Rf0lWSfhJzJwMAUOMVFxePWbVqVcyuXbvi6CFGdSgtLbVdu3bFrVq1Kqa4uLjCbBnq4HitpDhJFzrnciXNNrN6kkab2Thf2T7MrLGkiZJucs5NCVg0q9pbDAAADlmXLl0+XLx48Y0///zzvc65ZNWAb6tx1Ck1s/XFxcVjunTp8mFFlUIdhvtL+rBM6H1V0iOSekp6u4L1/uD7+UI1tg0AAFQjX0CpMKQAh0OoP4UdL2lZYIFzbo2kPN+yivxW0nJJV5rZOjMrMrMFZnZy9TUVAAAAR5tQh+EESdvKKc/xLatIsqQ2ku6W9FdJ50raJekDM2tSxW0EAADAUSrUYVjyLp4ryyoo94uQVEfSlc65l51zH0i6QFKJpBvLW8HMhprZQjNbuGnTpkNsMgAAAI4GoQ7DOZIalFNeX+X3GPtt9f2c5y/wjTteJKldeSs4555zznVzznVLTEw8mLYCAADgKBPqMLxMZcYGm9kxkmqrzFjiMn6Q13NcdiqW/U6qDAAAAAQKdRh+X1I/M6sbUHaRpHxJ8/ez3jvygu9p/gIzqy+pq6Ql1dBOAAAAHIVCHYafkVQo6Q0z62tmQyWNljQhcLo1M1thZn/3P3fOLZT0b0l/N7PLzexsSW9JKpL01OE8AAAAABy5ggrDZhZdlTt3zuVI6iMpUt6cwmPk3Uzj3jJVo3x1Av1J0puSJkiaIS8I9/ZtEwAAADigYG+6kWlmz0ua4pxbURUNcM59L6n3AeqklVO2U9J1vgcAAAAQtGCHSURIuk3ScjObbWYDzSzUd7EDAAAADkqwYThV3vCE/5M3vOE1SWvN7EEza1nVjQMAAACqU1Bh2Dm32zk33TnXS96UaJPkDbW4U9JPZvaemZ1vZqG+MA8AAAA4oIMOrc65H51zt0pqql97i8+U9IakNWY22sxSq6aZAAAAQNU75B5c59xuSe9KmiUpS978v6mSRklaaWaTzKzWoe4HAAAAqGqHFIbNrLtvdokseVOi1Zb0uKQTJf1Z0nJJN8kbTgEAAADUKEHPBOG7W9ylkq6R1EFeT/BiSZMlTXfO5fuqfmNmL0n6QNIgMQUaAAAAapigwrCZTZV3u+R4eXeOe0nS0865r8qr75wrMbN5OsA8wgAAAEAoBNsz/GdJP8u7jfLzzrmtlVhnnqT7gtwPAAAAUO2CDcP9nXMfBrOCc+4zSZ8FuR8AAACg2gU7z3BQQRgAAACoyQ7qVspmlibvIrrOkupL2i7vIrqXnXMrq6x1AAAAQDU6mNkkbpX0oKRoeTNJ+F0g6R4zu9M5N6FqmgcAAABUn2Bnk7hE0nhJOfLmE54nab2kZEmnSRomabyZZTrn/lW1TQUAAACqVrA9w7fKC8JdnHOrA8qXS5pvZi9IWiTpL5IIwwAAAKjRgr0DXTtJr5UJwnv4xgu/Jqn9oTYMAAAAqG7BhuEdkrYdoM42SbkH0xgAAADgcAo2DP9HUr+KFpqZSTrDVw8AAACo0YINw7dLSjCzV8ysReACM2suabqkBr56AAAAQI0W7AV0L8sbBvEHSQPNbI2kDZKaSGouKVLSN5Kme53EezjnXJ9Dbi0AAABQhYINw73KrHus7xGoUznruSD3AwAAAFS7oMKwcy7YYRUAAABAjUW4BQAAQNgiDAMAACBsHVQYNrOLzWyOmW0xs2Iz22pms83s4qpuIAAAAFBdghoz7JtH+EVJgyWZpBJJmyQ1ltRHUm8zO885N7iqGwoAAABUtWB7hq+R9EdJiyX1lRTrnEuRFOt7vkjSRWZ2bZW2EgAAAKgGwYbhP0taJamHc26uc65EkpxzJc65uZJ6+pZfWZWNBAAAAKpDsGG4naRZzrn88hb6yt+U1PYQ2wUAAABUu2DDsJM3Vnh/DrQcAAAAqBGCDcM/SLrQzOLKW+grv0DS94fYLgAAAKDaBRuG/yGpuaRPzKyPmUVJkplFmtlpkj6W1MJXDwAAAKjRgppaTdKzkk6VdImk/0gqNbOtkhrKC9Ym6TXn3DNV2koAAACgGgTVM+w8f5Q3vdpcSdvlBeHtvud/dM5x4w0AAAAcEYK96UYPSbnOuVckvVI9TQIAAAAOj2DHDH8saWh1NAQAAAA43IINw5sllTvHMAAAAHCkCTYMz5N0cjW0AwAAADjsgg3Dd0tqY2b3m1l0dTQIAAAAOFyCnVrtTklLJd0l6UozWyJpvbw70wVyzrkrq6B9AAAAQLUJNgwPCfg92fcoj5NEGAYAAECNFmwYblktrQAAAABCIKgw7JxbXV0NAQAAAA63oC6gM7NRvhtv7K/OqWY26tCaBQAAAFS/YGeTGC2p1wHq9JB078E0BgAAADicgg3DlRElqbQatgsAAABUqeoIw13l3akOAAAAqNEOeAGdmc0tUzTEzHqVUzVS0jGSWkh65ZBbBgAAAFSzyswm0Svgdycpzfcoq1TSFkn/kjT8ENsFAAAAVLsDhmHn3J6hFGZWKmm0c+6+am0VAAAAcBgEe9ONKyT9tzoaAgAAABxuwd5044XqaggAAABwuAXbMyxJMrMkSd0kJci7cG4fzrkXD6FdAAAAQLULKgybWbSkZyRdpoqnZTN5F9oRhgEAAFCjBdszfL+8ccM/S3pZ0lpJxVXdKAAAAOBwCDYMD5b0o6TOzrn8amgPAAAAcNgEewe6JEnvEYQBAABwNAg2DK+RVK86GgIAAAAcbsGG4WmS+ptZ/WpoCwAAAHBYBRuGH5b0qaQ5ZnaamdFLDAAAgCNWsBfQFfl+mqQ5kmRm5dVzzrmDmsMYAAAAOFyCDaz/J28OYQAAAOCIF+ztmHtVUzsAAACAw+6ghzKYWW1JrSXVcc79X9U1CQAAADg8gr2ATmbWzMxmSsqRtFDSxwHLTjGz782sVxDba2dmH5lZnpllmdl9ZhYZxPoRZrbIzJyZnRPEoQAAACDMBdUzbGYpkhZIaiLpLXk34TgpoMoCX9lFkuZVYnsJ8i7E+17S+ZLSJT0qL6TfXclmXSWpaSXrAgAAAHsE2zN8r7yw29c5d6Gk2YELnXNF8i6y+10lt3etpDhJFzrnZjvnnpE0RtKIykzb5gvTD0oaWflDAAAAADzBhuGzJL3lnJu3nzprJKVWcnv9JX3onMsNKHtVXkDuWYn175f0maSPKrk/AAAAYI9gw3ATST8doE6RpNqV3N7xkpYFFjjn1kjK8y2rkJmdIOkKSX+p5L4AAACAvQQbhrdKOuYAdVpLWl/J7SVI2lZOeY5v2f48Iekp59yKSu4LAAAA2EuwYfgzSeeZWXJ5C83sOElnKmCGiUoo7yYeVkG5fz8XS2oj6YHK7sTMhprZQjNbuGnTpiCaBwAAgKNVsGF4vKRYSfPNrL+keMmbc9j3/G1JpfJmhKiMHEkNyimvr/J7jGVm0b52PCIpwswaSPJfbFfbzOqWt55z7jnnXDfnXLfExMRKNg8AAABHs2DvQLfAzIZKekbSOwGL/BfAFUv6s3Puu0pucpnKjA02s2PkjTleVu4a3rJmkib4HoFelfSzpFaV3D8AAADCWNB3oHPOPW9mn0q6XlJ3SY0kbZf0paQnnXPLg9jc+5JuM7O6zrkdvrKLJOVLml/BOjslnVamLFnSK5LukjQ3iP0DAAAgjB3U7Zidcz9JGl4F+39G0jBJb5jZI5KOlTRa0oTA6dbMbIWk+c65K51zxSpzQw8zS/P9+q1zbkEVtAsAAABh4KDCcFVxzuWYWR9JT8obb7xN0kR5gThQlKRK36IZAAAAqIyQhmFJcs59L6n3AeqkHWD5KnkzUAAAAACVFuxsEgAAAMBRgzAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhbhGEAAACELcIwAAAAwhZhGAAAAGGLMAwAAICwRRgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhbhGEAAACELcIwAAAAwhZhGAAAAGGLMAwAAICwRRgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhbhGEAAACELcIwAAAAwlbIw7CZtTOzj8wsz8yyzOw+M4s8wDq/MbPnzWyFb73lZnavmcUernYDAADgyBcVyp2bWYKkOZK+l3S+pHRJj8oL6XfvZ9WLfHUfkfSTpBMk3e/7ObAamwwAAICjSEjDsKRrJcVJutA5lytptpnVkzTazMb5ysrziHNuU8DzeWZWIOlZM2vhnFtdze0GAADAUSDUwyT6S/qwTOh9VV5A7lnRSmWCsN9/fT+Tqq55AAAAOJqFOgwfL2lZYIFzbo2kPN+yYJwsqVTS8qppGgAAAI52oQ7DCZK2lVOe41tWKWaWLGmkpJf2M7QCAAAA2Euow7AkuXLKrILyfSuaxUh6TdJOScP3U2+omS00s4WbNpU3ygIAAADhJtRhOEdSg3LK66v8HuO9mJlJelFSe0lnOedyKqrrnHvOOdfNOdctMTHx4FoLAACAo0qoZ5NYpjJjg83sGEm1VWYscQUmypuS7XTnXGXqAwAAAHuEumf4fUn9zKxuQNlFkvIlzd/fimZ2p6SbJP3JOfdp9TURAAAAR6tQh+FnJBVKesPM+prZUEmjJU0IvBDOd6e5vwc8HyzpIXlDJDLNrHvAgzEQAAAAqJSQDpNwzuWYWR9JT0p6W9444YnyAnGgKEmBt2g+w/dziO8R6ApJ06q0oQAAADgqhXrMsJxz30vqfYA6aWWeD9G+IRgAAAAISqiHSQAAAAAhQxgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhbhGEAAACELcIwAAAAwhZhGAAAAGGLMAwAAICwRRgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhbhGEAAACELcIwAAAAwhZhGAAAAGGLMAwAAICwRRgGAABA2CIMAwAAIGwRhgEAABC2CMMAAAAIW4RhAAAAhC3CMAAAAMIWYRgAAABhizAMAACAsEUYBgAAQNgiDAMAACBsEYYBAAAQtgjDAAAACFuEYQAAAIQtwjAAAADCFmEYAAAAYSvkYdjM2pnZR2aWZ2ZZZnafmUVWYr36Zva8meWY2XYze9nMGh2ONgMAAODoEBXKnZtZgqQ5kr6XdL6kdEmPygvpdx9g9X9JaiPpKkmlkh6R9KakU6upuQAAADjKhDQMS7pWUpykC51zuZJmm1k9SaPNbJyvbB9mdpKkfpJ6Ouc+8ZVlSlpgZn2dc3MOU/sBAABwBAv1MIn+kj4sE3pflReQex5gvQ3+ICxJzrmvJK30LQMAAAAOKNRh+HhJywILnHNrJOX5llV6PZ8fDrAeAAAAsEeow3CCpG3llOf4llX1egAAAMAe5pwL3c7NiiT9xTn3WJnyTEnTnHMjK1hvtqSdzrkBZcpflpTmnPtdOesMlTTU97SNpOVVcAjVobGkzaFuxFGGc1q1OJ9Vi/NZtTifwWvhnEsMdSOAUAn1BXQ5khqUU15f5ff8Bq5X3j/cBhWt55x7TtJzwTQuFMxsoXOuW6jbcTThnFYtzmfV4nxWLc4ngGCFepjEMpUZ42tmx0iqrfLHBFe4nk9FY4kBAACAfYQ6DL8vqZ+Z1Q0ou0hSvqT5B1gv2cxO8ReYWTdJx/qWAQAAAAcU6jD8jKRCSW+YWV/fuN7RkiYETrdmZivM7O/+5865LyR9KOlFM7vQzC6Q9LKkT4+COYZr/FCOIxDntGpxPqsW57NqcT4BBCWkF9BJ3u2YJT0p6SR5432nShrtnCsJqLNK0jzn3JCAsgaSJkoaIC/UvyNpmHOOCycAAABQKSEPwwAAAECohHqYBMows2lmtjCI+qPNjN7wKmRmzsxuDHg+z8xmhLJNAACgeoR6ajXs6355t6OurKmS3q6mtqCGM7NRkq6RlCLpRUk/y7uVeYakupJaOudWhayBAADUcIThKmBmcc65/KrYlnPu5yDrr5O0rir2XVOZWaSkSOfc7lC3pSbxzaAyRtJdkuZJ2ihvFpYVkj6WdF7IGocqYWaxzrmCULcDAI5mDJMowz9MwcwuMLNlZlZgZp/6LvTz13FmNsLMJpnZJknf+spjzWycma01s0IzW2JmZ5Wzj6vN7FvftjeY2Qwzqx+4/4C6Dcxsqpll+eqvMbMpAcv3GSZhZi3N7E0zyzWzHWb2tpm1KlPHmdnNZvaQmW0ys41m9pSZ1aqyk3mQyrwG30kqkPRbMzvfV15gZut95zq6zLon+I53m5ntNLOvzOx037LaZvakmS03szwzW+k75nqhOM4q4J9r+ynn3Be+D1LNnXO9JE2peLWax8yC+Takqvd9kpm95fs3tsvM/mdmfyxTp4WZvWJmm31/O9+Y2eCA5XG+v8fVvn/7K81sbMDyvYbe+Mr2+rdrZkN89TJ8Q3PyJd3mW/aw7z1jp5mtM7OXzSy5nGMp973FzM42s1Iza1mmfktfeY394HSg16e890BfeXnnvML3XgDhizBcvhaSJsgbsjBY3h3xPjSz2IA6t8n7avpSScN8ZTMkDZH0kKRzJX0t6S0zO9G/kpndLelZeT14F0i6TtJ2SXUqaMsESadIGi6pn7xewAqvevSF2Y8ktZV0ta89LSXNN7OGZarfKilV0p8kjZf3dfvNFW37MEuTNE7SWElnyTuGNyR9Ja/Hc4y822sHBo7jJX0m73W5Vt5MI7MkHeOrEi8pUtJISf0l3SOpt6TXq/tgqpqZTZP0ku/pdt9//L2cc6WHuN1oM/ub70NXoS+AzDKzmIA6BwqGjc3sBTPb4ls+z7xe7MD9rDKzR83sHjNbJynXVx5hZneYN51ioZn9aGaXH8oxVUILeX83V8n7dztT0vNmdomvTUmSvpD0G0l/8dX5u3x/V2Zmkv4t79/yU/L+Xu+Vd1vgg/GKvNlxzvL9lKQkee8rZ0u6Rd6c6nPN+9ZEvnbs773lA0lZksqeyyGSNkl67yDbejjs9/WprIN47wUQLpxzPAIekqbJC5snB5S1kFQs6Vrfcyfpv2XW6+Mr71mm/BNJr/t+byApT948yvvb/8KA50sl3bSf+qMlbQ54fq2vrccGlDWTtFvSnQFlTtInZbb1pqQva9BrcKLvuUlaLen5MvX+LO8GLY18z1+RN2QkrpL7iZL0O9++mpc5NzcGPJ8naUaoz0uZtqfL+7DmJJ0mqbukegHLz/EtSwtyu6MkZcsLTT0k/cH3esT5lifJC1Ur5AWpPvI+QP01YBufSlov6Qp54eUTSTsktQqos8q3nznyPtxc6Ct/StJOSbdL6ivpEUklks45TOfVfH8Xz0qa6ysbK2mXpJQK1unnO9fn7We7e/1N+crK/tsd4qt38wHaGCmpqa9uD19ZAx34veUBSSv16yxC5nsd/hbqv+dDfH32Oo/lnfPKnB8ePHiE74Mxw+Xb6Jz73P/EObfazBbJuyjpGV/xu2XW6SsvAHxmZoHn9SN5/8lJ3lzKcZKeD6It/5N0m5mVSJrjnPvxAPUzJC12zv0S0P51ZvaZvB7mQP8p8/x7Sd1UM2Q65/7n+721pOaSXitzbudKipXUQV5vT29J/3T7Gb9tZpdKGiHpOHm3/fZrLWlNlbW+mjnnfjYz//jyr51zO6to0xmSpjvnXggoey3g9+Hyvinp6pzL9pV95F9oZmfK+4DRyzk331c2V17ouk3etw+BznG+MbHmDeW5TtIVAfufY2Yp8npa31E1MLMEed80nC8vZPp7WzN9P3tL+iDgeMvqLWmrc+6tKmpS2fcWmZn/m4z2kgKH9bSW92GjMu8t/5D3zVIveWPKT5P3QT+Y96PDrhKvT2UczHsvgDDBMInybaygLCXg+YYyyxtLSpZUVOYxWr9+Td/I97Oi/1TLc6O8HttRkpab2U9mdvF+6qeU0zZ/e8sOk9hW5vlueeGyJgg8Bv/Xze9p73O70lceeH4rPLdmNkDejAtfSPq9vN7UAb7FNeW4Q+1/koaY2e3mjb+2MssPFAwzJG3yB2FJcs7tkhdky34Y+8jtfXFYH0mlkmaZWZT/IS9snxg4JKCKTZN3G/jxks6QNxziH/r1b2K/f1eVWB6svf79mtlvJL0l71uPS+UFu+6+xYFt1P7a4fuAPE9ej718P79yzn1XJa2uPtO0/9enMg7mvRdAmKBnuHxJFZQF/qdRdtzuVnk9FRfsZ7tbfD9TJFVqbmDn3DZ5Y5KHmdkJ8r4+ftnMvnHOfV/OKtnyeo/KauJr45Ei8Pz62z1U0n/LqesPxVu09weWsn4vaYFz7np/gZn1PJRGHoUekBdIr5c3RCHTzMY75x7zLW8kbyx8RYL5MFbeB8pIeeM4K9p2lc6c4rsO4Gx5X6c/E1Ae2FFwoL+rAy2XvNvOx5QpK3s+/Mq+twyQN673Iuec87WvRTltkA783jJV0hQzu1PShfKuG6ixKvn6FKjMufX1JgcK+r0XQPigZ7h8SWZ2sv+JmTWX1EXexVsV+Uhez/BO59zCsg9fnS/kjXE9qAuCnHPfyPuqOUK/ziRQ1gJJXQOvGjezppJOljeW80i0XN4HjbTyzq1zzv8f3UeS/lDmQsdAcfJCSaA/llcxXDnnCpxzo5xzafK+gv+XpEm+4Q/SgYNftsr/MFneh7HyPlAWS/qtvN6/so/yvrE5VLXkBfA9fxdmVld7T0v3kaR+Ztakgm18JKmhmZ2zn/2sk3dRq38fEfJ62SsjTlKRPwj7lP27rex7yxvyvgF6Vd77yKuVbEOoVOb1WSepru99zu+MMts5pPdeAEc3eobLt1nSS2Z2j7w30Pvk/Uc8bT/rzJb0oaTZZvaIvF7kepJOlBTrnLvTObfNzO6X9KDv6vz35L3Zny1pjHNunzFwZvapvBkRlsoLD1fLu5inomA+TdJfJb1v3g0ZSuS7wETeRSdHHOdcqZndKu81qSfpfXn/oR8rryd+kHMuT964wq8lfWJmj8oLbp0lbXHO/UPea/SUmY2U96HhLHlfzaMczrmfzOwvkm6Q1E7ejAQfyfuWoolzrrwe4AWSxphZD+fcJ5JkZvHy/sZnHWCXc+UFn/rOudlVdRz745zbbmZfSxplZrnyesXvkNc77R+bO1HSZZL+z8welLRWXrCt7Zwbp1//7U83s/skLZb3gaGHc84/RnqWpBvM7L+SfpE3M0Jlp/SbLekWM5sk7wY7J8ubASbwOCr13uKcKzCzl+W9pq/4vnmqsSr5+nwg7336H75/9y3lXUgcuJ2g33sBhA/CcPlWy5vG6GF5F5gslHSJ28/k9845Z2YXyrtA5RZ5F3xtlTcG84mAemPNbKu8K/CvkZSjX6+2L88X8i7AS5MXbP8rqb/zbrZRXjsKzayvvCnZ/i7v6ut58q7WP5KGSezFOfcv33+Gd8mbRaJEXqh4R14wlnNuuZmdIu91m+pb9XvfOpL3YeBYeec+Vl7IGCzpy8N0GNXON+wjUVJXX1F/8+bC/r6CYTVl158laZG8v7N8SYPkvU984quy32DonPvQd7Hmv8zsDnkfSP4ir3dz/P727Xv9npH0qpmNk/fvLlbesJ/WzrmrKnsegjRY0nPyxpNvkfSkvGn4bvS1a5OZ/U7eVH+T5IWon+Sb1s/3b3+AvNk9bpF3/rMkTQ/Yxxh5PeYPyPt7fVLeB9y95sEtj3PuPTP7