decision_tree.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "df545e50",
   "metadata": {},
   "source": [
    "### Decision Tree Classification Model\n",
    "\n",
    "A decision tree model is a machine learning algorithm that can be used for classification tasks.\n",
    "<br>This classification model is easy to interpret and can handle both categorical and numerical features, thus explaining our choice for this algorithm. \n",
    "<br>The model represents a series of decisions that lead to a final classification decision as a tree-like graph with several nodes. \n",
    "<br>At each node of the tree, a decision is made based on a specific attribute, and each branch represents a different decision or outcome. \n",
    "<br>\n",
    "<br>In classification, the goal is to predict a categorical variable based on a set of input features.\n",
    "<br>For our case - predicting the cost of damage for a fire incident - the set of input features is as follows:\n",
    "<li><b>DateOfCall</b>: the month of the date when the fire incident was reported \n",
    "<li><b>PropertyType</b>: the type of location where the fire incident occured\n",
    "<li><b>NumPumpsAttending</b>: the number of total fire pumps that were deployed to the fire incident location\n",
    "<li><b>PumpHoursRoundUp</b>: the number of hours the fire pumps were used during the fire incident\n",
    "<li><b>mean_temp</b>: mean daily temperature in Celsius (Cº)\n",
    "<br>\n",
    "<br>\n",
    "The output of our classification model is the cost of damage in pound sterling (£).\n",
    "<br>The cost value was originally a continuous numerical variable, but we convert it to a categorical variable by dividing and categorizing the numerical value in intervals of £300. \n",
    "<br>For example, all records of cost between £0.00 and £300.00 fall under category 1, all records of cost between £300.01 and £500.00 fall under category 2, and so on. \n",
    "<br>All records with costs larger than £1100.00 are categorized as category 5. \n",
    "<br>\n",
    "\n",
    "#### Scikit-Learn Library\n",
    "\n",
    "Our prediction model using the decision tree classifier is implemented with <b><i>scikit-learn</i></b> machine learning library in Python.<br>\n",
    "Please follow the <b>scikit-learn</b>'s installation guide ([https://scikit-learn.org/stable/install.html](Hidden_landing_URL)) and have the library ready before running the code.<br>\n",
    "\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "5e49967b",
   "metadata": {},
   "source": [
    "#### 1. Importing Libraries\n",
    "\n",
    "Following libraries and function are necessary to implement the decision tree prediction model.\n",
    "<br>\n",
    "<br>\n",
    "<b> Pandas</b>: \n",
    "<li> Data manipulation library\n",
    "<br><br>\n",
    "<b> Numpy</b>: \n",
    "<li> Data manipulation library\n",
    "<br><br>\n",
    "<b> train_test_split </b> from sklearn.model_selection:\n",
    "<li> Dividing the data into training and testing sets for model training and peformance analysis\n",
    "<br><br>\n",
    "<b> tree </b> from sklearn:\n",
    "<li> Scikit-learn's decision tree classifier model library\n",
    "<br><br>\n",
    "<b> classification_report</b> from sklearn.metrics:\n",
    "<li> Visualizing and measure the performance of the prediction model\n",
    "<br><br>\n",
    "<b> GridSearchCV</b> from sklearn.model_selection:\n",
    "<li> Hyper parameter tuning\n",
    "<br><br>\n",
    "<b>pickle</b>:\n",
    "<li> Saving and loading machine learning model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c07b9daf",
   "metadata": {},
   "outputs": [],
   "source": [
    "#import all necessary libraries for decision tree model training and testing\n",
    "\n",
    "#libraries for the data manipulation\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "#library for decision tree model\n",
    "from sklearn import tree\n",
    "\n",
    "#library for measuring the performance metrics\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "#library for hyper parameter tuning\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "#library for saving the final classifier model\n",
    "import pickle\n",
    "\n",
    "# import library from sklearn\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "\n",
    "#visualization report\n",
    "import seaborn as sns\n",
    "from matplotlib import pyplot as plt"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "3d4e1fb7",
   "metadata": {},
   "source": [
    "#### 2. Load Dataset\n",
    "\n",
    "Loading the fire incident and weather dataset into <i>pandas</i> dataframe from CSV file.<br>\n",
    "The features present in the cleaned dataset are listed below, alongside the type of data each of them holds.\n",
    "<br>Only selected features will be used for the prediction model.\n",
    "<br>Please refer to below for each data types' equivalent in Python.\n",
    "<br><br>\n",
    "Pandas' datatypes and their Python equivalents:\n",
    "<li> int64 = int\n",
    "<li> float64 = float\n",
    "<li> object = string\n",
    "<br><br>*Please refer to the <b><i>preprocessing</i></b> folder for detailed implementation on data cleaning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "60aa0437",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "DateOfCall             int64\n",
       "CalYear                int64\n",
       "HourOfCall             int64\n",
       "IncidentGroup          int64\n",
       "PropertyCategory       int64\n",
       "PropertyType           int64\n",
       "NumPumpsAttending      int64\n",
       "PumpHoursRoundUp       int64\n",
       "Notional Cost (£)      int64\n",
       "Date                  object\n",
       "cloud_cover          float64\n",
       "sunshine             float64\n",
       "global_radiation     float64\n",
       "max_temp             float64\n",
       "mean_temp            float64\n",
       "min_temp             float64\n",
       "precipitation        float64\n",
       "pressure             float64\n",
       "snow_depth           float64\n",
       "CostCat                int64\n",
       "dtype: object"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load dataset, print types\n",
    "df = pd.read_csv('preprocessing/data/london_clean.csv')\n",
    "df.dtypes"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "9537f34b",
   "metadata": {},
   "source": [
    "#### 3. Feature Selection and Dataset Split\n",
    "\n",
    "First, the entire dataset is split into input and output features - X and y respectively.\n",
    "<br>We wish to predict feature in <i>y</i> based on features grouped in <i>X</i> used as input values.\n",
    "<br>\n",
    "<Br>\n",
    "Both groups of input and output features are split into two subsets for their respective use with the help of Scikit-Learn's <i>train_test_split</i> function:\n",
    "<li>67% of the dataset is used to train the decision tree model\n",
    "<li>33% of the dataset is used to test the performance of the decision tree model\n",
    "<br>\n",
    "<br>\n",
    "In <i>train_test_split</i> function, the <i>random_state</i> variable is specified to 42.\n",
    "<br>This specification allows <i>train_test_split</i> function to generate the identical training and testing subsets every time it is called.\n",
    "<br>This functionality allows all three prediction models to be trained on the same dataset, allowing better comparison between thier performance in the later step."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "2705cd45",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "         DateOfCall  PropertyType  NumPumpsAttending  PumpHoursRoundUp  \\\n",
      "706975            5            12                  1                 1   \n",
      "450954            9             6                  2                 1   \n",
      "525760            6            12                  2                 1   \n",
      "20577             3            83                  2                 1   \n",
      "1069644          11            37                  1                 1   \n",
      "\n",
      "         mean_temp  \n",
      "706975        15.0  \n",
      "450954        13.3  \n",
      "525760        16.4  \n",
      "20577          7.6  \n",
      "1069644        8.3  \n",
      "         CostCat\n",
      "706975         0\n",
      "450954         0\n",
      "525760         0\n",
      "20577          0\n",
      "1069644        1\n"
     ]
    }
   ],
   "source": [
    "# do train and test split\n",
    "X = df[['DateOfCall', 'PropertyType', 'NumPumpsAttending', 'PumpHoursRoundUp', 'mean_temp']]\n",
    "y = df[['CostCat']]\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)\n",
    "\n",
    "# print a small sample of train X and y\n",
    "print(X_train[0:5])\n",
    "print(y_train[0:5])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "54fe14f0",
   "metadata": {},
   "source": [
    "#### 4. Model implementation and training\n",
    "Scikit-Learn's deicision tree classifier offers a range of parameters that can be tuned to control the behaviour of the decision tree model.\n",
    "<br>\n",
    "The following are the parameters we are interested in:\n",
    "<ol>\n",
    "<li><b><i>max_depth</i></b>: \n",
    "<br> - Sets the maximum depth of the decision tree\n",
    "<br> - A deeper tree can capture more complex relationships in the data, but can also lead to overfitting\n",
    "<br> - Default value = \"None\"\n",
    "<li><b><i>min_samples_split</i></b>: \n",
    "<br> - Sets the minimum number of samples required to split an internal node. \n",
    "<br> - A higher value can prevent the tree from splitting too early, leading to more robust models.\n",
    "<br> - Default value = 2\n",
    "<li><b><i>criterion</i></b>: \n",
    "<br> - Specifies the function used to measure the quality of a split. \n",
    "<br> - The two options are \"gini\" and \"entropy\", which correspond to the Gini impurity and information gain criteria, respectively.\n",
    "<br> - Default value = \"gini\"\n",
    "</ol>\n",
    "\n",
    "\n",
    "The first training and testing of the decision tree model is done with default parameters set by the Scikit-Learn library.\n",
    "<br>\n",
    "These hyper parameters will be specified in the following step, once the decision tree model provides an acceptable performance score with the current training and testing datasets.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "7abac23e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# create and train decision tree\n",
    "clf = tree.DecisionTreeClassifier()\n",
    "clf = clf.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "09e9cec6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Default Decision Tree Model Description\n",
      "Depth: 52\n",
      "Number of Leaves: 86704\n",
      "*******************************************\n",
      "Classification Report of the default model\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.78      0.85      0.81    231390\n",
      "           1       0.71      0.61      0.66    140418\n",
      "           2       1.00      1.00      1.00     34444\n",
      "           3       0.72      0.75      0.73      5421\n",
      "           4       0.58      0.56      0.57      4962\n",
      "           5       0.92      0.92      0.92      7949\n",
      "\n",
      "    accuracy                           0.78    424584\n",
      "   macro avg       0.79      0.78      0.78    424584\n",
      "weighted avg       0.77      0.78      0.77    424584\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmUAAAI3CAYAAADaySzbAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy89olMNAAAACXBIWXMAAA9hAAAPYQGoP6dpAACMGklEQVR4nOzdd1gUVxcG8HdBWJCOSFMURBFRBEVF7IWAvURjYm+xBSzYIsaCRsXYNRYssURj7GKJDStRsYu9N2ygSBOUpex8f/gxcaWqwI7y/p5nH9k7Z+7cOc7CYebOIBMEQQARERERqZWGugdARERERCzKiIiIiCSBRRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAHF1D0AIiIiUh/t4a6Ftq2UOeGFtq0vEc+UEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCjLZw8fPoRMJsPq1as/ar1GjRqhUaNGBTImyl5UVBQ6duyIEiVKQCaTYd68eQWynV69esHW1rZA+qavi0wmQ0BAwEev96nfe3Kj7u9NAQEBkMlkKm1paWkYPXo0bGxsoKGhgXbt2gH49Nx9Ln6+Kb98dUXZ6tWrIZPJxJeOjg6sra3h7e2NBQsW4PXr1+oeomTY2tqq5Cq7V35/k/8YR48eVRmLXC6HhYUFGjVqhGnTpuHly5ef1b+fnx/2798Pf39/rF27Fs2aNcunkefszZs3CAgIwNGjR/MU/2EetLS0UK5cOfTo0QP3798v2MGq2fXr1xEQEICHDx8W2jbf/z5y/PjxTMsFQYCNjQ1kMhlatWpVaOPKT1FRURg5ciQcHR1RvHhx6Onpwc3NDVOmTEFcXJy6h5ejlStXYubMmejYsSPWrFkDPz+/At/ms2fPEBAQgPDw8ALfFhVdxdQ9gIIyefJk2NnZITU1FZGRkTh69CiGDRuGOXPmYOfOnahatWqBbLds2bJ4+/YttLS0Pmq9AwcOFMh4cjJv3jwkJiaK7/fs2YO///4bc+fOhZmZmdhep06dQh/bh4YMGYKaNWsiPT0dL1++xMmTJzFx4kTMmTMHmzZtQpMmTT6p38OHD6Nt27YYOXJkPo84Z2/evMGkSZMA4KPOQmTkITU1FRcuXMCyZcvwzz//4MqVK7C2ti6g0arX9evXMWnSJDRq1KjQz0bo6Ohg/fr1qFevnkr7sWPH8OTJE8jl8kIdT345e/YsWrRogcTERHTr1g1ubm4AgHPnzmH69OkIDQ1Vy/ekrIwbNw5jxoxRaTt8+DBKlSqFuXPnqrS/ffsWxYoVzI+1Z8+eYdKkSbC1tYWrq6vKsuXLl0OpVBbIdqlo+WqLsubNm6NGjRrie39/fxw+fBitWrVCmzZtcOPGDejq6ub7djPOzn0sbW3tfB9LbjJO+WeIjIzE33//jXbt2uX4wy8pKQl6enoFO7gP1K9fHx07dlRpu3TpEry8vNChQwdcv34dVlZWH93vixcvYGxsnE+jLHjv56F3795wcHDAkCFDsGbNGvj7+39yv4IgIDk5uUA+E1KVl+O4RYsW2Lx5MxYsWKDyw379+vVwc3NDdHR0QQ8z38XFxaF9+/bQ1NTExYsX4ejoqLJ86tSpWL58uZpGl1mxYsUyFVrZfW4/5XtvfvjYX8KJsvPVXb7MSZMmTTB+/Hg8evQI69atU1l28+ZNdOzYEaamptDR0UGNGjWwc+fOTH3ExcXBz88Ptra2kMvlKF26NHr06CF+c85qXkdkZCR69+6N0qVLQy6Xw8rKCm3btlW5HJPVvI0XL16gb9++sLCwgI6ODlxcXLBmzRqVmIztzZo1C8uWLYO9vT3kcjlq1qyJs2fPfl7C8G6uhL6+Pu7du4cWLVrAwMAAXbt2BQAolUrMmzcPlStXho6ODiwsLDBgwADExsZm6mfv3r2oX78+9PT0YGBggJYtW+LatWufNTYXFxfMmzcPcXFxWLhwocqyp0+fok+fPrCwsIBcLkflypWxcuVKcXnG5SlBELBo0SLxUhUAxMTEYOTIkXB2doa+vj4MDQ3RvHlzXLp0SWUbGX18eFkt41JjdpcmHz58iJIlSwIAJk2aJG77U+bCZJwhfPDgAQBg1apVaNKkCczNzSGXy+Hk5IQlS5ZkWs/W1hatWrXC/v37UaNGDejq6mLp0qWf1MfRo0fFPpydncX93rZtG5ydnaGjowM3NzdcvHgxUx+5fe5Wr16N7777DgDQuHFjMVfv5zYvx1ZOx3FOOnfujFevXiEkJERsS0lJwZYtW9ClS5cs10lKSsKIESNgY2MDuVyOihUrYtasWRAEQSVOoVDAz88PJUuWhIGBAdq0aYMnT55k2Wdux/PHWLp0KZ4+fYo5c+ZkKsgAwMLCAuPGjct2/ZSUFEyYMAFubm4wMjKCnp4e6tevjyNHjmSK3bBhA9zc3GBgYABDQ0M4Oztj/vz54vLU1FRMmjQJFSpUgI6ODkqUKIF69eqp5Pv9OWUZ3++OHDmCa9euZToesvocPX36FH379oW1tTXkcjns7OwwaNAgpKSkAMjb5/3o0aOoWbMmgHe/DH04tSOrOWV5PQ5kMhl8fX0RHByMKlWqiP+/+/bty/b/gL5eX+2Zsux0794dY8eOxYEDB9CvXz8AwLVr11C3bl2UKlUKY8aMgZ6eHjZt2oR27dph69ataN++PQAgMTER9evXx40bN9CnTx9Ur14d0dHR2LlzJ548eaJyye99HTp0wLVr1zB48GDY2trixYsXCAkJQURERLZnpN6+fYtGjRrh7t278PX1hZ2dHTZv3oxevXohLi4OQ4cOVYlfv349Xr9+jQEDBkAmk2HGjBn49ttvcf/+/c/+LS4tLQ3e3t6oV68eZs2aheLFiwMABgwYgNWrV6N3794YMmQIHjx4gIULF+LixYs4ceKEuN21a9eiZ8+e8Pb2xm+//YY3b95gyZIlqFevHi5evPhZl6Q6duyIvn374sCBA5g6dSqAd3NlateuLX6zK1myJPbu3Yu+ffsiISEBw4YNQ4MGDbB27Vp0794d33zzDXr06CH2ef/+fQQHB+O7776DnZ0doqKisHTpUjRs2BDXr1//7MuEJUuWxJIlSzBo0CC0b98e3377LQB80iX1e/fuAQBKlCgBAFiyZAkqV66MNm3aoFixYti1axd++uknKJVK+Pj4qKx769YtdO7cGQMGDEC/fv1QsWLFj+7j7t276NKlCwYMGIBu3bph1qxZaN26NYKCgjB27Fj89NNPAIDAwEB06tQJt27dgobGu98F8/K5a9CgAYYMGYIFCxZg7NixqFSpEgCI/37MsZXdcZwTW1tbeHh44O+//0bz5s0BvCsC4+Pj8cMPP2DBggUq8YIgoE2bNjhy5Aj69u0LV1dX7N+/H6NGjcLTp09VLrf9+OOPWLduHbp06YI6derg8OHDaNmyZaYx5OV4/hg7d+6Erq5upjPPeZWQkIAVK1agc+fO6NevH16/fo0//vgD3t7eOHPmjHhpLyQkBJ07d0bTpk3x22+/AQBu3LiBEydOiN+/AgICEBgYiB9//BG1atVCQkICzp07hwsXLuCbb77JtO2SJUti7dq1mDp1KhITExEYGAjgv+PhQ8+ePUOtWrUQFxeH/v37w9HREU+fPsWWLVvw5s0baGtr5+nzXqlSJUyePBkTJkxA//79Ub9+fQDZT+34mOMAAI4fP45t27bhp59+goGBARYsWIAOHTogIiJC/GxTESF8ZVatWiUAEM6ePZttjJGRkVCtWjXxfdOmTQVnZ2chOTlZbFMqlUKdOnWEChUqiG0TJkwQAAjbtm3L1KdSqRQEQRAePHggABBWrVolCIIgxMbGCgCEmTNn5jjuhg0bCg0bNhTfz5s3TwAgrFu3TmxLSUkRPDw8BH19fSEhIUFleyVKlBBiYmLE2B07dggAhF27duW43ffNnDlTACA8ePBAbOvZs6cAQBgzZoxK7L///isAEP766y+V9n379qm0v379WjA2Nhb69eunEhcZGSkYGRllav/QkSNHBADC5s2bs41xcXERTExMxPd9+/YVrKyshOjoaJW4H374QTAyMhLevHkjtgEQfHx8VOKSk5OF9PR0lbYHDx4IcrlcmDx5stiWcay9n6/3x3zkyBGxrWfPnkLZsmXF9y9fvhQACBMnTsx2v7Lqc+XKlcLLly+FZ8+eCf/8849ga2sryGQy8Xh/f98yeHt7C+XKlVNpK1u2rABA2LdvX6b4j+3j5MmTYtv+/fsFAIKurq7w6NEjsX3p0qWZcpLXz93mzZszrSsIH3dsZXccZ+f97yMLFy4UDAwMxLx89913QuPGjcUctGzZUlwvODhYACBMmTJFpb+OHTsKMplMuHv3riAIghAeHi4AEH766SeVuC5dumQ6LvJ6PH/4vSc7JiYmgouLS57yIAiZvzelpaUJCoVCJSY2NlawsLAQ+vTpI7YNHTpUMDQ0FNLS0rLt28XFRSV/WZk4caLw4Y+qhg0bCpUrV84U+2HuevToIWhoaGT58yDje3ZeP+9nz57NNr8ffr7zehxkjFlbW1ul7dKlSwIA4ffff8+0rYKg5edSaC/KWZG6fJlBX19fvAszJiYGhw8fRqdOnfD69WtER0cjOjoar169gre3N+7cuYOnT58CALZu3QoXFxfxzNn7PrxlO4Ouri60tbVx9OjRLC/rZWfPnj2wtLRE586dxTYtLS0MGTIEiYmJOHbsmEr8999/DxMTE/F9xm9y+XVn3qBBg1Teb968GUZGRvjmm2/EnEVHR8PNzQ36+vripYyQkBDExcWhc+fOKnGamppwd3fP8pLHx3r//1MQBGzduhWtW7eGIAgq2/T29kZ8fDwuXLiQY39yuVw8m5Oeno5Xr15BX18fFStWzHXdgtanTx+ULFkS1tbWaNmyJZKSkrBmzRpx/uT7c8Li4+MRHR2Nhg0b4v79+4iPj1fpy87ODt7e3pm28TF9ODk5wcPDQ3zv7u4O4N1l1TJlymRqzzgeP+Zzl51PObY+PI7zolOnTnj79i12796N169fY/fu3dleutyzZw80NTUxZMgQlfYRI0ZAEATs3btXjAOQKe7Ds175cTx/KCEhAQYGBh+1zvs0NTXFObBKpRIxMTFIS0tDjRo1VMZibGyMpKQklUuRHzI2Nsa1a9dw586dTx5PdpRKJYKDg9G6dWuV+cUZMr5nF8TnPa/HQQZPT0/Y29uL76tWrQpDQ8Ov/s5qyqzIXb4E3l2GNDc3B/Du8osgCBg/fjzGjx+fZfyLFy9QqlQp3Lt3Dx06dPiobcnlcvz2228YMWIELCwsULt2bbRq1Qo9evSApaVltus9evQIFSpUEL9ZZMg4Tf/o0SOV9vd/AAIQC7SPKQSzU6xYMZQuXVql7c6dO4iPjxfz+KEXL16IcQCyvTvS0NDws8eXmJgo/pB5+fIl4uLisGzZMixbtizHsWVHqVRi/vz5WLx4MR48eID09HRxmbovJUyYMAH169eHpqYmzMzMUKlSJZVJ0CdOnMDEiRMRFhaGN2/eqKwbHx8PIyMj8b2dnV2W2/iYPj487jKW2djYZNmecTx+zOcuOx97bGV1HOdFyZIl4enpifXr1+PNmzdIT0/P9tLfo0ePYG1tnano+fBz++jRI2hoaKj8IAYgXkLOkB/H84cMDQ0/+9FAa9aswezZs3Hz5k2kpqaK7e8fUz/99BM2bdqE5s2bo1SpUvDy8kKnTp1UHjszefJktG3bFg4ODqhSpQqaNWuG7t2758vd8S9fvkRCQgKqVKmSY1xBfN7zehxk+PBzBLz7Hp4f37/py1LkirInT54gPj4e5cuXBwDxNuaRI0dmedYAgBj7qYYNG4bWrVsjODgY+/fvx/jx4xEYGIjDhw+jWrVqn9V3Bk1NzSzbhQ8mlX6K93+TzKBUKmFubo6//vory3UyJrJn5Hft2rVZFqGfe/t6amoqbt++LX7jzdhet27d0LNnzyzXye0b/rRp0zB+/Hj06dMHv/76K0xNTaGhoYFhw4ap3Pae3dnR97+p5zdnZ2d4enpmuezevXto2rQpHB0dMWfOHNjY2EBbWxt79uzB3LlzM92yn9Wdlh/bR3bHXW7HY3587j722MrqOM6rLl26oF+/foiMjETz5s0L7Y7d/DieP+To6Ijw8HCkpKR80l3f69atQ69evdCuXTuMGjUK5ubm0NTURGBgoDjHEQDMzc0RHh6O/fv3Y+/evdi7dy9WrVqFHj16iDcsNWjQAPfu3cOOHTtw4MABrFixAnPnzkVQUBB+/PHHjx7bp8jr570gFeT3b/qyFLmibO3atQAg/iAoV64cgHeXBrP7YZfB3t4eV69e/aTt2tvbY8SIERgxYgTu3LkDV1dXzJ49O9NdoBnKli2Ly5cvQ6lUqvwguXnzprhcnezt7XHw4EHUrVs3x8coZJwJMDc3zzW/n2LLli14+/at+P+ZcSdbenr6J29vy5YtaNy4Mf744w+V9ri4OJWbOTLORn74oM0PfwvOSnYF3efYtWsXFAoFdu7cqfKb98dcIs6PPvLiYz532eWqoI+t97Vv3x4DBgzAqVOnsHHjxmzjypYti4MHD+L169cqZ0k+/NyWLVsWSqUS9+7dUzk7duvWLZX+8uN4/lDr1q0RFhaGrVu3qkyPyKstW7agXLly2LZtm8r/zcSJEzPFamtro3Xr1mjdujWUSiV++uknLF26FOPHjxeLblNTU/Tu3Ru9e/dGYmIiGjRogICAgM8uykqWLAlDQ8Ncv2fn9fP+MZ/ZvB4HRB8qUnPKDh8+jF9//RV2dnbi7fDm5uZo1KgRli5diufPn2da5/0nxnfo0AGXLl3C9u3bM8Vl9xvNmzdvkJycrNJmb28PAwMDKBSKbMfaokULREZGqvwASEtLw++//w59fX00bNgw550tYJ06dUJ6ejp+/fXXTMvS0tLEQsXb2xuGhoaYNm2aymWODJ/zRP5Lly5h2LBhMDExEe8K1NTURIcOHbB169YsvxnnZXuampqZ/j83b96caY5TRlEQGhoqtqWnp2d7mel9GXf+5eeT0zN+235/7PHx8Vi1alWh9pEXH/O5y3iW2Ie5Kshj60P6+vpYsmQJAgIC0Lp162zjWrRogfT09EyPaJk7dy5kMpl4B2fGvx/evfnhn/nKj+P5QwMHDoSVlRVGjBiB27dvZ1r+4sULTJkyJdv1szpGTp8+jbCwMJW4V69eqbzX0NAQz+plfO/7MEZfXx/ly5fP8XtjXmX8+aVdu3bh3LlzmZZnjD+vn/fsjsOs5PU4IPrQV3umbO/evbh58ybS0tIQFRWFw4cPIyQkBGXLlsXOnTtVHjK4aNEi1KtXD87OzujXrx/KlSuHqKgohIWF4cmTJ+LzakaNGoUtW7bgu+++Q58+feDm5oaYmBjs3LkTQUFBcHFxyTSO27dvo2nTpujUqROcnJxQrFgxbN++HVFRUfjhhx+yHX///v2xdOlS9OrVC+fPn4etrS22bNmCEydOYN68eZ81UTc/NGzYEAMGDEBgYCDCw8Ph5eUFLS0t3LlzB5s3b8b8+fPRsWNHGBoaYsmSJejevTuqV6+OH374ASVLlkRERAT++ecf1K1bN9M3rqz8+++/SE5OFifinjhxAjt37oSRkRG2b9+ucvlq+vTpOHLkCNzd3dGvXz84OTkhJiYGFy5cwMGDBxETE5Pjtlq1aoXJkyejd+/eqFOnDq5cuYK//vpLPLuToXLlyqhduzb8/f0RExMDU1NTbNiwAWlpabnuj66uLpycnLBx40Y4ODjA1NQUVapUyXX+S068vLzEMxMDBgxAYmIili9fDnNz8ywLn4LqI6/y+rlzdXWFpqYmfvvtN8THx0Mul4vPUcuPYyuvsrt8+L7WrVujcePG+OWXX/Dw4UO4uLjgwIED2LFjB4YNGyYW8q6urujcuTMWL16M+Ph41KlTB4cOHcLdu3cz9fm5x/OHTExMsH37drRo0QKurq4qT/S/cOEC/v77b5WbNz7UqlUrbNu2De3bt0fLli3x4MEDBAUFwcnJSeUvhPz444+IiYlBkyZNULp0aTx69Ai///47XF1dxblVTk5OaNSoEdzc3GBqaopz585hy5Yt8PX1/ah9ys60adNw4MABNGzYEP3790elSpXw/PlzbN68GcePH4exsXGeP+/29vYwNjZGUFAQDAwMoKenB3d39yznZub1OCDKpLBv9yxoGbeyZ7y0tbUFS0tL4ZtvvhHmz58vPkriQ/fu3RN69OghWFpaClpaWkKpUqWEVq1aCVu2bFGJe/XqleDr6yuUKlVK0NbWFkqXLi307NlTvF39w9vSo6OjBR8fH8HR0VHQ09MTjIyMBHd3d2HTpk0q/X5427kgCEJUVJTQu3dvwczMTNDW1hacnZ0z3Y6dsb2sHrmBj3jkgiBk/0gMPT29bNdZtmyZ4ObmJujq6goGBgaCs7OzMHr0aOHZs2cqcUeOHBG8vb0FIyMjQUdHR7C3txd69eolnDt3LscxZTwKIuOlpaUllCxZUmjQoIEwdepU4cWLF1muFxUVJfj4+Ag2NjaClpaWYGlpKTRt2lRYtmyZShyyeSTGiBEjBCsrK0FXV1eoW7euEBYWluX/0b179wRPT09BLpcLFhYWwtixY4WQkJBcH4khCIJw8uRJwc3NTdDW1s71/yovjwYRBEHYuXOnULVqVUFHR0ewtbUVfvvtN2HlypWZ/l8/fJRDfvaRVU6zO07z+rlbvny5UK5cOUFTUzNTbvNybOV2HH8oL4/WEYSsc/D69WvBz89PsLa2FrS0tIQKFSoIM2fOFB/BkOHt27fCkCFDhBIlSgh6enpC69athcePH2d5LOTleM7rIzEyPHv2TPDz8xMcHBwEHR0doXjx4oKbm5swdepUIT4+Xoz78LhXKpXCtGnThLJlywpyuVyoVq2asHv37kzH+JYtWwQvLy/B3Nxc0NbWFsqUKSMMGDBAeP78uRgzZcoUoVatWoKxsbGgq6srODo6ClOnThVSUlLEmM95JIYgCMKjR4+EHj16CCVLlhTkcrlQrlw5wcfHR3ysx8d83nfs2CE4OTkJxYoVU8l1Vp/vvB4HWX1eBOHdsdWzZ89M7QWBj8SQDpkgcCYhERFRUaU93LXQtpUyJ7zQtvUlKlJzyoiIiIikikUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERJIUGhqK1q1bw9raGjKZDMHBwSrLZTJZlq+ZM2eKMba2tpmWT58+XaWfy5cvo379+tDR0YGNjQ1mzJiRaSybN2+Go6MjdHR04OzsjD179qgsFwQBEyZMgJWVFXR1deHp6Yk7d+581P6yKCMiIiJJSkpKgouLCxYtWpTl8ufPn6