q6Sb5H0Y/kLebCE/lqlX2feWN+WF4X8caN81xIFen81mNlDS3+Qd2yLfOnv9vR/Eey+AMOGfYgc+5s3f2sE5V1NmVQDKZWZD5F0dX9c/m4SZzZN3O+ayxjjnRldim7fJu1jpOHlfo38v6SHn3L8D6rSQFwzPUEAwdM696lueKOlRedOqxcr7FuMvzrmvA7axSt50dX8ps3+TF1aultRK3vzD30v6u3PuxQO1Hwfm+6BxkbxbdR/SvNQAcDQgDJdBGAZwNDKzNvKGu7wo78PR30LcJACoERgmAQDh4Vl5Fye+JenxELcFAGoMeoaBMOEbgrC/uXpL+docABBumFoNCB89te9NYQIfo0LXNAAAQoOeYSBM+OZnbbOfKlnOuazD1R4AAGoCwjAAAADCFsMkAAAAELYIwwAAAAhbhGEA+zCzXmbmzKzDAepNM7OFAc8zzGx0NbUpxsxGm9mJ1bF9AEB4IgwDOBT3y7tduF+GpHuraV8xvm2fWE3bBwCEIW66AYQZM4t1zhVUxbaccz9XxXZCwcwiJUU653aHui0AgNChZxg4gpnZSWb2lpllmdkuM/ufmf0xYPkQ33CHDDObZ2b5km7zLTvBzN42s21mttPMvjKz08vsorGZve5b/ouZXV9m/3uGSZjZEElP+H53vse8gLodzOxdM9vhe7xuZsllttfIzJ41s2wzKzCz5WZ2i2/xDt/P5wO2n1bRkA7f8c4o21Yzu8DMvpNUIO+ObDKz833LCsxsvZmNM7PogHWbmdlrZrbRzPLN7Gczu79yrxIAoCajZxg4srWQ9JmkZ+SFu9/JC4ulzrlXAuq9ImmypDGStpnZ8b71lku6VtIWSd0kHVNm+1MkvSDpOUmXSHrKzBY6574qpy3vSnpU0q2STvKV5UqSmbXy7W+hpEvl3Qnvfklvm1mGc86ZWZykeZKSfO1cJqmV7yFJvSXNlfSAb1+SlC0prRLnyS9N0jhJ90naIGmlmf3Bd36elXSXpHRJY+V1FvzFt96LkuIkDZW0TdKxko4PYr8AgBqKMAwcwZxzr/p/991u+RNJzSRdLS/g+T3unHssoO4rkrZLOtU5l+8rnl3OLl5xzj3gW2eepHMlXShpnzDsnNtkZqt8v39ZZvG9ktZL6u8flmBm38gLvGfJC7eXSWovqYtz7n++9eYGbONr38+fA7fvHXalNZLU17993zkbL+lF59yeXm8zK5QX/Mc657bIGwt9iXPubV+VecHsFABQczFMAjiCmVmCmT1uZqv1622Vh0pqXabqu2We95b0r4AgXJH/+H9xzhVJ+kle2A5WX0mzJJWaWZSZRUlaKWmVvB5pf5v+GxCEq0Nmme23ltRc0mv+dvnaNldSrCT/0Iv/SRrrG3bSvBrbBwA4zAjDwJFtmqSL5PVuniHpN5L+IS/IBdpQ5nkjeUMMDmRbmee7y9l2ZTSW9Ff9Gtj9j2P169CMyrbpUJQ9D419P98r066VvnJ/2y6SN8RjoqTVvrHZfaq5rQCAw4BhEsARysxiJZ0t6Ubn3DMB5eV9yC173/UtklKqsXllbZXXMzy1nGWbfT+36NfxwcHwz4wRU6a8YcC2/cqeh62+n0Ml/becba+UJOdcpqQhvnObIWm0pLfMrLlvGAUA4AhFGAaOXLXkXYhW6C8ws7qSztO+oa+sjyT9wcxGVtU0az7+8cBlp2/7SN6Qg0XOuYra9pGk35vZCc65byratvbtmV7n+9lW0mLf/o+R1EbSjwdo73JJmZLSnHNTDlBXzrlSSV+a2RhJn8u7gJEwDABHMMIwcIRyzm03s68ljTKzXEmlku6Qd2FcvQOsPkbeBWmfmNmj8gJdZ0lbnHP/OIRmLfP9vNnM5krKdc4tl9eT+pWkd83sH/J6bJtKOl3SNOfcPHkzNtwg6T++u9gtl9RSUmvn3B3Oud1mtlJeiF8qr0f4G+fcOt95uN/M8uQN/7pLv/b6Vsg5V2pmt0p6yczqSXpfXug+VtIFkgZJipb0oa99P8r7EHKrvAsCfzjYEwUAqBkYMwwc2QbL+yr/RUmPSZrp+32/fAH1FHmhdKq8IQyDJK0+xPb8n7zxyzdLWiBvujI5536U1F1Snrxp2t6XF8gLJa3w1SmQdxHd2/KmPntf0u2SsgK2f628cb5z5IX5VF/5YElrJP1T0kO+9ZdXpsHOuX9JOl/ene1el/SGpOvl9TLvlhe6v/Ud01vypprLk3RGJS5ABADUcFbxN5YAAADA0Y2eYQAAAIQtwjAAAADCFmEYAAAAYYswDAAAgLBFGAYAAEDYIgwDAAAgbBGGAQAAELYIwwAAAAhb/w+5Y/JR3rV9mgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x444.96 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x444.96 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "start = time.perf_counter()\n",
    "file_months = 'predict_month_1'\n",
    "data_label = 'train'\n",
    "x_train = np.load('/Users/yunxiazhao/Documents/AAA_fire_pred/data/origin_2/' + file_months + '/data_' + data_label + '_x.npy')\n",
    "y_train = np.load('/Users/yunxiazhao/Documents/AAA_fire_pred/data/origin_2/' + file_months + '/data_' + data_label + '_y.npy')\n",
    "y_train = np.ravel(y_train)\n",
    "data_label = 'val'\n",
    "x_test = np.load('/Users/yunxiazhao/Documents/AAA_fire_pred/data/origin_2/' + file_months + '/data_' + data_label + '_x.npy')\n",
    "y_test = np.load('/Users/yunxiazhao/Documents/AAA_fire_pred/data/origin_2/' + file_months + '/data_' + data_label + '_y.npy')\n",
    "y_test = np.ravel(y_test)\n",
    "used_fields = ['Precip', 'LAI', 'ET', 'PET', 'S_rad', 'L_rad', 'LE', 'SH', 'Wind', 'Sf_SM', 'Sub_SM', 'P_SM', 'POP', 'Ele', 'LST'] \n",
    "MT_model_pool = ['logistic', 'svm', 'tree', 'XGboost', 'network', 'randomforest']\n",
    "\n",
    "print_opt = 1\n",
    "ML_Performance(x_train, x_test, y_train, y_test, MT_model_pool, print_opt)\n",
    "print_opt = 0\n",
    "feature_importance = calImportance(x_train, x_test, y_train, y_test, MT_model_pool, used_fields, print_opt)\n",
    "\n",
    "end = time.perf_counter()\n",
    "print((start-end)/60)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}