u8Vq5cCZlMhg4dOqjETZ48WSVu8ODB4rKEhAR4eXmhbNmyOH/+PGbOnImAgACVv2N88uRJdO7cGX379sXFixfRrl07tGvXTuVv0s6YMQMLFixAUFAQTp8+DT09PXh7e2f6+9c5+aqf6F+YTyn+krz/ROVDT/dkH1hENS3VQvw6IvGeGkciTWX0//u7fcxP1pijnDE/OXs/P4XhS3miv0wmw/bt29GuXbtsY9q1a4fXr1/j0KFDYputrS2GDRuGYcOGZbnOkiVL8MsvvyAyMhLa2toAgDFjxiA4OBg3b94EAHz//fdISkrC7t27xfVq164NV1dXBAUFQRAEWFtbY8SIERg5ciQAID4+HhYWFli9enWOf+v6fTxTRkRERIVCoVAgISFB5aVQKPKl76ioKPzzzz/o27dvpmXTp09HiRIlUK1aNcycORNpaWnisrCwMDRo0EAsyADA29sbt27dQmxsrBjj6emp0qe3tzfCwsIAAA8ePEBkZKRKjJGREdzd3cWYvGBRRkRERIUiMDAQRkZGKq/AwMB86XvNmjUwMDDAt99+q9I+ZMgQbNiwAUeOHMGAAQMwbdo0jB49WlweGRkJCwsLlXUy3kdGRuYY8/7y99fLKiYviuU5koiIiOgz+Pv7Y/jw4Sptcrk8X/peuXIlunbtCh0dHZX297dXtWpVaGtrY8CAAQgMDMy3becXnikjIiKiQiGXy2FoaKjyyo/C6N9//8WtW7fw448/5hrr7u6OtLQ0PHz4EABgaWmJqKgolZiM95aWljnGvL/8/fWyiskLFmVERET0Rfvjjz/g5uYGFxeXXGPDw8OhoaEBc3NzAICHhwdCQ0ORmpoqxoSEhKBixYowMTERY96/eSAjxsPDAwBgZ2cHS0tLlZiEhAScPn1ajMkLXr4kIiIiSUpMTMTdu3fF9w8ePEB4eDhMTU1RpkwZAO+Kn82bN2P27NmZ1g8LC8Pp06fRuHFjGBgYICwsDH5+fujWrZtYcHXp0gWTJk1C37598fPPP+Pq1auYP38+5s6dK/YzdOhQNGzYELNnz0bLli2xYcMGnDt3Tnxshkwmw7BhwzBlyhRUqFABdnZ2GD9+PKytrXO8W/RDLMqIiIhIks6dO4fGjRuL7zPmh/Xs2ROrV68GAGzYsAGCIKBz586Z1pfL5diwYQMCAgKgUChgZ2cHPz8/lXlmRkZGOHDgAHx8fODm5gYzMzNMmDAB/fv3F2Pq1KmD9evXY9y4cRg7diwqVKiA4OBgVKlSRYwZPXo0kpKS0L9/f8TFxaFevXrYt29fpjluOeFzyoogPqcsZ3xOWc74jKncMUc5Y35yxueUFV2cU0ZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikoBi6h6AFNQrVx0jGvdEtdKVYG1kjo4r/bDz6hFxubm+Kaa1GgbPirVhrGuAf+9fgN+233A3OkKMCflpBRqWr6HS77KTm+G7Zar4PmVOeKZtd/vzZ2wK3y++H1j3e/xU73uUNbXG49hITD+4AuvO7RaXF9Mohp89+6BbjdYoZWSO2y8fYuzu+Thw82R+pCJPQnecQOiuE4iJjAEAWNlaokV3b1R2rwQAmOu3EHcu3VNZp15rD3Tx6wQACNt3Bmtn/J1l379tnQwDEwMAwLHg4zga/C9iImNhYm6MZt2+QW2vmmLsswfPsXv1PkTcfoyYqFh0/KkdmnRsmO/7q247Nu3C5j+3IuZVLOwr2MFn9CA4Vqmo7mFJBvOTM+Ynd8wRSQWLMgB62rq4/Ow2Vp8JxubeczMt39JnLlLT09BhpR9eJydiaKPu2DswCC4zvsWblGQxbkXYVkzat1h8//6yDH3/noADN0+I7+Pevha/7l/nO0xpORiDNk3GuYhrqFmmCpZ0moDYNwn453ooAGByCx90dmuJQZsm41bUA3zjWAebe89BwwU9Ef70Vr7kIzfGJY3Q7sdWMC9dEoIg4NSBswga/wf8l46AtZ0VAKBuy9po1bu5uI62XFv82q2xK5xqOar0ufa3v5GakioWZKE7TmDHit3oMuJ72Fa0wcObEfhr9iYU19dF1TpVAAApilSYWZVA9YYu2LI4uID3Wj2OHjiGpXOWY8hYX1Sq4oht64Ph7zseK7ctg4mpsbqHp3bMT86Yn9wxRyQlvHwJYP/NE5i4dxF2XDmSaVmFkmVQ29YFg7dMw/nH13D75SP4bpkKXS0dfF+tuUrsm9RkRL1+Jb5eK5Iy9Rf/9rVKjCItRVzW1a0VlodtxebwA3gQ8xSbwvdjxamtGNmktxjTxa0lfjv4B/bdOI4HMU+x7ORm7LtxHMMa9cjHjOSsap0qqFLbCealS8LCxhxt+7aEXFeOBzceiTHacm0YmRqKL109nWyXaWho4NbFO6jT3F2MOR1yDvVa1UGNxtVgZm2GGk2qo14rD4RsOCzG2DqWwbcD26BGk+oopvV1/n6xdd12NG/fDM3aeKFsuTIYOtYXch059u84oO6hSQLzkzPmJ3fMEUmJWouy6OhozJgxA+3bt4eHhwc8PDzQvn17zJw5Ey9fvlTn0ETyYu/O8CSnKcQ2QRCgSEtBXbtqKrGdqzfHs8lHcHHUFkxpORi6Wjr40PwO/ng2+QhODFuHnrXafrAtLShSFSptyakK1CxTBcU0ionjeX8sAPA2VYE6H4ylsCjTlTh3+AJSkhUo52Qrtp89dB6j2o3Dr31+Q/Dy3UhJTsm2j9MHzkJbroVqDV3EtrTUNGhpqxZaWtpaeHgzAulp6fm+H1KUmpqK2zfvonotV7FNQ0MD1Wu54vqVm+obmEQwPzljfnLHHJHUqO30wtmzZ+Ht7Y3ixYvD09MTDg4OAICoqCgsWLAA06dPx/79+1GjRo0c+1EoFFAoVIsUuVwOuVyeL+O8GfUQj2KeYUrLIfhp869ISnmLoQ27wcbEEpaGZmLchgt7ERH7DM8TXsLZygFTWw2FQ0lbdFo9QowJ2LsIR+6cxdvUt/Cs6IHfO4yFvrw4Fv37bn5VyK0w9K7dHjuuHsHFJzdQvbQTeru3h3YxLZjpGSPydTRCboVhWMPuOH7vAu69eowmFdzRzrkJNDU082V/8+rp/WeY5TsfqSlpkOtqo/+kPrCytQQA1GxaHaYWpjAqYYin958jeNkuRD1+gQGT+2TZ18m9p1GjqZvKJU6nmo44secUXOo5w6ZCaUTcfoyTe04hPS0difGJMCphVCj7qU7xcQlQpithUsJEpd2khDEeP3ysplFJB/OTM+Ynd8wRSY3airLBgwfju+++Q1BQEGQymcoyQRAwcOBADB48GGFhYTn2ExgYiEmTJqm0TZw4EQEBAfkyzjRlGjqtHoFl3wfgxdR/kZaehkN3TmPvjeN4f9R/nNoqfn31+V08T3iJAz8tR7kSpXH/1RMAwLSQ5WJM+NNb0NPWxfBGPcWibGrIMlgYlMDxoX9CBhmiEmOw7twujGzSG0pBCQAYvn0GgjpNwJUx2yEIAu6/eoI1Z3ail7vqWbeCZmFjDv/lI5GclIwLxy7hz9/Ww2+uL6xsLVGvVR0xrlQ5axiZGmL+yMV4+TQaJUuZqfRz/9pDRD6KQi//rirtzbt/g4SYBMzwmQcIgIGJAdy9ayJkw2HIZLzqTkREXx+1FWWXLl3C6tWrMxVkACCTyeDn54dq1XK/JOfv74/hw4ertOXXWbIMF5/cQM3Z38NQRx/amlqITorF8aFrcf7x9WzXORNxBQBgb2YjFmWZYh5dxS9eA6CtqYWU9FQkpyrQf2MAfto8BRYGpnieEI0fPTogITkRL5NiAQDRSbHouMoP8mLaKKFnjGfxLzCt1VA8ePU0X/c5N8W0isG8VEkAQBkHGzy6FYEj20LRZXinTLG2lcoAAF4+y1yUndhzCqXLl0IZBxuVdm25NrqP7owuwzshIfY1jEwNcXx3GHSKy6FvrFdAeyUtRsaG0NDUQOyrWJX22FdxMDEzVdOopIP5yRnzkzvm6J2K9ja5B1GhUNspB0tLS5w5cybb5WfOnIGFhUWu/cjlchgaGqq88rsoy5CQnIjopFiUNysDNxsn7Lp6NNtYF+t3dxdGJkRnH1OqImLexCMlPVWlPU2ZhqfxL6AUlOhUzRt7rv8LQRBUYhRpKXgW/wLFNIqhXdWmOY6lMAhKAWmpaVkue3LvXcFoaGqo0p78VoELR8NVJvh/SLOYJkxKGkNDUwPnjlxEldqVoaFRNM6UaWlpwcGxPC6evSS2KZVKXDwbDidnxxzWLBqYn5wxP7ljjkhq1HambOTIkejfvz/Onz+Ppk2bigVYVFQUDh06hOXLl2PWrFmFMhY9bV2UNysjvrc1LQUX63cF0+O4SHRw+QYvE2PxOPY5qlhVwOz2o7Hz6hEcvP3u0mq5EqXxQ/Xm2HvjOGKS4uFsXQEz245E6L1zuPL8DgCgpVMDmBuUwJlHl5GcloKmDrXxc9O+mHv0T3G7FUqWQc0yVXDm0VUYFzfEsIbdUNmyPPqunyDG1CxTBaWMzHHp6S1YG5ljvPdAaMg0MOvw6kLJFQAEL9+NyrUqwdTCBMlvknH20AXcuXQPvr8NwMun0Th7+AKquFeCnqEent57hi2Lg1G+qj1K21ur9HP+yEUo05Wo9U3meYNRj1/g4c0I2FUqizev3+DQlmN4/vA5eo7pIsakpabh+aMoAEB6WjriouPx+O5TyHW1xbN4X7oO3dpjxsQ5cKhUARWrOGD7+h1IfquAd5tv1D00SWB+csb85I45IilRW1Hm4+MDMzMzzJ07F4sXL0Z6+rs76jQ1NeHm5obVq1ejU6fMl8IKgptNZRz0WSG+n9VuJADgzzM78eOGCbA0NMOMNiNgYVACzxNe4q9zuzE1ZJkYn5KeiiYO7hjcoCv0tHXxOC4KwZcPqcwhS1WmYVDd7zGr7UjIZDLci36MUTtn4Y9T28QYTZkmhjXsAQfzskhNT8Oxu+fQcEFPPIp9JsboaMkxqbkP7EqURqLiDfbdOI7e68chPvm/550VtNdxiVgz/S8kxCRAR08XpcpZwfe3AahUoyJiXsTi5vnbOLL1GBRvU2BibgzXBlXRvJtXpn5O7jkN1/rOKK6vm2mZUing0OajiHr8AprFNOHgWh4jFwxFCcv/LinEv0pAYP//CveDm47g4KYjqOBiD7+5vgWz84WskVdDxMUmYE3QWsS+ioW9QzlM+31yponJRRXzkzPmJ3fMEUmJTPjwupgapKamIjr63WU+MzMzaGlp5Uu/2sNd86Wfr837f1ng0NM96huIRDUt1UL8OiLxXg6RRVMZfXvxa+Yna8xRzpifnL2fn8LgvKh1oW3ris+uQtvWl0gST9zU0tKClZWVuodBREREpDZFY8Y0ERERkcSxKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlREREJEmhoaFo3bo1rK2tIZPJEBwcrLK8V69ekMlkKq9mzZqpxMTExKBr164wNDSEsbEx+vbti8TERJWYy5cvo379+tDR0YGNjQ1mzJiRaSybN2+Go6MjdHR04OzsjD179qgsFwQBEyZMgJWVFXR1deHp6Yk7d+581P6yKCMiIiJJSkpKgouLCxYtWpRtTLNmzfD8+XPx9ffff6ss79q1K65du4aQkBDs3r0boaGh6N+/v7g8ISEBXl5eKFu2LM6fP4+ZM2ciICAAy5YtE2NOnjyJzp07o2/fvrh48SLatWuHdu3a4erVq2LMjBkzsGDBAgQFBeH06dPQ09ODt7c3kpOT87y/MkEQhDxHExER0VfFeVHrQtvWFZ9dn7yuTCbD9u3b0a5dO7GtV69eiIuLy3QGLcONGzfg5OSEs2fPokaNGgCAffv2oUWLFnjy5Amsra2xZMkS/PLLL4iMjIS2tjYAYMyYMQgODsbNmzcBAN9//z2SkpKwe/duse/atWvD1dUVQUFBEAQB1tbWGDFiBEaOHAkAiI+Ph4WFBVavXo0ffvghT/vIM2VERERUKBQKBRISElReCoXis/o8evQozM3NUbFiRQwaNAivXr0Sl4WFhcHY2FgsyADA09MTGhoaOH36tBjToEEDsSADAG9vb9y6dQuxsbFijKenp8p2vb29ERYWBgB48OABIiMjVWKMjIzg7u4uxuQFizIiIiIqFIGBgTAyMlJ5BQYGfnJ/zZo1w59//olDhw7ht99+w7Fjx9C8eXOkp6cDACIjI2Fubq6yTrFixWBqaorIyEgxxsLCQiUm431uMe8vf3+9rGLyolieI79Ah57uyT2oCGpaqoX49dxLs9Q4Emnycxkpfh2ReE+NI5GmMvr24tfMT9aYo5wxPzl7Pz9fG39/fwwfPlylTS6Xf3J/718WdHZ2RtWqVWFvb4+jR4+iadOmn9yvuvBMGRERERUKuVwOQ0NDldfnFGUfKleuHMzMzHD37l0AgKWlJV68eKESk5aWhpiYGFhaWooxUVFRKjEZ73OLeX/5++tlFZMXLMqIiIjoq/DkyRO8evUKVlZWAAAPDw/ExcXh/PnzYszhw4ehVCrh7u4uxoSGhiI1NVWMCQkJQcWKFWFiYiLGHDp0SGVbISEh8PDwAADY2dnB0tJSJSYhIQGnT58WY/KCRRkRERFJUmJiIsLDwxEeHg7g3YT68PBwREREIDExEaNGjcKpU6fw8OFDHDp0CG3btkX58uXh7e0NAKhUqRKaNWuGfv364cyZMzhx4gR8fX3xww8/wNraGgDQpUsXaGtro2/fvrh27Ro2btyI+fPnq1xmHTp0KPbt24fZs2fj5s2bCAgIwLlz5+Dr6wvg3Z2hw4YNw5QpU7Bz505cuXIFPXr0gLW1tcrdorn5queUERER0Zfr3LlzaNy4sfg+o1Dq2bMnlixZgsuXL2PNmjWIi4uDtbU1vLy88Ouvv6pcEv3rr7/g6+uLpk2bQkNDAx06dMCCBQvE5UZGRjhw4AB8fHzg5uYGMzMzTJgwQeVZZnXq1MH69esxbtw4jB07FhUqVEBwcDCqVKkixowePRpJSUno378/4uLiUK9ePezbtw86Ojp53t+v+jllnOifNU70zxkn+ueMk7RzxxzljPnJWWFP9P9SnlNWFPDyJREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBxdQ9ACnavXof9vy5X6XNwsYcE9f4q7QJgoBF/stw/cxN9J/cB671nAEAT+49xYH1h3D36gMkxSfB1NIE9VvXQZMODcV1b4ffxbzhizJtO3DLJBiZGmZq37/+IHas+AeNv22A73zb58dufjKlUolzmy7gzr938CbuLfRMi6NiQwdU71ANMplMjIt9EotTf53B8+vPoVQKMCltDK8R38DATB8AcP3gDdw5fg/RD6KR+jYVvVf1gFxPnml7jy5E4PyWC3j1KAaa2pqwrmSFZqO9MsUlv07G5lHbkBSTlG1fX6Idm3Zh859bEfMqFvYV7OAzehAcq1RU97Akg/nJGfOTO+aIpIJFWTasbC0xZNYg8b2mZuaTioe3HIMMskztEbefQN9EH73GdoVpSWPcu/YQ6+dsgoaGBhq1r68SO3GNP3T0dMT3Bsb6mfp7eDMCx3eHoVQ568/ZpXwTHnwJ10Ouo7FPI5iUNsHL+y9xdHEotItrw7lFFQBAfGQCgifsgmOTiqjZyQ1autqIfRKLYlqaYj9pijSUcS2NMq6lcXr92Sy3df/UAxxb+i9qda6BUlWsoVQKiImIyTL26JJQmJY1RVJMUv7vtJocPXAMS+csx5CxvqhUxRHb1gfD33c8Vm5bBhNTY3UPT+2Yn5wxP7ljjkhKePkyG5qaGjAyNRRf+kaqxdLju09xaPNRdBv9Q6Z16zR3Ryffb+HgUh5m1mZw/6YGPJrVQvi/lzPFGpgYqGxHQ0P1vyT5rQKrp61D1xGdUNxAN3938hNF3o6CbY2yKFu9DAzNDWBfuxxKVy2FF3dfijFnNpxFmWo28OjmDjM7MxhZGsK2RlnoGv23D1VbOqNaO1eYVzDPcjvKdCVOrA5D7e61UNnLCcbWxjAtbYLydewzxV47cB2KNylwbe2c/zusRlvXbUfz9s3QrI0XypYrg6FjfSHXkWP/jgPqHpokMD85Y35yxxyRlLAoy8aLp9Hw/24ixnf9FaumrkVMVKy4LCU5BaumrsX3QztkeakxK2+TkqFnWDxT+7R+MzGm4wQsGLUE967ez7R84/wtqOJeCY5u0jmVbulggSdXnyHuWRwAIPrhK0TeioJNNRsAgKAUEHHhMYytjLB76h6s/nEtto0NxoMzDz9qOy8fRCMpJgkymQybR2/Dn/3X4Z9pezOdKYt5EovzWy6giW8jQJb5zOWXKjU1Fbdv3kX1Wq5im4aGBqrXcsX1KzfVNzCJYH5yxvzkjjkiqZF0Ufb48WP06dOn0LdrV6kseozuDJ/pA9B52HeIfh6DOUN/R/KbZADAlsXBKFfZFi5183ZW5t7VBzh/5CLqtvQQ24xMDdHZ7zv0n9Qb/QJ6w6SkMeb6LULE7cdizLnDF/D4zlO07dcqf3fwM1Vr54rydeyxwW8zlnVegS0/b4NziypwqF8eAPA24S1Sk1Nxcccl2LjYoNW4FrCrZYv9s0Pw7PrzPG/nddRrAMC5zRfg9m01NB/jDbmeHDsn7UZy4rv/i/TUdByafxi1u7mLc9W+FvFxCVCmK2FSwkSl3aSEMWKjs76EW5QwPzljfnLHHJHUSHpOWUxMDNasWYOVK1dmG6NQKKBQKFTa5HI55PJPn+Rd2b3Sf2/srWFbqSzGdZ6M80fDYWCkj1sX78B/2cg89fXswXMsHf8HWvbwhlNNR7Hdoow5LMr8d9nOvoodXj57hcNbjqHX2G6IeRGLzYu2Y/CMQdDS1vrkfSkI98Lu487xu/Ac0gQmNiZ49fAVTqwOg55JcVRs5ABBKQAAbGuUhUurd4WrmW0JRN6KwvUDN2DtZJWn7QjCu36qf+uKcrXtAACNf2qItQPX437YAzh9Uwmn15+BcSljODSoUAB7SkREVHjUWpTt3Lkzx+X372e+nPehwMBATJo0SaVt4sSJCAgI+JyhqSiurwvz0iXx8mk0nt1/juhnrzCy9ViVmOUBq1DeuRz85vqKbc8fRmL+yMWo28oDzbtnvlvwQ7aOZcRLmBG3n+B1bCKmD5gtLlcqlbh7+T6OBR/Hgv0zoZHFzQeFIWzdaVRr64Lydd/N7SpRxhSvX77GxeBwVGzkAB1DHWhoymBS2lhlPZNSxnh+KyrP2ylu/O5yr0np/36L1dTShKGFAV5HJwIAnl59hpiIWCw9teJdwLs6Dqv7rkX1b6uhZie3T9xL9TMyNoSGpgZiX8WqtMe+ioOJmamaRiUdzE/OmJ/cMUckNWotytq1aweZTCaeEcmKLJc5Qv7+/hg+fLhK2+ecJctK8lsFop+9gtE3hqjeyBV1W9ZWWT6l7wx0/KkdnD0qi23PHjzH/JGL4e5VE237tszTdp7cewrD/89Rc6xeAeP+GK2y/M8Zf8PSxhxenZuqrSAD3t01KdNQ/X+RaWiI/4+axTRR0r4k4p7Fq8TEPY//qEuMJcuZQVNLE3HP4mDlaAkASE9T4vXLRBiUfNeP14hvkJ6SJq7z4t5LHF0SiraTW8PIIm/z/aRKS0sLDo7lcfHsJdRtXAfAu8L84tlwtO3UWs2jUz/mJ2fMT+6YI5IatRZlVlZWWLx4Mdq2bZvl8vDwcLi55Xym43MvVWZl65IdcK5TGSUsTBEXHY9/1uyDhoYMNZpUh4GxfpaT+03MTWBmVQLAu4Js3ojFcKpREU2/a4T4mAQA7yaQZjzy4vCWYyhhZQorW0ukpaThxD+ncOviHQyeMRAAoFNcB9Z2qpf55Dra0DPUy9Re2Mq6lcGFbeHQN9OHSWkTvHoYjcu7r8CxsYMY49qmKkLmHoZVJSuUqmKFx+FP8Oh8BNoE/Dc/7k3cG7yJe4uEyHf5iYmIgZauNvTN9KCjrwPt4tpw+qYSzm26AP0S+jAoqY/wne/uYLWvXQ4AYGSp+n/x9vW7uWYmpYy/iueUdejWHjMmzoFDpQqoWMUB29fvQPJbBbzbfKPuoUkC85Mz5id3zBFJiVqLMjc3N5w/fz7boiy3s2gFJS46HqumrEVSQhL0jfRh71wOoxYOy/IZYlm5cOwSEuMScebgeZw5eF5sN7UwwZS/JwAA0tLSsG3JTsRFx0NbRwvW5awxZOYgVKwm/blR9frUwdmN5/HvihN4G//u4bFO3zjCrWN1Mcaulh0a9KuHC8HhOLHqJIytjeA1wlM84wUA1w7cwPktF8T3OybuBgA0+qkhHBu9K/Bqd3OHhoYMhxceRVpKGszLm6P1hBaQ63/5BVdeNPJqiLjYBKwJWovYV7GwdyiHab9PzjQxuahifnLG/OSOOSIpkQnqqHr+799//0VSUhKaNWuW5fKkpCScO3cODRs2zHJ5bg493fM5w/tqNS3VQvx67qVZahyJNPm5/HcTR0TiPTWORJrK6P/3nDjmJ2vMUc6Yn5y9n5/C4Lyo8C7VXvHZVWjb+hKp9UxZ/fr1c1yup6f3yQUZERER0ZdE0s8pIyIiIioqWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBav2D5ERERKReTnal1D0E+j+eKSMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiIJCk0NBStW7eGtbU1ZDIZgoODxWWpqan4+eef4ezsDD09PVhbW6NHjx549uyZSh+2traQyWQqr+nTp6vEXL58GfXr14eOjg5sbGwwY8aMTGPZvHkzHB0doaOjA2dnZ+zZs0dluSAImDBhAqysrKCrqwtPT0/cuXPno/aXRRkRERFJUlJSElxcXLBo0aJMy968eYMLFy5g/PjxuHDhArZt24Zbt26hTZs2mWInT56M58+fi6/BgweLyxISEuDl5YWyZcvi/PnzmDlzJgICArBs2TIx5uTJk+jcuTP69u2Lixcvol27dmjXrh2uXr0qxsyYMQMLFixAUFAQTp8+DT09PXh7eyM5OTnP+8s/SE5ERESS1Lx5czRv3jzLZUZGRggJCVFpW7hwIWrVqoWIiAiUKVNGbDcwMIClpWWW/fz1119ISUnBypUroa2tjcqVKyM8PBxz5sxB//79AQDz589Hs2bNMGrUKADAr7/+ipCQECxcuBBBQUEQBAHz5s3DuHHj0LZtWwDAn3/+CQsLCwQHB+OHH37I0/7yTBkREREVCoVCgYSEBJWXQqHIt/7j4+Mhk8lgbGys0j59+nSUKFEC1apVw8yZM5GWliYuCwsLQ4MGDaCtrS22eXt749atW4iNjRVjPD09Vfr09vZGWFgYAODBgweIjIxUiTEyMoK7u7sYkxcsyoiIiKhQBAYGwsjISOUVGBiYL30nJyfj559/RufOnWFoaCi2DxkyBBs2bMCRI0cwYMAATJs2DaNHjxaXR0ZGwsLCQqWvjPeRkZE5xry//P31sorJC16+JCIiokLh7++P4cOHq7TJ5fLP7jc1NRWdOnWCIAhYsmSJyrL3t1e1alVoa2tjwIABCAwMzJdt56evuihrWqqFuocgeX4uI9U9BEkro2+v7iFIGvOTO+YoZ8xP0SKXy/O9EMooyB49eoTDhw+rnCXLiru7O9LS0vDw4UNUrFgRlpaWiIqKUonJeJ8xDy27mPeXZ7RZWVmpxLi6uuZ5X3j5koiIiL5IGQXZnTt3cPDgQZQoUSLXdcLDw6GhoQFzc3MAgIeHB0JDQ5GamirGhISEoGLFijAxMRFjDh06pNJPSEgIPDw8AAB2dnawtLRUiUlISMDp06fFmLz4qs+UERER0ZcrMTERd+/eFd8/ePAA4eHhMDU1hZWVFTp27IgLFy5g9+7dSE9PF+dvmZqaQltbG2FhYTh9+jQaN24MAwMDhIWFwc/PD926dRMLri5dumDSpEno27cvfv75Z1y9ehXz58/H3Llzxe0OHToUDRs2xOzZs9GyZUts2LAB586dEx+bIZPJMGzYMEyZMgUVKlSAnZ0dxo8fD2tra7Rr1y7P+ysTBEHIh7xJUkTiPXUPQZLev1zAHGX2fn4OPd2TQ2TR9P60AB4/WeNnLGfMT84K+5Lu93sGFtq2NrYI+qj4o0ePonHjxpnae/bsiYCAANjZ2WW53pEjR9CoUSNcuHABP/30E27evAmFQgE7Ozt0794dw4cPV7mMevnyZfj4+ODs2bMwMzPD4MGD8fPPP6v0uXnzZowbNw4PHz5EhQoVMGPGDLRo8d/3Q0EQMHHiRCxbtgxxcXGoV68eFi9eDAcHhzzvL4uyIojfEHPGoixnLMpyx89YzpifnLEoK7o4p4yIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUlAMXUP4Gu3Y9MubP5zK2JexcK+gh18Rg+CY5WK6h6WZHyN+QndcQKhu04gJjIGAGBla4kW3b1R2b2SSpwgCFjkvwzXz9xE/8l94FrPOVNfifFJmNZvJuKi4zFr5zQU19fNFHPv6n3MHbYI1naWGLt8VJZj2r/+IHas+AeNv22A73zb58NeSsfXeAzlJ+Ynd8wRSQXPlBWgoweOYemc5ejWvwuW/PU7yjmUg7/veMTGxKl7aJLwtebHuKQR2v3YCmOCRuDnJcPhUK0Cgsb/gWcPnqvEHd5yDDLIcuxr3awNKFXOOtvlbxLfYk3gelSsXiHbmIc3I3B8d1iO/XypvtZjKL8wP7ljjkhKWJQVoK3rtqN5+2Zo1sYLZcuVwdCxvpDryLF/xwF1D00Svtb8VK1TBVVqO8G8dElY2Jijbd+WkOvK8eDGIzHm8d2nOLT5KLqN/iHbfkJ3nMDbxLfw7NQ425i/525GzabVYedkm+Xy5LcKrJ62Dl1HdEJxg8xn2b50X+sxlF+Yn9wxRyQlLMoKSGpqKm7fvIvqtVzFNg0NDVSv5YrrV26qb2ASUVTyo0xX4tzhC0hJVqDc/wunlOQUrJq6Ft8P7QAjU8Ms13v+MBJ71u5HzzFdIdPI+mxa2N7TiH72Ci16eme7/Y3zt6CKeyU4un19l2KKyjH0qZif3DFHJDWcU1ZA4uMSoExXwqSEiUq7SQljPH74WE2jko6vPT9P7z/DLN/5SE1Jg1xXG/0n9YGVrSUAYMviYJSrbAuXupnnkAFAakoaVk5Zi/YD2sDUwgTRz19linnx5CWCV+zG8HmDoampmWU/5w5fwOM7T/HzEr/82zEJ+dqPoc/F/OSOOSKpUfuZsrdv3+L48eO4fv16pmXJycn4888/c1xfoVAgISFB5aVQKApquER5YmFjDv/lIzF68TDUb1MXf/62Hs8fRuLyiau4dfEOOvpkP9l+x4rdsCxrAfdvamS5XJmuxMqpa9GyZzNY2JhnGRPzIhabF21Hr7HdoKWtlS/7REREBUutZ8pu374NLy8vREREQCaToV69etiwYQOsrKwAAPHx8ejduzd69OiRbR+BgYGYNGmSStvEiRMREBBQkEPPlZGxITQ0NRD7KlalPfZVHEzMTNU0Kun42vNTTKsYzEuVBACUcbDBo1sROLItFFraWoh+9gojW49ViV8esArlncvBb64vbl+8g6cPnuPisUsAAAECAGB0u3Fo1s0TTTo2QsStx3hy5yk2Ldj2LkYQIAgCfD1HYPCMgXj7JhmvYxMxfcBscRtKpRJ3L9/HseDjWLB/JjQ01f472Wf52o+hz8X85I45IqlRa1H2888/o0qVKjh37hzi4uIwbNgw1K1bF0ePHkWZMmXy1Ie/vz+GDx+u0iaXywtiuB9FS0sLDo7lcfHsJdRtXAfAux+KF8+Go22n1moenfoVtfwISgFpqWlo2asZ6rasrbJsSt8Z6PhTOzh7VAYA9AvojdSUVHH5o5sRWDtzA4bPH4yS1iWgU1yOcX+MVunj2I4TuH3xDvoF9EIJS1MIgpAp5s8Zf8PSxhxenZt+8QUZUPSOoY/F/OSOOSKpUWtRdvLkSRw8eBBmZmYwMzPDrl278NNPP6F+/fo4cuQI9PT0cu1DLpdLogjLSodu7TFj4hw4VKqAilUcsH39DiS/VcC7zTfqHpokfK35CV6+G5VrVYKphQmS3yTj7KELuHPpHnx/GwAjU8MsJ/ebmJvAzKoEAKBkKTOVZYnxSQAAy7IW4nPKrO2sVGIMjPWhpV1Mpf3DGLmONvQM9TK1f8m+1mMovzA/uWOOSErUWpS9ffsWxYr9NwSZTIYlS5bA19cXDRs2xPr169U4us/XyKsh4mITsCZoLWJfxcLeoRym/T4506TSouprzc/ruESsmf4XEmISoKOni1LlrOD72wBUqvH13QGpbl/rMZRfmJ/cMUckJTJBEAR1bbxWrVoYPHgwunfvnmmZr68v/vrrLyQkJCA9Pf2T+o9IvPe5Q/wqldG3F79mjjJ7Pz+Hnu5R40ikqWmpFuLXPH6yxs9YzpifnL2fn8Lw/Z6BhbatjS2CCm1bXyK1Tixp3749/v777yyXLVy4EJ07d4Yaa0YiIiKiQqPWoszf3x979mR/JmLx4sVQKpWFOCIiIiIi9fjyb8EiIiIi+gqwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikoBieQnauXNnnjts06bNJw+GiIiIqKjKU1HWrl27PHUmk8mQnp7+OeMhIiIiKpLyVJTxj4ITERERFSzOKSMiIiKSgDydKftQUlISjh07hoiICKSkpKgsGzJkSL4MjIiIiKgo+eii7OLFi2jRogXevHmDpKQkmJqaIjo6GsWLF4e5uTmLMiIiIqJP8NGXL/38/NC6dWvExsZCV1cXp06dwqNHj+Dm5oZZs2YVxBiJiIiIvnofXZSFh4djxIgR0NDQgKamJhQKBWxsbDBjxgyMHTu2IMZIRERE9NX76KJMS0sLGhrvVjM3N0dERAQAwMjICI8fP87f0REREREVER9dlFWrVg1nz54FADRs2BATJkzAX3/9hWHDhqFKlSr5PkAiIiIqmkJDQ9G6dWtYW1tDJpMhODhYZbkgCJgwYQKsrKygq6sLT09P3LlzRyUmJiYGXbt2haGhIYyNjdG3b18kJiaqxFy+fBn169eHjo6OePXvQ5s3b4ajoyN0dHTg7OyMPXv2fPRYcvPRRdm0adNgZWUFAJg6dSpMTEwwaNAgvHz5EsuWLfvY7oiIiIiylJSUBBcXFyxatCjL5TNmzMCCBQsQFBSE06dPQ09PD97e3khOThZjunbtimvXriEkJAS7d+9GaGgo+vfvLy5PSEiAl5cXypYti/Pnz2PmzJkICAhQqWlOnjyJzp07o2/fvrh48SLatWuHdu3a4erVqx81ltzIBEEQPiZBX5KIxHvqHoIkldG3F79mjjJ7Pz+Hnu7JIbJoalqqhfg1j5+s8TOWM+YnZ+/npzB8v2dgoW1rY4ugT15XJpNh+/bt4l8ZEgQB1tbWGDFiBEaOHAkAiI+Ph4WFBVavXo0ffvgBN27cgJOTE86ePYsaNWoAAPbt24cWLVrgyZMnsLa2xpIlS/DLL78gMjIS2traAIAxY8YgODgYN2/eBAB8//33SEpKwu7du8Xx1K5dG66urggKCsrTWPKCD48lIiKiQqFQKJCQkKDyUigUn9TXgwcPEBkZCU9PT7HNyMgI7u7uCAsLAwCEhYXB2NhYLMgAwNPTExoaGjh9+rQY06BBA7EgAwBvb2/cunULsbGxYsz728mIydhOXsaSFx9dlNnZ2aFcuXLZvoiIiIiyEhgYCCMjI5VXYGDgJ/UVGRkJALCwsFBpt7CwEJdFRkbC3NxcZXmxYsVgamqqEpNVH+9vI7uY95fnNpa8+OiHxw4bNkzlfWpqKi5evIh9+/Zh1KhRH9sdERERFRH+/v4YPny4SptcLlfTaKTno4uyoUOHZtm+aNEinDt37rMHRERERF8nuVyeb0WYpaUlACAqKkq8ATHjvaurqxjz4sULlfXS0tIQExMjrm9paYmoqCiVmIz3ucW8vzy3seRFvs0pa968ObZu3Zpf3RERERFly87ODpaWljh06JDYlpCQgNOnT8PDwwMA4OHhgbi4OJw/f16MOXz4MJRKJdzd3cWY0NBQpKamijEhISGoWLEiTExMxJj3t5MRk7GdvIwlL/KtKNuyZQtMTU3zqzsiIiIq4hITExEeHo7w8HAA7ybUh4eHIyIiAjKZDMOGDcOUKVOwc+dOXLlyBT169IC1tbV4h2alSpXQrFkz9OvXD2fOnMGJEyfg6+uLH374AdbW1gCALl26QFtbG3379sW1a9ewceNGzJ8/X+Uy69ChQ7Fv3z7Mnj0bN2/eREBAAM6dOwdfX18AyNNY8uKjH4lRrVo1yGQy8b0gCIiMjMTLly+xePFilWd/EBERkbRJ+ZEYR48eRePGjTO19+zZE6tXr4YgCJg4cSKWLVuGuLg41KtXD4sXL4aDg4MYGxMTA19fX+zatQsaGhro0KEDFixYAH19fTHm8uXL8PHxwdmzZ2FmZobBgwfj559/Vtnm5s2bMW7cODx8+BAVKlTAjBkz0KLFf48IystYcvPRRVlAQIBKUaahoYGSJUuiUaNGcHR0/JiuiIiISM3mXppVaNvycxlZaNv6En30RP+AgIACGAYRERFR0fbRRZmmpiaeP3+e6bkfr169grm5OdLT0/NtcJ+LT4rOGp+mnTPmJ2fMT+6Yo5wxPzkr7Cf6k3R89ET/7K52KhQKlafhEhEREVHe5flM2YIFCwC8u8NgxYoVKhPk0tPTERoayjllRERERJ8oz0XZ3LlzAbw7UxYUFARNTU1xmba2NmxtbREU9Ol/aJSIiIioKMtzUfbgwQMAQOPGjbFt2zbxgWpERERE9Pk+eqL/kSNHCmIcREREREXaR0/079ChA3777bdM7TNmzMB3332XL4MiIiIiKmo+uigLDQ1VeYJthubNmyM0NDRfBkVERERU1Hx0UZaYmJjloy+0tLSQkJCQL4MiIiIiKmo+uihzdnbGxo0bM7Vv2LABTk5O+TIoIiIioqLmoyf6jx8/Ht9++y3u3buHJk2aAAAOHTqE9evXY8uWLfk+QCIiIqKi4KOLstatWyM4OBjTpk3Dli1boKurCxcXFxw+fBimpqYFMUYiIiKir95HF2UA0LJlS7Rs2RIAkJCQgL///hsjR47E+fPnJfW3L4mIiIi+FB89pyxDaGgoevbsCWtra8yePRtNmjTBqVOn8nNsREREREXGR50pi4yMxOrVq/HHH38gISEBnTp1gkKhQHBwMCf5ExEREX2GPJ8pa926NSpWrIjLly9j3rx5ePbsGX7//feCHBsRERFRkZHnM2V79+7FkCFDMGjQIFSoUKEgx0RERERU5OT5TNnx48fx+vVruLm5wd3dHQsXLkR0dHRBjo2IiIioyMhzUVa7dm0sX74cz58/x4ABA7BhwwZYW1tDqVQiJCQEr1+/LshxEhEREX3VPvruSz09PfTp0wfHjx/HlStXMGLECEyfPh3m5uZo06ZNQYyRiIiI6Kv3yY/EAICKFStixowZePLkCf7+++/8GhMRERFRkfNZRVkGTU1NtGvXDjt37syP7oiIiIiKnHwpyoiIiIjo87AoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAF5/oPk9Gl2bNqFzX9uRcyrWNhXsIPP6EFwrFJR3cOSDOYnd0UxRxtWbcIfC1ejfee2+GnkAABAiiIFQXOX4+iBUKSmpKKGR3UMGeMDkxImmdZPiEvAgM4+iH7xCtuPboK+gT4A4OrFa1j++0o8fvgEimQFLCzN0bJDc3To2r5Q9+9TXL5wBZv/3IrbN+4iJjoGAbPGoW7jOuLyP5euw9H9oXgZ9RLFtLRQoVJ59P6pByo5OwIALp27jJEDxmTZ98I/56FiZQcAwNmT5/Hn0nV4dD8C2tpacK5eBQP8+sHS2qLgd1JNiuJnjKSJZ8oK0NEDx7B0znJ0698FS/76HeUcysHfdzxiY+LUPTRJYH5yVxRzdOvabfyzbS/KVbBTaV8yexlOhZ7B+On+mL38N7x6GYOAUVOy7GP25Hmw+2B9ANDR1UHbTq0xZ/kM/LFlKbr8+ANWL/4T/2zbWyD7kp+S3yajnIMdBv/8U5bLS5cpBd+fB2HZxsWY+8dMWFiZY4zPOMTFxgMAnFwqYeP+dSqv5u28YVnKEg5OFQAAz59GYuKIyXCt6YKg9QsRuHAK4uMSMGlk1nn+GhTFzxhJF4uyArR13XY0b98Mzdp4oWy5Mhg61hdyHTn27zig7qFJAvOTu6KWo7dv3iJw3Az4jRsCfUN9sT3pdRL27TiAgcP7oVotVzhUqoCRE/1w/dINXL9yU6WPXZv/QWJiEr7r9m2m/ss72qNJs0awtS8LS2sLeLZoAjcPN1y5eLXA9+1z1apbE71/6ol6TepkubxJ88ao7l4NVqWtYGtfFgOH98ebpDe4f+cBAEBLSwumZqbiy9DIEGHHTsG7tSdkMhkA4M6Nu1CmK9H7px6wtrFChUrl8V33Drh3+z7SUtMKbV8LU1H7jJG0sSgrIKmpqbh98y6q13IV2zQ0NFC9lmumHyJFEfOTu6KYo9+nL4Z7vVqo7l5Npf32jTtIS0tDdXdXsa2MnQ3MLUvixuUbYtuj+xFYt3w9fp40AhoauX97u3vzHq5fvoGq1Z3zbR+kIDU1FXu27YWevh7sszhjCABhoaeQEP8a3m28xLYKlcpDQ0OG/TtDkJ6ejqTXSTj4zyFUq+WKYlpf32yXovgZI2lT+6fsxo0bOHXqFDw8PODo6IibN29i/vz5UCgU6NatG5o0aZLj+gqFAgqFQqVNLpdDLpcX5LBzFR+XAGW6MtN8F5MSxnj88LGaRiUdzE/uilqOjuw/hjs372LR2vmZlsW+ioWWVjFxblgGkxImiHkVCwBISUnFtLG/od+wvjC3Msfzp5HZbqtz8+6Ij41HeroS3ft3QYv2zfJ3Z9TkVOhpTB37GxTJCpiameK3xVNhZGKUZezeHQfg5lEdJS3MxDarUpYIXDQVU8YEYt6036FMV8KpaiVMXTCpsHahUBW1zxhJn1rPlO3btw+urq4YOXIkqlWrhn379qFBgwa4e/cuHj16BC8vLxw+fDjHPgIDA2FkZKTyCgwMLKQ9IKL88CLyJRbPWgr/qaOhLdf+pD5WLlyFMnY28GyR8y9yADBnxUwsWjsfQ/19se3vHTi87+gnbVNqXGq6IOjvhZi3ajZq1nHDlDGBWc6NehkVjfNhF9C8rZdKe0x0DOZOmQ+vVk2x6M/5mL38NxTTKobJo6dBEIRC2guiokutZ8omT56MUaNGYcqUKdiwYQO6dOmCQYMGYerUqQAAf39/TJ8+PcezZf7+/hg+fLhKm7rPkgGAkbEhNDQ1EPv/3+IzxL6Kg4mZqZpGJR3MT+6KUo7u3LiDuJg4DOo6WGxTpitx5cJV7Ni0C4ELpyA1NQ2JrxNVzpbFvoqF6f/Pclw8exkP7z5E6KFW7xb+v4bo0PQHdOnzA3oO7CauZ1XKEgBgV8EOsTGxWLvsLzRp1qhgd7IQ6OrqoJSNNUrZWMPJ2RE92/2IfcH70bnP9ypx+3cegKGRATwa1FZp37lpN/T09dBvaF+xbcyvo9ClRQ/cuHoLTv+/k/NrUZQ+Y/RlUGtRdu3aNfz5558AgE6dOqF79+7o2LGjuLxr165YtWpVjn1I4VJlVrS0tODgWB4Xz14Sb1tXKpW4eDYcbTu1VvPo1I/5yV1RylG1Wq5YtnGxStusSXNhY1sa3/f8DuYWJVGsWDFcPBOO+k3rAQAeP3yCF5EvUalqJQDAxBm/qExluHX9NmZPmoe5K2bCqrRVtttWKgWkpqQWwF6pn6BUIjVVdd8EQcD+XQfh2bJppnliyckKaPx/0n+GjLl5glJZsINVg6L0GaMvg9rnlGXc9aOhoQEdHR0YGf03/8HAwADx8fHqGtpn69CtPWZMnAOHShVQsYoDtq/fgeS3Cni3+UbdQ5ME5id3RSVHxfWKw668rUqbjq4ODI0MxfZmbb0QNGc5DAwNUFy/OBbNCIJT1Uri2RtrG9XCKyEuAcC7GwIyzq7t2LQL5pbmsLEtDQC4cuEqtqzbinY/tCnAvcsfb9+8xdPHz8T3kc+icPfWPRgaGsDA2BDr/9gAj4a1UcLMBPFxCdi5aTeiX75CA8/6Kv1cPHsJkU8j0bydd6ZtuNeriW3rg7F22Xo0btYQb5PeYuWiNbCwMkf5ivYFvo/qUFQ+Y/RlUGtRZmtrizt37sDe/t2HPSwsDGXKlBGXR0REwMoq+99wpa6RV0PExSZgTdBaxL6Khb1DOUz7fXKWD7ssipif3DFH/xk0oj9kGjJMHj0VqSmpcPNww5AxWT+zKzuCUsDKhasR+TQSGpqasC5thR8H90HLDs0LaNT55/b1OyoPfw2asxwA8E0rTwwb64vHD58gZPdUJMTFw8DIEBUrO2DuipmwtS+r0s++4P1wcqmEMnY2mbZRrZYr/KeOxqY1W7Dpzy3Q0ZGjUtVKmPb7r5DrSO+KRH7gZ4ykRCaocfZmUFAQbGxs0LJlyyyXjx07Fi9evMCKFSs+qf+IxHufM7yvVhn9/37jZY4yY35yxvzkjjnKGfOTs/fzUxjmXppVaNvycxlZaNv6Eqn1TNnAgQNzXD5t2rRCGgkRERGRevHhsUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlREREJDm2traQyWSZXj4+PgCARo0aZVo2cOBAlT4iIiLQsmVLFC9eHObm5hg1ahTS0tJUYo4ePYrq1atDLpejfPnyWL16daaxLFq0CLa2ttDR0YG7uzvOnDlTIPvMooyIiIgk5+zZs3j+/Ln4CgkJAQB89913Yky/fv1UYmbMmCEuS09PR8uWLZGSkoKTJ09izZo1WL16NSZMmCDGPHjwAC1btkTjxo0RHh6OYcOG4ccff8T+/fvFmI0bN2L48OGYOHEiLly4ABcXF3h7e+PFixf5vs8syoiIiEhySpYsCUtLS/G1e/du2Nvbo2HDhmJM8eLFVWIMDQ3FZQcOHMD169exbt06uLq6onnz5vj111+xaNEipKSkAACCgoJgZ2eH2bNno1KlSvD19UXHjh0xd+5csZ85c+agX79+6N27N5ycnBAUFITixYtj5cqV+b7PLMqIiIioUCgUCiQkJKi8FApFruulpKRg3bp16NOnD2Qymdj+119/wczMDFWqVIG/vz/evHkjLgsLC4OzszMsLCzENm9vbyQkJODatWtijKenp8q2vL29ERYWJm73/PnzKjEaGhrw9PQUY/ITizIiIiIqFIGBgTAyMlJ5BQYG5rpecHAw4uLi0KtXL7GtS5cuWLduHY4cOQJ/f3+sXbsW3bp1E5dHRkaqFGQAxPeRkZE5xiQkJODt27eIjo5Genp6ljEZfeSnYvneIxEREVEW/P39MXz4cJU2uVye63p//PEHmjdvDmtra7Gtf//+4tfOzs6wsrJC06ZNce/ePdjb2+ffoAsRizIiIiIqFHK5PE9F2PsePXqEgwcPYtu2bTnGubu7AwDu3r0Le3t7WFpaZrpLMioqCgBgaWkp/pvR9n6MoaEhdHV1oampCU1NzSxjMvrIT7x8SURERJK1atUqmJubo2XLljnGhYeHAwCsrKwAAB4eHrhy5YrKXZIhISEwNDSEk5OTGHPo0CGVfkJCQuDh4QEA0NbWhpubm0qMUqnEoUOHxJj8xKKMiIiIJEmpVGLVqlXo2bMnihX77+LevXv38Ouvv+L8+fN4+PAhdu7ciR49eqBBgwaoWrUqAMDLywtOTk7o3r07Ll26hP3792PcuHHw8fERz9YNHDgQ9+/fx+jRo3Hz5k0sXrwYmzZtgp+fn7it4cOHY/ny5VizZg1u3LiBQYMGISkpCb179873/eXlSyIiIpKkgwcPIiIiAn369FFp19bWxsGDBzFv3jwkJSXBxsYGHTp0wLhx48QYTU1N7N69G4MGDYKHhwf09PTQs2dPTJ48WYyxs7PDP//8Az8/P8yfPx+lS5fGihUr4O3tLcZ8//33ePnyJSZMmIDIyEi4urpi3759mSb/5weZIAhCvvcqERGJ99Q9BEkqo//fBEjmKDPmJ2fMT+6Yo5wxPzl7Pz+FYe6lWYW2LT+XkYW2rS8RL18SERERScBXffmysH/b+BIxRzljfnLG/OSOOcoZ80P0H54pIyIiIpIAFmVEREREEvBVX77kBNKscZJtzpifnDE/uWOOcsb85IyXdIsunikjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkoJi6B0BERETqU9XMSd1DoP/jmTIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBPAPkhewHZt2YfOfWxHzKhb2FezgM3oQHKtUVPewJIP5yV1RyNHlC1ew+c+tuH3jLmKiYxAwaxzqNq4jLp8xcQ5Cdh9UWaeGhxsCF/6aqa+UlFQM7umH+7fvY8n631G+oj0A4NK5y9i6Phi3rt7Cm6Q3sC5TCp26d0DTFo0Ldufywd8rN+L4kZN4/PAJ5HJtOFWthB+H9IGNbWkAQOSzKHRv3TvLdcdN90fDb+oDAG5du40Vv6/CnRt3IZPJULGyA/oN7QN7h3IAgD+XrsPaZesz9aGjI8euE9sLaO8KTvSLaKxYsApnTp6DIlkB69JWGBngh4pODgDe7e/R/aF4GfUSxbS0UKFSefT+qQcqOTuKfYz3m4R7t+4jLjYOBgb6qObuih+H9IFZyRLq2i36irEoK0BHDxzD0jnLMWSsLypVccS29cHw9x2PlduWwcTUWN3DUzvmJ3dFJUfJb5NRzsEO3m28MGnUlCxjatZxw8iJfuJ7LW2tLOOWz/8DJUqa4v7t+yrt1y7fQLnytvi+Z0eYmJrg1L+nMWPibOjpF0ftBu75tzMF4PKFq2jzXStUrOyA9PR0rFy4BmN8fsGKLUuhq6uDkhZm2Lh/nco6/2zbh81rt6JW3RoAgLdv3sJ/8Hh4NHDHkDE+SE9Px59L18HfdzzW/7MGxbSK4bvuHdCqQwuVfkYPGguH/xcxX5LXCa8xrM9IuNSoimkLJsPIxAhPI57BwMBAjCldphR8fx4Eq1KWUChSsPWv7RjjMw5rdvwBYxMjAIBrjaro3Od7lDAzQfSLV1g27w/8Onoa5q+ara5do6+Y5IoyQRAgk8nUPYx8sXXddjRv3wzN2ngBAIaO9cXp42exf8cB/NC7k5pHp37MT+6KSo5q1a2JWnVr5hijpaUFUzPTHGPOnDiL86cuYuLMX3D2xDmVZV36fK/y/tsu7XD+1EUcP3JS8kXZh2cER00aju88O+POjTuoWt0ZmpqamXJz4uhJNPymPnSL6wIAIh4+xuv41+g5sDvMLUsCALr364L+P/ggKvIFStlYQ7e4rhgPAPdu38ej+xEYOta3gPcw/21cvQUlLUpiVMBwsc2qlKVKTJPmqmdJBw7vj307DuD+nQeoXssVANCha3txuYWVBb7v9R0CRvyKtNQ0FNOS3I9Q+sJJbk6ZXC7HjRs31D2Mz5aamorbN++KH2wA0NDQQPVarrh+5ab6BiYRzE/umCNVl85fwXeendH7236YP20hEuISVJbHvorF3CkL8POvIyDXkeepz6TEJBgYGuQeKDFJiUkAkO3Yb9+4g3u37qNZWy+xzaZsaRgaGWLfjv1ITU2FIlmBvTsOoIydDSytLLLsZ2/wfpQuWwrO1ark/04UsLDQU3BwqoDJo6fhO8/OGNjFF3u27cs2PjU1FXu27YWevh7sK9hlGZMQ/xqH9x6BU9VKLMioQKjtqBo+fHiW7enp6Zg+fTpKlHh3vX7OnDmFOax8Ex+XAGW6EiYlTFTaTUoY4/HDx2oalXQwP7ljjv5Ts44b6jWpAytrCzx78hwrF63B2CETMH/VbGhqakIQBMwMmINWHVqgopMDIp9F5drnsQOhuH39Nob9MrgQ9iD/KJVKLJm1FJVdnGBX3jbLmH3B74qtyi5OYltxveKYtWw6Akb8ir9WbAAAlLKxRuCiX6FZTDNTHymKFBzeewTf9/quQPajoD1/GoldW/5Bh67t0aXP97h1/TYWzQpCMa1i8GrtKcadCj2NqWN/gyJZAVMzU/y2eCqM/n/pMsPyBSuxc+MuJCcrUMnZEVPmBRTy3lBRobaibN68eXBxcYGxsbFKuyAIuHHjBvT09PJ0GVOhUEChUKi0yeVyyOV5+02ZiKSvsXdD8Wu7CnYoV8EOPdr2xaXzV1C9liuCN+zEm6S3eb6kG372EmZNmgu/cUNha1+2oIZdIH6fvhgP7z3C3D9mZblckazA4X1H0fXHzpna50yeh8ouThg77WekK5XYsnYrxg0NwMI/52U6u3j8yEm8SXoLr1ae+BIJSgEOThXQ17cXAKC8oz0e3n2E3Vv3qBRlLjVdEPT3QsTHJWDv9n2YMiYQC9bMVZmz2al7BzRv64Wo5y+wdtl6/DZhNqbMD/hqptqQdKjt8uW0adMQHx+P8ePH48iRI+JLU1MTq1evxpEjR3D48OFc+wkMDISRkZHKKzAwsBD2IGdGxobQ0NRA7KtYlfbYV3EwyWVeTFHA/OSOOcqeVWkrGBkb4tnjZwDeFVk3rtxEC4+28K7VCj3b9QUA+HQfihkTVCdkXzp/BeP9JmHgiP74plXTQh/75/j9t8U4ffwMZi6djpIWZlnGhB46DkWyItO+Hd53FJHPX7y7+7CyA5ycHeE/dTQin0bi5LFTmfrZG7wftevXynSm9kthamaCMnY2Km1l7GzwIvKlSpuurg5K2VjDydkRIyYMg4amJvYF71eJMTIxQumypeFWuzp+CRyDMyfO4kYRnEJABU9tRdmYMWOwceNGDBo0CCNHjkRqauon9ePv74/4+HiVl7+/fz6P9uNpaWnBwbE8Lp69JLYplUpcPBsOp/duty6qmJ/cMUfZexkVjYT41+Lkdp9RAxH090IErX/3mjp/MgBgXOAY9PbpKa536dxljBs6ET8O6Y2W3zZXy9g/hSAI+P23xThxJAwzggIzTVh/374dB+DR0F28ezCDIlkBDZlM5eyOhkwDkMkgKJUqsc+fRuLSucsqc9K+NJVdnPDk0VOVticRT2FhZZ7jeoJSmePPo4xcpaZ82s8sopyodaZizZo1cf78efj4+KBGjRr466+/Pvp0sJQvVXbo1h4zJs6BQ6UKqFjFAdvX70DyWwW823yj7qFJAvOTu6KSo7dv3uLp/896Ae+eu3X31j0YGhrAwMgAa5etR72mdWFawgTPnjzHivkrYW1jhRoebgAA8w9+0GbcQWhV2ko8oxR+9hLGDwtAu85tUb9JXcRExwAAimlpwdBI2pP9f5++GIf3HcWkORNQvLiuOHY9fT2Vy45PHz/DlQtXMXXBpEx9VHevhmXz/8Dv0xej7Q+tISgFbFi9CZqamnCp4aISu3/HAZiamaLm/x+n8SXq0LU9hvYegfUrN6LhN/Vx6+ot7Nm2F8N+GQIAePs2Gev/2ACPhrVRwswE8XEJ2LlpN6JfvkIDz3fPdbtx5SZuXb+DKq5OMDDUx7PHz7E6aC2sS1uhUtVK6tw9+kqp/fYRfX19rFmzBhs2bICnpyfS09PVPaR808irIeJiE7AmaC1iX8XC3qEcpv0++Yu9HJDfmJ/cFZUc3b5+ByMHjBHfB81ZDgD4ppUnhvr74P6dBwjZfRCJr5NQoqQp3GpXR69B3aGdzbPKsnJg9yEkJyuwYdUmbFi1SWyv6uaM2ct+y7+dKQC7tvwDABjZ/2eV9pET/VQK9H07DsDM3Axutatn6qOMnQ1+nTsRa5etx9BeI6ChIYN9RXtMW/grSpT873K4UqnEgd0H4dXaE5qamW8A+FJUrOyAgFnj8MfC1Vi3fD0srS0xaMQA8WHBmhoaePzwCUJ2T0VCXDwMjAxRsbID5q6YKc4z1NGR48ThE/hz6Tokv01GCTNT1PBwQ9fpP3zUsUeUVzJBEAR1DyLDkydPcP78eXh6ekJPT++z+4tIvJcPo/r6lNG3F79mjjJjfnLG/OSOOcoZ85Oz9/NTGA493VNo22paqkXuQUWY2s+Uva906dIoXbq0uodBREREVOgk9/BYIiIioqKIRRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUERERkeQEBARAJpOpvBwdHcXlycnJ8PHxQYkSJaCvr48OHTogKipKpY+IiAi0bNkSxYsXh7m5OUaNGoW0tDSVmKNHj6J69eqQy+UoX748Vq9enWksixYtgq2tLXR0dODu7o4zZ84UyD6zKCMiIiJJqly5Mp4/fy6+jh8/Li7z8/PDrl27sHnzZhw7dgzPnj3Dt99+Ky5PT09Hy5YtkZKSgpMnT2LNmjVYvXo1JkyYIMY8ePAALVu2ROPGjREeHo5hw4bhxx9/xP79+8WYjRs3Yvjw4Zg4cSIuXLgAFxcXeHt748WLF/m+vyzKiIiISJKKFSsGS0tL8WVmZgYAiI+Pxx9//IE5c+agSZMmcHNzw6pVq3Dy5EmcOnUKAHDgwAFcv34d69atg6urK5o3b45ff/0VixYtQkpKCgAgKCgIdnZ2mD17NipVqgRfX1907NgRc+fOFccwZ84c9OvXD71794aTkxOCgoJQvHhxrFy5Mt/3l0UZERERFQqFQoGEhASVl0KhyDb+zp07sLa2Rrly5dC1a1dEREQAAM6fP4/U1FR4enqKsY6OjihTpgzCwsIAAGFhYXB2doaFhYUY4+3tjYSEBFy7dk2Meb+PjJiMPlJSUnD+/HmVGA0NDXh6eoox+YlFGRERERWKwMBAGBkZqbwCAwOzjHV3d8fq1auxb98+LFmyBA8ePED9+vXx+vVrREZGQltbG8bGxirrWFhYIDIyEgAQGRmpUpBlLM9YllNMQkIC3r59i+joaKSnp2cZk9FHfiqW7z0SERERZcHf3x/Dhw9XaZPL5VnGNm/eXPy6atWqcHd3R9myZbFp0ybo6uoW6DjVhWfKiIiIqFDI5XIYGhqqvLIryj5kbGwMBwcH3L17F5aWlkhJSUFcXJxKTFRUFCwtLQEAlpaWme7GzHifW4yhoSF0dXVhZmYGTU3NLGMy+shPLMqIiIhI8hITE3Hv3j1YWVnBzc0NWlpaOHTokLj81q1biIiIgIeHBwDAw8MDV65cUblLMiQkBIaGhnBychJj3u8jIyajD21tbbi5uanEKJVKHDp0SIzJTyzKiIiISHJGjhyJY8eO4eHDhzh58iTat28PTU1NdO7cGUZGRujbty+GDx+OI0eO4Pz58+jduzc8PDxQu3ZtAICXlxecnJzQvXt3XLp0Cfv378e4cePg4+Mjnp0bOHAg7t+/j9GjR+PmzZtYvHgxNm3aBD8/P3Ecw4cPx/Lly7FmzRrcuHEDgwYNQlJSEnr37p3v+8w5ZURERCQ5T548QefOnfHq1SuULFkS9erVw6lTp1CyZEkAwNy5c6GhoYEOHTpAoVDA29sbixcvFtfX1NTE7t27MWjQIHh4eEBPTw89e/bE5MmTxRg7Ozv8888/8PPzw/z581G6dGmsWLEC3t7eYsz333+Ply9fYsKECYiMjISrqyv27duXafJ/fpAJgiDke68SEZF4T91DkKQy+vbi18xRZsxPzpif3DFHOWN+cvZ+fgrDoad7Cm1bTUu1KLRtfYl4+ZKIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJOCrvvuSiIiIcsa7L6WDZ8qIiIiIJIBFGREREZEEfNVP9OdDCbPGBzfmjPnJGfOTO+YoZ8xPzgr74bEkHTxTRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBERERFJAIsyIiIiIglgUUZEREQkASzKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgkgEUZERERkQSwKCMiIiKSABZlRERERBLAooyIiIhIAliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEFFP3AL52OzbtwuY/tyLmVSzsK9jBZ/QgOFapqO5hSQbzkzvmKGdFOT/RL6KxYsEqnDl5DopkBaxLW2FkgB8qOjmIMY8eRGDFglW4fP4KlOnpKFOuDCbO+AXmVuZIiH+NP5euw/lTF/Ai8iWMjI1Qt5EHeg3qDj0DPTXu2cfr1qoXop6/yNTe+ruW6DWoR57289a121jx+yrcuXEXMpkMFSs7oN/QPrB3KAcA+HPpOqxdtj7TNnR05Nh1YnvB7RwVGSzKCtDRA8ewdM5yDBnri0pVHLFtfTD8fcdj5bZlMDE1Vvfw1I75yR1zlLOinJ/XCa8xrM9IuNSoimkLJsPIxAhPI57BwMBAjHn2+Dn8+o5C87Ze6DmgG4rrFcfD+4+gJdcGALx6+QqvXr5C/2E/oqxdGUQ9j8L8wIV4Ff0KE2b8oq5d+yQL186HMj1dfP/w3iP8/NMvaOhZP0/7+fbNW/gPHg+PBu4YMsYH6enp+HPpOvj7jsf6f9agmFYxfNe9A1p1aKGy3dGDxsLhvSKY6HPw8mUB2rpuO5q3b4ZmbbxQtlwZDB3rC7mOHPt3HFD30CSB+ckdc5Szopyfjau3oKRFSYwKGA7HKhVhVcoSNTyqw9rGSoxZtXgNatWtgX5D+6K8oz2sbaxQp2FtsWC1K2+LiTPHwaOBO6xtrFCtlit6/9QTp0JPIz0tPZstS5OxiRFMzUzF16l/z8C6tBWqujnnaT8jHj7G6/jX6DmwO2xsS8PWviy69+uC2FexiIp8dwZOt7iuyjZiY+Lw6H4EmrfzUueu01eERVkBSU1Nxe2bd1G9lqvYpqGhgeq1XHH9yk31DUwimJ/cMUc5K+r5CQs9BQenCpg8ehq+8+yMgV18sWfbPnG5UqnE6eNnUbpMKYzxGYfvPDtjcI9hOHHkZI79JiUmobhecWgW0yzoXSgwqampOLTnCLzbekEmk2UZ8+F+2pQtDUMjQ+zbsR+pqalQJCuwd8cBlLGzgaWVRZZ97A3ej9JlS8G5WpUC2xcqWliUFZD4uAQo05UwKWGi0m5Swhix0TFqGpV0MD+5Y45yVtTz8/xpJHZt+QelylgjcOEUtO7YEotmBeHAroMAgLiYOLx98xYbV29GzTpuCFw0BXUb18GkUVNx6fyVLPuMj43HXyv+RotvmxfmruS7k0fCkJiYCK/Wnlkuz2o/i+sVx6xl03FozxG0qtMebep3wLmT5zHt98lZFqgpihQc3nsEzdryLBnlH0nNKUtKSsKmTZtw9+5dWFlZoXPnzihRokSO6ygUCigUCpU2uVwOuVxekEMlIlIrQSnAwakC+vr2AgCUd7THw7uPsHvrHni19oRSEAAAHg1ro0PX9u9iKtrj2uUb2L11D1zcnFX6S0p8g3FDJ6JsuTLo0b9roe5Lftu74wBq1akBs5KZf35kt5+KZAXmTJ6Hyi5OGDvtZ6QrldiydivGDQ3Awj/nQa6j+jPl+JGTeJP0Fl6tsi78viQVjIrGjTFfArWeKXNyckJMzLvfaB8/fowqVarAz88PISEhmDhxIpycnPDgwYMc+wgMDISRkZHKKzAwsDCGnyMjY0NoaGog9lWsSnvsqziYmJmqaVTSwfzkjjnKWVHPj6mZCcrY2ai0lbGzwYvIlwDe5UdTUxNly5XJIkb1LsU3SW8wdvB46OoVR8Cs8SimJanf1z9K1PMoXDwTjubtvDMty2k/D+87isjnL97dvVrZAU7OjvCfOhqRTyNx8tipTH3tDd6P2vVrZTpTS/Q51FqU3bx5E2lpaQAAf39/WFtb49GjRzhz5gwePXqEqlWr4pdfcr4DyN/fH/Hx8Sovf3//whh+jrS0tODgWB4Xz14S25RKJS6eDYeTs6MaRyYNzE/umKOcFfX8VHZxwpNHT1XankQ8hYWVOYB3+alY2QGPHz1RiXn66CksLM3F90mJbzDGZxyKaRXD5DkToP3/OzO/VPt3hsDYxAju9WqptOe2n4pkBTRkMpU5aBoyDUAmg6BUqsQ+fxqJS+cu89Il5TvJzCkLCwtDQEAAjIyMAAD6+vqYNGkSjh8/nuN6crkchoaGKi+pXLrs0K099mzfhwO7DuLRgwgsCFyE5LcKeLf5Rt1DkwTmJ3fMUc6Kcn46dG2PG1duYv3KjXj6+BkO7z2CPdv2os13rcSY77p3wLED/2LPtn14+vgZgjfuQti/p8WYd4XKL0h+m4wR44fhTdIbxETHICY6BunpX9bdl8C7onz/zhB808pTZR5YXvazuns1vH6diN+nL8ajBxF4eO8RZk6aA01NTbjUcFHZzv4dB2BqZoqadWsU6v7R10/t56gzfitJTk6GlZWVyrJSpUrh5cuX6hhWvmjk1RBxsQlYE7QWsa9iYe9QDtN+n8zT3f/H/OSOOcpZUc5PxcoOCJg1Dn8sXI11y9fD0toSg0YMQNMWjcWYek3qYOhYX/y9ahMWzQpC6bKlMXHGL6hSrTIA4O7Nu7h59RYAoGe7vir9r921CpbWWd91KFUXTofjReRLNGurWpTnZT/L2Nng17kTsXbZegztNQIaGjLYV7THtIW/okTJ/y6HK5VKHNh9EF6tPaGp+eXeoUrSJBOE/88GVQMNDQ1UqVIFxYoVw507d7B69Wp06NBBXB4aGoouXbrgyZMnOfSSvYjEe/k11K9KGX178WvmKDPmJ2fMT+6Yo5wxPzl7Pz+FoTD/Dwp73740aj1TNnHiRJX3+vr6Ku937dqF+vXrF+aQiIiIiNRCUkXZh2bOnFlIIyEiIiJSL8lM9CciIiIqyliUEREREUkAizIiIiIiCWBRRkRERCQBLMqIiIiIJIBFGREREZEEsCgjIiIikgAWZUREREQSwKKMiIiISAJYlBEREZHkBAYGombNmjAwMIC5uTnatWuHW7duqcQ0atQIMplM5TVw4ECVmIiICLRs2RLFixeHubk5Ro0ahbS0NJWYo0ePonr16pDL5ShfvjxWr16daTyLFi2Cra0tdHR04O7ujjNnzuT7PrMoIyIiIsk5duwYfHx8cOrUKYSEhCA1NRVeXl5ISkpSievXrx+eP38uvmbMmCEuS09PR8uWLZGSkoKTJ09izZo1WL16NSZMmCDGPHjwAC1btkTjxo0RHh6OYcOG4ccff8T+/fvFmI0bN2L48OGYOHEiLly4ABcXF3h7e+PFixf5us8yQRCEfO1RQgrzL99/Scro24tfM0eZMT85Y35yxxzljPnJ2fv5KQyF+X/wOfv28uVLmJub49ixY2jQoAGAd2fKXF1dMW/evCzX2bt3L1q1aoVnz57BwsICABAUFISff/4ZL1++hLa2Nn7++Wf8888/uHr1qrjeDz/8gLi4OOzbtw8A4O7ujpo1a2LhwoUAAKVSCRsbGwwePBhjxoz55H36EM+UERERUaFQKBRISEhQeSkUijytGx8fDwAwNTVVaf/rr79gZmaGKlWqwN/fH2/evBGXhYWFwdnZWSzIAMDb2xsJCQm4du2aGOPp6anSp7e3N8LCwgAAKSkpOH/+vEqMhoYGPD09xZj8wqKMiIiICkVgYCCMjIxUXoGBgbmup1QqMWzYMNStWxdVqlQR27t06YJ169bhyJEj8Pf3x9q1a9GtWzdxeWRkpEpBBkB8HxkZmWNMQkIC3r59i+joaKSnp2cZk9FHfimWr70RERERZcPf3x/Dhw9XaZPL5bmu5+Pjg6tXr+L48eMq7f379xe/dnZ2hpWVFZo2bYp79+7B3r5wLwPnBxZlREREVCjkcnmeirD3+fr6Yvfu3QgNDUXp0qVzjHV3dwcA3L17F/b29rC0tMx0l2RUVBQAwNLSUvw3o+39GENDQ+jq6kJTUxOamppZxmT0kV94+ZKIiIgkRxAE+Pr6Yvv27Th8+DDs7OxyXSc8PBwAYGVlBQDw8PDAlStXVO6SDAkJgaGhIZycnMSYQ4cOqfQTEhICDw8PAIC2tjbc3NxUYpRKJQ4dOiTG5BeeKSMiIiLJ8fHxwfr167Fjxw4YGBiI87eMjIygq6uLe/fuYf369WjRogVKlCiBy5cvw8/PDw0aNEDVqlUBAF5eXnByckL37t0xY8YMREZGYty4cfDx8RHP2A0cOBALFy7E6NGj0adPHxw+fBibNm3CP//8I45l+PDh6NmzJ2rUqIFatWph3rx5SEpKQu/evfN1n1mUERERkeQsWbIEwLvHXrxv1apV6NWrF7S1tXHw4EGxQLKxsUGHDh0wbtw4MVZTUxO7d+/GoEGD4OHhAT09PfTs2ROTJ08WY+zs7PDPP//Az88P8+fPR+nSpbFixQp4e3uLMd9//z1evnyJCRMmIDIyEq6urti3b1+myf+fi88pK4L4jKCcMT85Y35yxxzljPnJGZ9TVnRxThkRERGRBLAoIyIiIpIAFmVEREREEsCijIiIiEgCWJQRERERSQCLMiIiIiIJYFFGREREJAEsyoiIiIgk4Kt+eCwRERHljA+PlQ6eKStgCoUCAQEBUCgU6h6KZDFHOWN+csb85I45yhnzQ1LBM2UFLCEhAUZGRoiPj4ehoaG6hyNJzFHOmJ+cMT+5Y45yVtTzwzNl0sEzZUREREQSwKKMiIiISAJYlBERERFJAIuyAiaXyzFx4kTI5XJ1D0WymKOcMT85Y35yxxzljPkhqeBEfyIioiKME/2lg2fKiIiIiCSARRkRERGRBLAoIyIiIpIAFmVEREREEsCirIAtWrQItra20NHRgbu7O86cOaPuIUlGaGgoWrduDWtra8hkMgQHB6t7SJISGBiImjVrwsDAAObm5mjXrh1u3bql7mFJxpIlS1C1alUYGhrC0NAQHh4e2Lt3r7qHJVnTp0+HTCbDsGHD1D0UyQgICIBMJlN5OTo6qntYVISxKCtAGzduxPDhwzFx4kRcuHABLi4u8Pb2xosXL9Q9NElISkqCi4sLFi1apO6hSNKxY8fg4+ODU6dOISQkBKmpqfDy8kJSUpK6hyYJpUuXxvTp03H+/HmcO3cOTZo0Qdu2bXHt2jV1D01yzp49i6VLl6Jq1arqHorkVK5cGc+fPxdfx48fV/eQqAjjIzEKkLu7O2rWrImFCxcCAJRKJWxsbDB48GCMGTNGzaOTFplMhu3bt6Ndu3bqHopkvXz5Eubm5jh27BgaNGig7uFIkqmpKWbOnIm+ffuqeyiSkZiYiOrVq2Px4sWYMmUKXF1dMW/ePHUPSxICAgIQHByM8PBwdQ9FrfhIDOngmbICkpKSgvPnz8PT01Ns09DQgKenJ8LCwtQ4MvpSxcfHA3hXeJCq9PR0bNiwAUlJSfDw8FD3cCTFx8cHLVu2VPleRP+5c+cOrK2tUa5cOXTt2hURERHqHhIVYcXUPYCvVXR0NNLT02FhYaHSbmFhgZs3b6ppVPSlUiqVGDZsGOrWrYsqVaqoeziSceXKFXh4eCA5ORn6+vrYvn07nJyc1D0sydiwYQMuXLiAs2fPqnsokuTu7o7Vq1ejYsWKeP78OSZNmoT69evj6tWrMDAwUPfwqAhiUUb0BfDx8cHVq1c53+UDFStWRHh4OOLj47Flyxb07NkTx44dY2EG4PHjxxg6dChCQkKgo6Oj7uFIUvPmzcWvq1atCnd3d5QtWxabNm3iJXBSCxZlBcTMzAyampqIiopSaY+KioKlpaWaRkVfIl9fX+zevRuhoaEoXbq0uocjKdra2ihfvjwAwM3NDWfPnsX8+fOxdOlSNY9M/c6fP48XL16gevXqYlt6ejpCQ0OxcOFCKBQKaGpqqnGE0mNsbAwHBwfcvXtX3UOhIopzygqItrY23NzccOjQIbFNqVTi0KFDnPNCeSIIAnx9fbF9+3YcPnwYdnZ26h6S5CmVSigUCnUPQxKaNm2KK1euIDw8XHzVqFEDXbt2RXh4OAuyLCQmJuLevXuwsrJS91CoiOKZsgI0fPhw9OzZEzVq1ECtWrUwb948JCUloXfv3uoemiQkJiaq/Eb64MEDhIeHw9TUFGXKlFHjyKTBx8cH69evx44dO2BgYIDIyEgAgJGREXR1ddU8OvXz9/dH8+bNUaZMGbx+/Rrr16/H0aNHsX//fnUPTRIMDAwyzT/U09NDiRIlOC/x/0aOHInWrVujbNmyePbsGSZOnAhNTU107vy/9u43tMryj+P4+6TttLbZMudM2dQc2QSbqTUGkQ5m7olYIxL741YZlGbm6d/2IJ1JTh8I9o9NsDajFCVryBSHDWYaWaQYEbpyKCpMnUXBDDfdsQfR+f1Opj9/0Xbube8X7MG57utc1/fcMPhwXRf3PTfRpWmAMpT1oDlz5tDe3s7SpUs5deoUkyZNYufOnZcd/h+ovvnmGwoLC2OfI5EIAKWlpdTV1SWoquCorq4GYPr06XHttbW1lJWV9X5BAXPmzBnmzZtHW1sbN910E3feeSeNjY3MmDEj0aWpjzh58iRz587lp59+IiMjg3vvvZd9+/aRkZGR6NI0QPmcMkmSBjCfUxYcnimTJEkKAEOZJElSABjKJEmSAsBQJkmSFACGMkmSpAAwlEmSJAWAoUySJCkADGWSJEkBYCiTdFVlZWU88MADsc/Tp0/nhRde6PU6mpubCYVC/PLLL70+tyT1BkOZ1EeVlZURCoUIhUIkJSWRk5PD66+/zsWLF3t03k8++YQVK1ZcU1+DlCRdO999KfVhxcXF1NbW0tnZyY4dO1i4cCHXX389FRUVcf26urpISkr6V+YcOnTovzKOJCmeK2VSHxYOhxkxYgSjR4/m2WefpaioiG3btsW2HN944w1GjhzJ+PHjAThx4gQPP/ww6enpDB06lNmzZ3Ps2LHYeN3d3UQiEdLT07nlllt45ZVX+Ovrcf+6fdnZ2cmrr75KVlYW4XCYnJwc3nvvPY4dOxZ74fzNN99MKBSKvUg9Go1SVVXF2LFjSU5OJi8vj48//jhunh07dnD77beTnJxMYWFhXJ2S1B8ZyqR+JDk5ma6uLgCamppoaWlh165dNDQ0cOHCBWbOnElaWhp79uzhiy++IDU1leLi4th31qxZQ11dHe+//z579+7l559/5tNPP73qnPPmzWPTpk289dZbHDp0iHXr1pGamkpWVhZbt24FoKWlhba2Nt58800Aqqqq+OCDD6ipqeH7779nyZIlPPbYY+zevRv4IzyWlJQwa9YsDh48yPz58ykvL++p2yZJgeD2pdQPXLp0iaamJhobG1m0aBHt7e2kpKSwfv362Lblhx9+SDQaZf369YRCIQBqa2tJT0+nubmZ+++/n7Vr11JRUUFJSQkANTU1NDY2XnHeH374gS1btrBr1y6KiooAuO2222LX/9zqHD58OOnp6cAfK2srV67ks88+o6CgIPadvXv3sm7dOqZNm0Z1dTXjxo1jzZo1AIwfP57vvvuO1atX/4t3TZKCxVAm9WENDQ2kpqZy4cIFotEojzzyCJWVlSxcuJCJEyfGnSP79ttvOXLkCGlpaXFjnD9/ntbWVn799Vfa2trIz8+PXRs8eDBTp069bAvzTwcPHmTQoEFMmzbtmms+cuQIv/32GzNmzIhr7+rq4q677gLg0KFDcXUAsQAnSf2VoUzqwwoLC6muriYpKYmRI0cyePB//qVTUlLi+nZ0dDBlyhQ++uijy8bJyMj4R/MnJyf/39/p6OgAYPv27YwaNSruWjgc/kd1SFJ/YCiT+rCUlBRycnKuqe/kyZPZvHkzw4cPZ8iQIX/b59Zbb+Wrr77ivvvuA+DixYvs37+fyZMn/23/iRMnEo1G2b17d2z78r/9uVLX3d0da5swYQLhcJjjx49fcYUtNzeXbdu2xbXt27fvf/9ISerDPOgvDRCPPvoow4YNY/bs2ezZs4ejR4/S3NzM888/z8mTJwFYvHgxq1ator6+nsOHD7NgwYKrPmNszJgxlJaW8uSTT1JfXx8bc8uWLQCMHj2aUChEQ0MD7e3tdHR0kJaWxksvvcSSJUvYsGEDra2tHDhwgLfffpsNGzYA8Mwzz/Djjz/y8ssv09LSwsaNG6mrq+vpWyRJCWUokwaIG2+8kc8//5zs7GxKSkrIzc3lqaee4vz587GVsxdffJHHH3+c0tJSCgoKSEtL48EHH7zquNXV1Tz00EMsWLCAO+64g6effppz584BMGrUKJYvX055eTmZmZk899xzAKxYsYLXXnuNqqoqcnNzKS4uZvv27YwdOxaA7Oxstm7dSn19PXl5edTU1LBy5coevDuSlHihS1c6wStJkvq94x2tvTZXduq4XpurL3KlTJIkKQAMZZIkSQFgKJMkSQoAQ5kkSVIAGMokSZICwFAmSZIUAIYySZIUWO+++y5jxozhhhtuID8/n6+//jrRJfUYQ5kkSQqkzZs3E4lEWLZsGQcOHCAvL4+ZM2dy5syZRJfWI3x4rCRJA1iQHx6bn5/P3XffzTvvvANANBolKyuLRYsWUV5e3hMlJpQrZZIkKXC6urrYv38/RUVFsbbrrruOoqIivvzyywRW1nMGJ7oASZI0MHR2dtLZ2RnXFg6HCYfDl/U9e/Ys3d3dZGZmxrVnZmZy+PDhHq0zUQxlkiQNYL35PsrKykqWL18e17Zs2TIqKyt7rYYgM5RJkqReUVFRQSQSiWv7u1UygGHDhjFo0CBOnz4d13769GlGjBjRYzUmkmfKJElSrwiHwwwZMiTu70qhLCkpiSlTptDU1BRri0ajNDU1UVBQ0Fsl9ypXyiRJUiBFIhFKS0uZOnUq99xzD2vXruXcuXM88cQTiS6tRxjKJElSIM2ZM4f29naWLl3KqVOnmDRpEjt37rzs8H9/4XPKJEmSAsAzZZIkSQFgKJMkSQoAQ5kkSVIAGMokSZICwFAmSZIUAIYySZKkADCUSZIkBYChTJIkKQAMZZIkSQFgKJMkSQoAQ5kkSVIAGMokSZIC4HevuxR483MMowAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 700x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# test\n",
    "y_pred = clf.predict(X_test).round()\n",
    "print(\"Default Decision Tree Model Description\")\n",
    "print(\"Depth: %d\" % clf.get_depth())\n",
    "print(\"Number of Leaves: %d\" % clf.get_n_leaves())\n",
    "print(\"*******************************************\")\n",
    "print(\"Classification Report of the default model\")\n",
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "#vizualizing classification report\n",
    "cmap =sns.color_palette(\"Greens\")\n",
    "cm = confusion_matrix(y_test, y_pred)\n",
    "plt.figure(figsize=(7,7))\n",
    "sns.heatmap(cm, annot=True, fmt='g', cmap=cmap, \n",
    "            annot_kws={'fontsize':10}, linewidth=1, square=True)\n",
    "plt.xlabel(\"Predicted\")\n",
    "plt.ylabel(\"Actual\")\n",
    "plt.title('Decision Tree Default Parameter Model Classification')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "137bd5ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "# saving the values in variables to export them later\n",
    "report_default = classification_report(y_test, y_pred,output_dict=True)\n",
    "macro_f1_default = report_default['macro avg']['f1-score']\n",
    "weighted_f1_default = report_default['weighted avg']['f1-score']\n",
    "class_0_default = report_default['0']['f1-score']\n",
    "class_1_default = report_default['1']['f1-score']\n",
    "class_2_default = report_default['2']['f1-score']\n",
    "class_3_default = report_default['3']['f1-score']\n",
    "class_4_default = report_default['4']['f1-score']\n",
    "class_5_default = report_default['5']['f1-score']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "2fce405d",
   "metadata": {},
   "source": [
    "#### 5. Hyper-parameter Tuning\n",
    "\n",
    "Now that the decision tree model has proven to produce acceptable performance, we will try to specify its hyper parameters to determine the combination that maxmizes the model's performance score.\n",
    "\n",
    "Scikit-Learn offers an exhuastive search algorithm enabling us to easily determine the best hyper parameter combination called <b><i>GridSearchCV</i></b>\n",
    "<br>The algorithm iterate over every combination possible of the input hyper parameter values and select the best one.\n",
    "\n",
    "The following are the values of hyperparameters tested:\n",
    "<ol>\n",
    "<li><b><i>max_depth</i></b>: \n",
    "<br> - The depth of the decision tree for the default model grew until depth = 52, thus we selected a range of numbers between 45 and 55 to include this value\n",
    "<li><b><i>min_samples_split</i></b>: \n",
    "<br> - range from 2 to 10: (2, 3, 4, 5, 6, 7, 8, 9, 10)\n",
    "<br> - range from 10 to 100: (10, 20, 30, 40, 50, 60, 70, 80, 90, 100)\n",
    "<li><b><i>criterion</i></b>: \n",
    "<br> - \"gini\" and \"entropy\"\n",
    "</ol>\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "597eeed1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameters for decision tree classification:  {'criterion': 'gini', 'max_depth': 47, 'min_samples_split': 90}\n"
     ]
    }
   ],
   "source": [
    "#specify the value of parameters of interest\n",
    "tree_params = {'criterion':['gini','entropy'],'max_depth': list(range(45, 56)),'min_samples_split':list(range(2, 10))+np.arange(10, 100, 10).tolist()}\n",
    "\n",
    "#determine the best hyper parameter\n",
    "dtc_top = GridSearchCV(tree.DecisionTreeClassifier(), tree_params, cv=5)\n",
    "\n",
    "# Training the model for classification with the same dataset \n",
    "model2 = dtc_top.fit(X_train, y_train)\n",
    "\n",
    "print(\"Best hyperparameters for decision tree classification: \", dtc_top.best_params_)\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1f36d286",
   "metadata": {},
   "source": [
    "#### 6. Best Performing Model\n",
    "\n",
    "The following are the selected hyperparameter values:\n",
    "<ol>\n",
    "<li><b><i>max_depth</i></b>: 47\n",
    "<li><b><i>min_samples_split</i></b>: 90\n",
    "<li><b><i>criterion</i></b>: gini\n",
    "</ol>\n",
    "\n",
    "With the best hyperparameter combination found in the previous step, we can now build our final decision tree classifier and measure its performance.\n",
    "<br>Using the <i>pickle</i> library, the model with the best hyperparameter is saved to the project file for future use.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7d9d4e3e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#saving the model with the best performing hyperparameters\n",
    "filename = \"models/DT_model.pickle\"\n",
    "\n",
    "#train the model with the best hyperparameters\n",
    "model_final = tree.DecisionTreeClassifier(criterion = \"gini\", max_depth = 47, min_samples_split = 90)\n",
    "model_final.fit(X_train, y_train)\n",
    "\n",
    "# save model\n",
    "pickle.dump(model_final, open(filename, \"wb\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "05f4b14e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report of the default model\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.79      0.85      0.82    231390\n",
      "           1       0.71      0.62      0.66    140418\n",
      "           2       1.00      1.00      1.00     34444\n",
      "           3       0.70      0.80      0.75      5421\n",
      "           4       0.59      0.47      0.53      4962\n",
      "           5       0.91      0.93      0.92      7949\n",
      "\n",
      "    accuracy                           0.78    424584\n",
      "   macro avg       0.78      0.78      0.78    424584\n",
      "weighted avg       0.78      0.78      0.78    424584\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 700x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Measure performance of the model with the best hyper parameter \n",
    "y_top_pred = model_final.predict(X_test)\n",
    "print(\"Classification Report of the default model\")\n",
    "print(classification_report(y_test, y_top_pred))\n",
    "\n",
    "#vizualizing classification report\n",
    "cmap =sns.color_palette(\"Greens\")\n",
    "cm = confusion_matrix(y_test, y_top_pred)\n",
    "plt.figure(figsize=(7,7))\n",
    "sns.heatmap(cm, annot=True, fmt='g', cmap=cmap, \n",
    "            annot_kws={'fontsize':10}, linewidth=1, square=True)\n",
    "plt.xlabel(\"Predicted\")\n",
    "plt.ylabel(\"Actual\")\n",
    "plt.title('Decision Tree Parameter Tuning Model Classification')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "cd40d95e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# saving the values in variables to export them later\n",
    "report_top = classification_report(y_test, y_top_pred,output_dict=True)\n",
    "macro_f1_top = report_top['macro avg']['f1-score']\n",
    "weighted_f1_top = report_top['weighted avg']['f1-score']\n",
    "class_0_top = report_top['0']['f1-score']\n",
    "class_1_top = report_top['1']['f1-score']\n",
    "class_2_top = report_top['2']['f1-score']\n",
    "class_3_top = report_top['3']['f1-score']\n",
    "class_4_top = report_top['4']['f1-score']\n",
    "class_5_top = report_top['5']['f1-score']\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "018ceffd",
   "metadata": {},
   "source": [
    "#### 7. Training and Testing the model with Undersampled Data\n",
    "\n",
    "After the hyper parameter tuning, the decision tree model's performance only slightly increased: its weighted f1-score changed from 0.77 to 0.78.\n",
    "<br> In an attempt to increase the performance of our model, we loaded an undersampled dataset to train and test the model.\n",
    "<br> The results of this section will let us know whether the imbalanced distribution of dataset has any correlation with the performance of our model.\n",
    "<br> This section uses the same methodology and libraries as the previous sections 3, 4, 5, and 6, but with <b><i>london_clean_undersample.csv</i></b> dataset.\n",
    "<br> For details, please refer to the previous sections.\n",
    "\n",
    "##### 7.1 Undersampled Dataset: Load and Split Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "5da39e9e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# load dataset, print types\n",
    "df1 = pd.read_csv('preprocessing/data/london_clean_undersample.csv')\n",
    "\n",
    "# do train and test split\n",
    "X = df1[['DateOfCall', 'PropertyType', 'NumPumpsAttending', 'PumpHoursRoundUp', 'mean_temp']]\n",
    "y = df1[['CostCat']]\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "a2ea8d9f",
   "metadata": {},
   "source": [
    "##### 7.2 Undersampled Dataset: Train and Test Model with Default Parameter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Default Decision Tree Model Description\n",
      "Depth: 42\n",
      "Number of Leaves: 27097\n",
      "*******************************************\n",
      "Classification Report of the default model\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.76      0.82      0.79     22940\n",
      "           1       0.66      0.57      0.61     14064\n",
      "           2       1.00      1.00      1.00     10240\n",
      "           3       0.73      0.75      0.74      5454\n",
      "           4       0.59      0.58      0.59      5179\n",
      "           5       0.92      0.91      0.92      8126\n",
      "\n",
      "    accuracy                           0.78     66003\n",
      "   macro avg       0.78      0.77      0.77     66003\n",
      "weighted avg       0.78      0.78      0.78     66003\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlwAAAI3CAYAAABOL4UpAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy89olMNAAAACXBIWXMAAA9hAAAPYQGoP6dpAACFIUlEQVR4nOzdd1gUV9sG8BsQFqQjXRERpSmiYgmJBRXBXqImtlhjC8bYDXmjoiZiL7FhiSUqMfbesGvELvaKIhaKUgXpO98ffkxc6cqwK96/69pL9syZM888zK4PZ2Zn1QRBEEBEREREklFXdgBEREREZR0LLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIiklg5ZQdAREREyqM1unapbStjXmipbUvVcIaLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuEpYeHg41NTUsHbt2mKt5+npCU9PT0liovxFR0eja9euqFChAtTU1LBgwQJJttOvXz9UqVJFkrGpbFFTU4O/v3+x1/vQ957CKPu9yd/fH2pqagptWVlZGD9+PGxsbKCuro5OnToB+PDcfSy+vqkoylzBtXbtWqipqYkPbW1tWFtbw8fHB3/88Qdev36t7BBVRpUqVRRyld+jpN/Ai+PEiRMKschkMlhYWMDT0xPTp0/Hy5cvP2r8UaNG4dChQ/Dz88P69evRqlWrEoq8YG/evIG/vz9OnDhRpP7v50FTUxNVq1ZFnz598OjRI2mDVbLbt2/D398f4eHhpbbNd99Hzpw5k2u5IAiwsbGBmpoa2rVrV2pxlaTo6GiMHTsWTk5OKF++PHR1deHu7o7ffvsNCQkJyg6vQKtXr8bs2bPRtWtXrFu3DqNGjZJ8my9evIC/vz9CQ0Ml3xaVTeWUHYBUpk6dCjs7O2RmZiIqKgonTpzAyJEjMW/ePOzevRu1atWSZLu2trZITU2FpqZmsdY7fPiwJPEUZMGCBUhOThaf79+/H3///Tfmz58PU1NTsf3LL78s9djeN2LECNSvXx/Z2dl4+fIlzp49i8mTJ2PevHnYvHkzmjdv/kHjHjt2DB07dsTYsWNLOOKCvXnzBlOmTAGAYs0e5OQhMzMTV65cwYoVK7Bv3z7cuHED1tbWEkWrXLdv38aUKVPg6elZ6rMI2traCAoKQqNGjRTaT548iWfPnkEmk5VqPCXl4sWLaNOmDZKTk9G7d2+4u7sDAC5duoQZM2bg1KlTSnlPysuvv/6Kn3/+WaHt2LFjqFixIubPn6/QnpqainLlpPlv7cWLF5gyZQqqVKmC2rVrKyxbuXIl5HK5JNv93J06dQqzZ8/G5cuXERkZiR07dogzmgByzX7mmDVrFsaNGwfg7eTCkydPFJYHBAQoHFfXr1+Hr68vLl68CDMzM/z4448YP368wjpbtmzBxIkTER4ejurVq2PmzJlo06ZNkfelzBZcrVu3Rr169cTnfn5+OHbsGNq1a4cOHTrgzp070NHRKfHt5syqFZeWllaJx1KYdw9aAIiKisLff/+NTp06FfgfW0pKCnR1daUN7j2NGzdG165dFdquXbsGb29vdOnSBbdv34aVlVWxx42JiYGRkVEJRSm9d/PQv39/ODg4YMSIEVi3bh38/Pw+eFxBEJCWlibJa0JVFeU4btOmDbZs2YI//vhD4T/yoKAguLu749WrV1KHWeISEhLQuXNnaGho4OrVq3ByclJY/vvvv2PlypVKii63cuXK5Sqi8nvdfsh7b0ko7h/YVHQpKSlwc3PDgAED8PXXX+daHhkZqfD8wIEDGDhwILp06aLQPnXqVAwaNEh8rq+vL/6clJQEb29veHl5ITAwEDdu3MCAAQNgZGSEwYMHAwDOnj2LHj16ICAgAO3atUNQUBA6deqEK1euoGbNmkXalzJ3SrEgzZs3x8SJE/HkyRNs2LBBYdndu3fRtWtXmJiYQFtbG/Xq1cPu3btzjZGQkIBRo0ahSpUqkMlkqFSpEvr06SO+8eZ1HUVUVBT69++PSpUqQSaTwcrKCh07dlQ4RZLXdRIxMTEYOHAgLCwsoK2tDTc3N6xbt06hT8725syZgxUrVsDe3h4ymQz169fHxYsXPy5heHttgp6eHsLCwtCmTRvo6+ujV69eAAC5XI4FCxagRo0a0NbWhoWFBYYMGYL4+Phc4xw4cACNGzeGrq4u9PX10bZtW9y6deujYnNzc8OCBQuQkJCAxYsXKyx7/vw5BgwYAAsLC8hkMtSoUQOrV68Wl+ecMhIEAUuWLBFPHwFAXFwcxo4dC1dXV+jp6cHAwACtW7fGtWvXFLaRM8b7p7pyTv/ld7owPDwcZmZmAIApU6aI2/6Qa09yZvYeP34MAFizZg2aN28Oc3NzyGQyuLi4YNmyZbnWq1KlCtq1a4dDhw6hXr160NHRwfLlyz9ojBMnTohjuLq6ivu9fft2uLq6QltbG+7u7rh69WquMQp73a1duxbdunUDADRr1kzM1bu5LcqxVdBxXJAePXogNjYWwcHBYltGRga2bt2Knj175rlOSkoKxowZAxsbG8hkMjg6OmLOnDkQBEGhX3p6OkaNGgUzMzPo6+ujQ4cOePbsWZ5jFnY8F8fy5cvx/PlzzJs3L1exBQAWFhb49ddf810/IyMDkyZNgru7OwwNDaGrq4vGjRvj+PHjufpu2rQJ7u7u0NfXh4GBAVxdXbFw4UJxeWZmJqZMmYLq1atDW1sbFSpUQKNGjRTy/e41XDnvd8ePH8etW7dyHQ95vY6eP3+OgQMHwtraGjKZDHZ2dhg2bBgyMjIAFO31fuLECdSvXx/A2z903r/cIq9ruIp6HKipqWH48OHYuXMnatasKf5+Dx48mO/v4HPSunVr/Pbbb+jcuXOeyy0tLRUeu3btQrNmzVC1alWFfvr6+gr93v1ja+PGjcjIyMDq1atRo0YNdO/eHSNGjMC8efPEPgsXLkSrVq0wbtw4ODs7Y9q0aahbt26u/3sK8lkVXADw3XffAVA8hXfr1i188cUXuHPnDn7++WfMnTsXurq66NSpE3bs2CH2S05ORuPGjbFo0SJ4e3tj4cKFGDp0KO7evZvvGyUAdOnSBTt27ED//v2xdOlSjBgxAq9fv0ZERES+66SmpsLT0xPr169Hr169MHv2bBgaGqJfv34Kb1g5goKCMHv2bAwZMgS//fYbwsPD8fXXXyMzM/ND0qQgKysLPj4+MDc3x5w5c8S/HIYMGYJx48bhq6++wsKFC9G/f39s3LgRPj4+Cttdv3492rZtCz09PcycORMTJ07E7du30ahRo4++Lqdr167Q0dFR+H1GR0fjiy++wJEjRzB8+HAsXLgQ1apVw8CBA8WL4ps0aYL169cDAFq2bIn169eLzx89eoSdO3eiXbt2mDdvHsaNG4cbN26gadOmePHixUfFCwBmZmZiAdO5c2dx23n99VaYsLAwAECFChUAAMuWLYOtrS1++eUXzJ07FzY2Nvjhhx+wZMmSXOveu3cPPXr0QMuWLbFw4ULxNElxxnj48CF69uyJ9u3bIyAgAPHx8Wjfvj02btyIUaNGoXfv3pgyZQrCwsLwzTffKJx2KcrrrkmTJhgxYgQA4JdffhFz5ezsDKB4x1Z+x3FBqlSpAg8PD/z9999i24EDB5CYmIju3bvn6i8IAjp06ID58+ejVatWmDdvHhwdHTFu3DiMHj1aoe/333+PBQsWwNvbGzNmzICmpibatm2ba8yiHM/FsXv3bujo6OSaMS6qpKQkrFq1Cp6enpg5cyb8/f3x8uVL+Pj4KFzfFBwcjB49esDY2BgzZ87EjBkz4OnpiX///Vfs4+/vjylTpqBZs2ZYvHgx/ve//6Fy5cq4cuVKnts2MzPD+vXr4eTkhEqVKuU6Ht734sULNGjQAJs2bcK3336LP/74A9999x1OnjyJN2/eACja693Z2RlTp04FAAwePFjcbpMmTfLcbnGOAwA4c+YMfvjhB3Tv3h2zZs1CWloaunTpgtjY2MJ/IZ+Y9PR0JCUlKTzS09NLZOzo6Gjs27cPAwcOzLVsxowZqFChAurUqYPZs2cjKytLXBYSEoImTZoonGny8fHBvXv3xAmEkJAQeHl5KYzp4+ODkJCQogcolDFr1qwRAAgXL17Mt4+hoaFQp04d8XmLFi0EV1dXIS0tTWyTy+XCl19+KVSvXl1smzRpkgBA2L59e64x5XK5IAiC8PjxYwGAsGbNGkEQBCE+Pl4AIMyePbvAuJs2bSo0bdpUfL5gwQIBgLBhwwaxLSMjQ/Dw8BD09PSEpKQkhe1VqFBBiIuLE/vu2rVLACDs2bOnwO2+a/bs2QIA4fHjx2Jb3759BQDCzz//rND39OnTAgBh48aNCu0HDx5UaH/9+rVgZGQkDBo0SKFfVFSUYGhomKv9fcePHxcACFu2bMm3j5ubm2BsbCw+HzhwoGBlZSW8evVKoV/37t0FQ0ND4c2bN2IbAMHX11ehX1pampCdna3Q9vjxY0EmkwlTp04V23KOtXfz9W7Mx48fF9v69u0r2Nrais9fvnwpABAmT56c737lNebq1auFly9fCi9evBD27dsnVKlSRVBTUxOP93f3LYePj49QtWpVhTZbW1sBgHDw4MFc/Ys7xtmzZ8W2Q4cOCQAEHR0d4cmTJ2L78uXLc+WkqK+7LVu25FpXEIp3bOV3HOfn3feRxYsXC/r6+mJeunXrJjRr1kzMQdu2bcX1du7cKQAQfvvtN4XxunbtKqipqQkPHz4UBEEQQkNDBQDCDz/8oNCvZ8+euY6Loh7P77/35MfY2Fhwc3MrUh4EIfd7U1ZWlpCenq7QJz4+XrCwsBAGDBggtv3000+CgYGBkJWVle/Ybm5uCvnLy+TJk4X3/6tq2rSpUKNGjVx9389dnz59BHV19Tz/P8h5zy7q6/3ixYv55vf913dRj4OcmLW0tBTarl27JgAQFi1alGtbUtAc5VZqj5zf57uPor4PAhB27NiR7/KZM2cKxsbGQmpqqkL73LlzhePHjwvXrl0Tli1bJhgZGQmjRo0Sl7ds2VIYPHiwwjq3bt0SAAi3b99+myNNTSEoKEihz5IlSwRzc/MixS4IgvDZzXABgJ6envhpxbi4OBw7dgzffPMNXr9+jVevXuHVq1eIjY2Fj48PHjx4gOfPnwMAtm3bBjc3tzynNvO7cE9HRwdaWlo4ceJEnqfa8rN//35YWlqiR48eYpumpiZGjBiB5ORknDx5UqH/t99+C2NjY/F548aNAaDEPsE2bNgwhedbtmyBoaEhWrZsKebs1atXcHd3h56ennh6ITg4GAkJCejRo4dCPw0NDTRs2DDP0xDF9e7vUxAEbNu2De3bt4cgCArb9PHxQWJiYr5/PeeQyWRQV3/70sjOzkZsbCz09PTg6OhY6LpSGzBgAMzMzGBtbY22bdsiJSUF69atE69XfPcarMTERLx69QpNmzbFo0ePkJiYqDCWnZ0dfHx8cm2jOGO4uLjAw8NDfN6wYUMAb091Vq5cOVd7zvFYnNddfj7k2Hr/OC6Kb775Bqmpqdi7dy9ev36NvXv35ns6cf/+/dDQ0BBn5XKMGTMGgiDgwIEDYj8AufqNHDlS4XlJHM/vS0pKUrh+pbg0NDTEmQC5XI64uDhkZWWhXr16CrEYGRkhJSVF4fTg+4yMjHDr1i08ePDgg+PJj1wux86dO9G+fXuF63lz5LxnS/F6L+pxkMPLywv29vbi81q1asHAwKBMfgLZz88PiYmJCo+Puf70XatXr0avXr1yXcs3evRoeHp6olatWhg6dCjmzp2LRYsWldjMWlGV2YvmC5KcnAxzc3MAb0+JCIKAiRMnYuLEiXn2j4mJQcWKFREWFlak0xDvkslkmDlzJsaMGQMLCwt88cUXaNeuHfr06QNLS8t813vy5AmqV68uvhHkyJk6f/8TF+/+5wZALL6KU+Tlp1y5cqhUqZJC24MHD5CYmCjm8X0xMTFiPwD5forQwMDgo+NLTk4W/wN5+fIlEhISsGLFCqxYsaLA2PIjl8uxcOFCLF26FI8fP0Z2dra4LOfUnbJMmjQJjRs3hoaGBkxNTeHs7KxwQfG///6LyZMnIyQkRDxlkiMxMRGGhobiczs7uzy3UZwx3j/ucpbZ2Njk2Z5zPBbndZef4h5beR3HRWFmZgYvLy8EBQXhzZs3yM7Ozvd03JMnT2BtbZ2roHn/dfvkyROoq6sr/CcLAI6OjgrPS+J4fp+BgcFH3x5n3bp1mDt3Lu7evatw+cC7x9QPP/yAzZs3o3Xr1qhYsSK8vb3xzTffKNx6ZerUqejYsSMcHBxQs2ZNtGrVCt99912JfIr85cuXSEpKKvSCZile70U9DnK8/zoC3r6Hl8T7t6qRyWSSfLr39OnTuHfvHv75559C+zZs2BBZWVkIDw+Ho6MjLC0tER0drdAn53nO/9P59Sno//H3fXYF17Nnz5CYmIhq1aoBgHhNydixY/P8ax+A2PdDjRw5Eu3bt8fOnTtx6NAhTJw4EQEBATh27Bjq1KnzUWPn0NDQyLNdeO8CzQ/x7l+AOeRyOczNzbFx48Y818m5KDwnv+vXr8/zwPzYj3BnZmbi/v374ptqzvZ69+6Nvn375rlOYW/m06dPx8SJEzFgwABMmzYNJiYmUFdXx8iRIxWuQcpvVvPdN+yS5urqmus6ghxhYWFo0aIFnJycMG/ePNjY2EBLSwv79+/H/Pnzc31sPa9PJBZ3jPyOu8KOx5J43RX32MrrOC6qnj17YtCgQYiKikLr1q1L7ZOtJXE8v8/JyQmhoaHIyMj4oE9Hb9iwAf369UOnTp0wbtw4mJubQ0NDAwEBAeI1hQBgbm6O0NBQHDp0CAcOHMCBAwewZs0a9OnTR/zwT5MmTRAWFoZdu3bh8OHDWLVqFebPn4/AwEB8//33xY7tQxT19S4lKd+/Pxd//vkn3N3d4ebmVmjf0NBQqKurixMGHh4e+N///ofMzEzxE6fBwcFwdHQUJy88PDxw9OhRhVno4OBghRn+wnx2BVfOhdE5b/I5n2TQ1NTM9z+yHPb29rh58+YHbdfe3h5jxozBmDFj8ODBA9SuXRtz587N9WnJHLa2trh+/TrkcrnCfxJ3794VlyuTvb09jhw5gq+++qrAWwnk/AVvbm5eaH4/xNatW5Gamir+PnM+8ZWdnf3B29u6dSuaNWuGP//8U6E9ISFB4f5kOS/E928S+f5fr3nJr1j7GHv27EF6ejp2796t8BdzcU7blsQYRVGc111+uZL62HpX586dMWTIEJw7d67Av6BtbW1x5MgRvH79WmF24/3Xra2tLeRyOcLCwhRmte7du6cwXkkcz+9r3749QkJCsG3bNoVLFopq69atqFq1KrZv367wu5k8eXKuvlpaWmjfvj3at28PuVyOH374AcuXL8fEiRPFgtrExAT9+/dH//79kZycjCZNmsDf3/+jCy4zMzMYGBgU+p5d1Nd7cV6zRT0OqHDJycl4+PCh+Pzx48cIDQ2FiYmJ+B6VlJSELVu2YO7cubnWDwkJwfnz59GsWTPo6+sjJCRE/EBPznt4z549MWXKFAwcOBATJkzAzZs3sXDhQoX7vP30009o2rQp5s6di7Zt22LTpk24dOlSvjPPefmsruE6duwYpk2bBjs7O/Ej4ebm5vD09MTy5ctz3c8DgMKdzLt06YJr164pfHIxR35/ibx58wZpaWkKbfb29tDX1y/w/HGbNm0QFRWl8OaelZWFRYsWQU9PD02bNi14ZyX2zTffIDs7G9OmTcu1LCsrSyxCfHx8YGBggOnTp+f5icmPuVP8tWvXMHLkSBgbG8PX1xfA278Uu3Tpgm3btuX5RluU7WloaOT6fW7ZsiXXNUU5/+GfOnVKbMvOzi7SC7B8+fIAchdrHyPnr+R3Y09MTMSaNWtKdYyiKM7rLufj2+/nSspj6316enpYtmwZ/P390b59+3z7tWnTBtnZ2bk+Kj5//nyoqamhdevWACD++8cffyj0e/9ThyVxPL9v6NChsLKywpgxY3D//v1cy2NiYvDbb7/lu35ex8j58+dzfVrr/U/Yqauri7NxOe997/fR09NDtWrVSuTampyv/NmzZw8uXbqUa3lO/EV9ved3HOalqMcBFe7SpUuoU6eOeDZo9OjRqFOnDiZNmiT22bRpEwRByPMPCJlMhk2bNqFp06aoUaMGfv/9d4waNUrhfdrQ0BCHDx/G48eP4e7ujjFjxmDSpEniPbiAtzcADwoKwooVK+Dm5oatW7eKt/IoqjI7w3XgwAHcvXsXWVlZiI6OxrFjxxAcHAxbW1vs3r1b4aK6JUuWoFGjRnB1dcWgQYNQtWpVREdHIyQkBM+ePRPvxzJu3Dhs3boV3bp1w4ABA+Du7o64uDjs3r0bgYGBeU5l3r9/Hy1atMA333wDFxcXlCtXDjt27EB0dHSeHyvPMXjwYCxfvhz9+vXD5cuXUaVKFWzduhX//vsvFixY8FEXvZaEpk2bYsiQIQgICEBoaCi8vb2hqamJBw8eYMuWLVi4cCG6du0KAwMDLFu2DN999x3q1q2L7t27w8zMDBEREdi3bx+++uqrIt3H5PTp00hLSxMvav3333+xe/duGBoaYseOHQqnlGbMmIHjx4+jYcOGGDRoEFxcXBAXF4crV67gyJEjiIuLK3Bb7dq1w9SpU9G/f398+eWXuHHjBjZu3Jjrvi41atTAF198AT8/P8TFxcHExASbNm1S+LhxfnR0dODi4oJ//vkHDg4OMDExQc2aNYv14n2ft7e3OKMwZMgQJCcnY+XKlTA3N8+zqJFqjKIq6uuudu3a0NDQwMyZM5GYmAiZTCbeJ6wkjq2iyu+U3rvat2+PZs2a4X//+x/Cw8Ph5uaGw4cPY9euXRg5cqRYpNeuXRs9evTA0qVLkZiYiC+//BJHjx5V+Es+x8cez+8zNjbGjh070KZNG9SuXVvhTvNXrlzB33//XeBpknbt2mH79u3o3Lkz2rZti8ePHyMwMBAuLi4K31zx/fffIy4uDs2bN0elSpXw5MkTLFq0CLVr1xavZXJxcYGnpyfc3d1hYmKCS5cuYevWrRg+fHix9ik/06dPx+HDh9G0aVMMHjwYzs7OiIyMxJYtW3DmzBkYGRkV+fVub28PIyMjBAYGQl9fH7q6umjYsGGe10IW9Tigwnl6ehZ6anXw4MEKxdG76tati3PnzhW6nVq1auH06dMF9unWrZt4X8APUuTPM34icj7OnfPQ0tISLC0thZYtWwoLFy4Ub6fwvrCwMKFPnz6CpaWloKmpKVSsWFFo166dsHXrVoV+sbGxwvDhw4WKFSsKWlpaQqVKlYS+ffuKH9l+/6PZr169Enx9fQUnJydBV1dXMDQ0FBo2bChs3rxZYdz3P3otCIIQHR0t9O/fXzA1NRW0tLQEV1fXXB9JztleXredQDE+bisI+d8WQldXN991VqxYIbi7uws6OjqCvr6+4OrqKowfP1548eKFQr/jx48LPj4+gqGhoaCtrS3Y29sL/fr1Ey5dulRgTDm3Q8h5aGpqCmZmZkKTJk2E33//XYiJiclzvejoaMHX11ewsbERNDU1BUtLS6FFixbCihUrFPohn9tCjBkzRrCyshJ0dHSEr776SggJCcnzdxQWFiZ4eXkJMplMsLCwEH755RchODi40NtCCIIgnD17VnB3dxe0tLQK/V0V5fYYgiAIu3fvFmrVqiVoa2sLVapUEWbOnCmsXr061+/1/dsZlOQYeeU0v+O0qK+7lStXClWrVhU0NDRy5bYox1Zhx/H7inJ7GUHIOwevX78WRo0aJVhbWwuamppC9erVhdmzZ4u3IciRmpoqjBgxQqhQoYKgq6srtG/fXnj69Gmex0JRjuei3hYix4sXL4RRo0YJDg4Ogra2tlC+fHnB3d1d+P3334XExESx3/vHvVwuF6ZPny7Y2toKMplMqFOnjrB3795cx/jWrVsFb29vwdzcXNDS0hIqV64sDBkyRIiMjBT7/Pbbb0KDBg0EIyMjQUdHR3BychJ+//13ISMjQ+zzMbeFEARBePLkidCnTx/BzMxMkMlkQtWqVQVfX1/x1hbFeb3v2rVLcHFxEcqVK6eQ67xe30U9DvJ6vQjC22Orb9++udqlUJq3hficqQkCr8ojIiL6XGmNrl1q28qYF1pq21I1n9U1XERERETKwIKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkVma/SxEo3bvnfkrevdPv0ef7lReIimpRsY34c0RymBIjUU2V9f77HjjmJ2/MUcGYn4K9mx8qOzjDRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEiun7ABUUaOqdTGmWV/UqeQMa0NzdF09CrtvHheX62rp4Pd2P6FDzWaooGuI8NjnWHz6b6wM2QoAsDW2xoOJ+/Mcu8e6cdh2LVh8/l39DhjZtDeqm9kiKS0F264F46ftAQWO02jhd7jw5EZJ7nKJOhR0BLtW7UOzr5ug2/DOAIAze8/i4tErePrgGdLepGPO7ukor6cjrnM/9CEWjF6S53jjl45CFafKiI2Kw8Se03ItH7f4J9i5VJFkX1TBrs17sOWvbYiLjYd9dTv4jh8Gp5qOyg5LZTA/BWN+CsccUWlgwZUHXS0dXH9xH2sv7MSW/vNzLZ/dcSw8q9dHv43/w5O4F/By9MCiLn6ITHqJvbdO4mlCFGwmt1BY53uPLhjt2RcH75wR235q2hsjPfvAb898XHhyA7paOrA1sc61PZ9lg3E7Kkx8HpuSWIJ7W7LC70bgzN4QVKyquB8ZaZlwqe8El/pO2LVqX671qtaogoCtUxTa9q4+gLtX78PW0UahfcScYbCqYik+1zPQLcE9UC0nDp/E8nkrMeKX4XCu6YTtQTvhN3wiVm9fAWMTI2WHp3TMT8GYn8IxR1RaeEoxD4fu/ovJB5Zg143jeS73qOKGDRf34FTYJTyJf4E/z23D9Rf3Ub9yTQCAXJAj+nWswqNjzebYeu0wUjJSAQBGOvqY0toXA4J+xaYrB/Ao9hluRD7A3lsnc20vLiVRYawseZZ0O/8R0lLTsXb6BvQa8w3K6+soLGvetSl8enrlOxNVTrMcDE0MxIeegS6unb0Jj1YNoaamptBX10BXoa9GOQ2pdknptm3YgdadW6FVB2/YVq2Mn34ZDpm2DId2HVZ2aCqB+SkY81M45ohKi1ILrlevXmHWrFno3LkzPDw84OHhgc6dO2P27Nl4+fKlMkMrUEj4NbSr4QlrQ3MAQNNq9VDdzBbB90Ly7F+nkjNqV3LCmvM7xTYvBw+oq6mjoqE5rk/YjkeTDiGozyxUMrLItf62gQvwbMoxHB++Bu1qNJVkn0rCPwu3omZDZzi5f/xU/PWzN5GSlAKPVg1yLQv8dRXGfz0Rc0f8gev/3vzobamqzMxM3L/7EHUb1Bbb1NXVUbdBbdy+cVd5gakI5qdgzE/hmCMqTUoruC5evAgHBwf88ccfMDQ0RJMmTdCkSRMYGhrijz/+gJOTEy5dulToOOnp6UhKSlJ4pKenSxr7yO0zcCf6EcInH0bK7IvYO3gpftoegDOPruTZv3/DzrgTFYZz4dfENrsKFaGupo4JLQZizM7Z6L5uLIzLG+DAkEBoarw905uc8Qbjds1Bj3Xj0HHVjzj7+Cq29p+vkkXXpWNX8PTBc3Qc1K5Exju7/zxc6jnB2MxIbJPpaKHLsI74fnI//DB9EOxr2mH5pNVltuhKTEiCPFsO4wrGCu3GFYwQ/ypOSVGpDuanYMxP4ZgjKk1Ku4brxx9/RLdu3RAYGJjrlJEgCBg6dCh+/PFHhITkPWuUIyAgAFOmKF77M3nyZPj7+5d0yCLfxj3Q0NYVnVeNQER8JBrZ18XCr/3wIvEljj04r9BXW1OG7nVbY/rhFQrt6mrq0CqniVE7ZuHI/bf7+N16PzydcgSe1eoj+F4IYlMSsPDkBnGdy09vwcrADKOb9c3z1KOyxMXEY8uSHfhx1jBoaml+9HjxLxNw+9JdfD+pr0K7nqEeWnTzFJ9XcaqMhNgkBG8+hlpf1fzo7RIREUlFaQXXtWvXsHbt2lzFFgCoqalh1KhRqFOnTqHj+Pn5YfTo0QptMpmsxOJ8n7amDNPa/Ihua0bjwJ3TAIAbkQ/gZu2IUc365Cq4utTyQnlNbWy4tFehPTLpFQDgTvR/F8O/SonHq5QE2Bhb5bv9CxE30cLxi5LanRIRcf8ZXscnY8aQuWKbXC7Hw+uPcHLnGfxxaDbUNYo+mRpy8AJ0DXRR68vCiyg7Z1vcvXzvg+JWdYZGBlDXUEd8bLxCe3xsAoxNTZQUlepgfgrG/BSOOXrL0d6m8E700ZR2StHS0hIXLlzId/mFCxdgYZH7eqb3yWQyGBgYKDykLLg01ctBq5wm5IJcoT1bkENdLXc6+zXsjL23TuBViuILOiT8KgDAwbyK2GZc3gCmukaIiIvMd/tuFR0R9f/Fmqpwqlsdv/45Hr+sHCs+KjvaoH6Luvhl5dhiFVuCICDk4Hk0bFmvSBfDP3v4HIYmBh8TvsrS1NSEg1M1XL3436louVyOqxdD4eLqpMTIVAPzUzDmp3DMEZUmpc1wjR07FoMHD8bly5fRokULsbiKjo7G0aNHsXLlSsyZM0cpselq6aCaaWXxeRWTinCzdkTcm0Q8TYjCyYeXMKP9KKRmpiMi/gUa29dD73rtMG7XXIVx7E1t0LhqXXRYNTzXNh68jMDuG8cxr9N4DNsyDa/TkvFb2xG4FxOOEw8vAgC+q9ceGdmZCH3+9uLNTq4t0K9BRwz5Z6qEe1982uW1YW2nOCsn09aCroGu2J4Yl4SkuNd4+fxtsfji0QvIymvDxNwIuu/c1uHe1QeIjYzDV21zz+KdO3QBGuXKwaZ6RQBA6OnrOHvwPHqP+VaqXVO6Lr07Y9bkeXBwrg7Hmg7YEbQLaanp8OnQUtmhqQTmp2DMT+GYIyotSiu4fH19YWpqivnz52Pp0qXIzs4GAGhoaMDd3R1r167FN998o5TY3G1q4IjvKvH5nE5jAQB/XdiN7zdNQu/1E/Bb2xFY13s6TMobICIuEpP2L8aKs1sUxunXoBOeJUbn++nF/kG/Yk6nsdj1/SLIBTlOh11GuxU/KNz24ZeWg1DZ2BpZ8izciwlHr78mYPv1IxLstbRO7z6L/X8dEp/PG7kYAPDd+B4Kn0Q8u/88qtaoAsvKec9uHthwGHHR8VDXUIeljTkGTuyDuk1rSxq7Mnl6N0VCfBLWBa5HfGw87B2qYvqiqbku8v1cMT8FY34KxxxRaVETBEFQdhCZmZl49ertzIepqSk0NT/+wmsA0Bpdu0TGKWsy5oWKPx99nvcd8T9nLSq2EX+OSA4roOfnqbKevfgz85M35qhgzE/B3s1PaXBd0r7UtnXDd0+pbUvVqMSd5jU1NWFllf+F4kRERESfMt5pnoiIiEhiLLiIiIiIJMaCi4iIiEhiLLiIiIiIJMaCi4iIiEhiLLiIiIiIJMaCi4iIiEhiLLiIiIiIJMaCi4iIiEhiLLiIiIhIJZ06dQrt27eHtbU11NTUsHPnToXl/fr1g5qamsKjVatWCn3i4uLQq1cvGBgYwMjICAMHDkRycrJCn+vXr6Nx48bQ1taGjY0NZs2alSuWLVu2wMnJCdra2nB1dcX+/cX7ajwWXERERKSSUlJS4ObmhiVLluTbp1WrVoiMjBQff//9t8LyXr164datWwgODsbevXtx6tQpDB48WFyelJQEb29v2Nra4vLly5g9ezb8/f2xYsUKsc/Zs2fRo0cPDBw4EFevXkWnTp3QqVMn3Lx5s8j7ohLfpUhERET0vtatW6N169YF9pHJZLC0tMxz2Z07d3Dw4EFcvHgR9erVAwAsWrQIbdq0wZw5c2BtbY2NGzciIyMDq1evhpaWFmrUqIHQ0FDMmzdPLMwWLlyIVq1aYdy4cQCAadOmITg4GIsXL0ZgYGCR9oUzXERERFQq0tPTkZSUpPBIT0//qDFPnDgBc3NzODo6YtiwYYiNjRWXhYSEwMjISCy2AMDLywvq6uo4f/682KdJkybQ0tIS+/j4+ODevXuIj48X+3h5eSls18fHByEhIUWOkwUXERERlYqAgAAYGhoqPAICAj54vFatWuGvv/7C0aNHMXPmTJw8eRKtW7dGdnY2ACAqKgrm5uYK65QrVw4mJiaIiooS+1hYWCj0yXleWJ+c5UXBU4pERERUKvz8/DB69GiFNplM9sHjde/eXfzZ1dUVtWrVgr29PU6cOIEWLVp88LhS4AwXERERlQqZTAYDAwOFx8cUXO+rWrUqTE1N8fDhQwCApaUlYmJiFPpkZWUhLi5OvO7L0tIS0dHRCn1ynhfWJ79rx/LCgouIiIjKhGfPniE2NhZWVlYAAA8PDyQkJODy5ctin2PHjkEul6Nhw4Zin1OnTiEzM1PsExwcDEdHRxgbG4t9jh49qrCt4OBgeHh4FDk2FlxERESkkpKTkxEaGorQ0FAAwOPHjxEaGoqIiAgkJydj3LhxOHfuHMLDw3H06FF07NgR1apVg4+PDwDA2dkZrVq1wqBBg3DhwgX8+++/GD58OLp37w5ra2sAQM+ePaGlpYWBAwfi1q1b+Oeff7Bw4UKFU58//fQTDh48iLlz5+Lu3bvw9/fHpUuXMHz48CLvCwsuIiIiUkmXLl1CnTp1UKdOHQDA6NGjUadOHUyaNAkaGhq4fv06OnToAAcHBwwcOBDu7u44ffq0wmnKjRs3wsnJCS1atECbNm3QqFEjhXtsGRoa4vDhw3j8+DHc3d0xZswYTJo0SeFeXV9++SWCgoKwYsUKuLm5YevWrdi5cydq1qxZ5H3hRfNERESkkjw9PSEIQr7LDx06VOgYJiYmCAoKKrBPrVq1cPr06QL7dOvWDd26dSt0e/nhDBcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUlMTSjojmJERERUprkuaV9q27rhu6fUtqVqOMNFREREJDEWXEREREQSK9PfpXj0+X5lh6CSWlRsI/48/9ocJUaimka5jRV/jkgOU2Ikqqmynr34M/OTN+aoYMxPwd7ND5UdnOEiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJseAiIiIikhgLLiIiIiKJlVN2AJ+KhJcJ2LFyL25fuIOMtEyYVTTFd+O7w9axMgBAEATsXXsQ/+4LQWpyGqrWrIIeI7vBvJKZOMay/63Cs7DneB2fjPL6OnCq64BOg9vDyNRQ7CMIAo5sPoF/94UgLjoOugZ6aNLxK7Tu3bLU97mo5HI5Lm2+ggenH+BNQip0TcrDsakD6napAzU1NQBv9+vS5su4c/Qu0lMyYOlkgcbfN4KR1dt9T4p5jSvbruD5zRfiGNUbV0fdr2tDo5yGuK2noU9xccsVxD+Nh4amBqycLeHR5wsYmOsrZd9Lw67Ne7Dlr22Ii42HfXU7+I4fBqeajsoOS2UwPwVjfgrHHFFp4AxXEbx5/QZzRvwBDQ0N+AYMxsQ1E/D10A4or1de7BO86RhObD+FHqO6YdySkZBpy7BoQiAyMzLFPg61q+H7SX0xeZ0fBvn3x8sXsVjpv1ZhW1sW78DZ/efw9ZAOmLTWD8N+H4gqTpVLa1c/SOjOa7gdfBuNBn6Fb+d3Q8NeDRC6+zpuHrj1X59d13DjwC00HtQIX0/vCE2ZJvb9fgBZGVkAgIQXCRAEoMngxvh2Xld82dcDt4Pv4ELQRXGMpJgkHJwdjIo1rNF11tdo+7/WSHudhsNzg0t9n0vLicMnsXzeSvQe3BPLNi5CVYeq8Bs+EfFxCcoOTSUwPwVjfgrHHFFpYcFVBIf/PgpjcyP0mdADVZxtYWpVAS71nWBW0RTA29mbY9tOolVvb7h95YpK9tbo+3NPJL5KwrUzN8RxWnTzhJ1LFVSwNIF9TTv49GiB8DtPkJ2VDQCIfBKNU7v/xdBpA1Hrq5owtaqAyg42cK6n2n9pRd2PRpV6trCtWxkG5vqw/6IqKtWqiJiHLwG8zc+N/TdR9+s6sKtfBRVsK6DZcE+8iX+D8ItPAACVa9ug2Q9NYeNWCQYWBqhSzxZu7V3x6EK4uJ2Xj15BkMvRoHs9GFoawKyqKdza18Kr8FhkZ8mVseuS27ZhB1p3boVWHbxhW7UyfvplOGTaMhzadVjZoakE5qdgzE/hmCMqLSy4iuB6yC3YOthgpf9ajP96IqYPnoMze0PE5bGRsUiKew0ndwexTUdPB1WcbfHodnieY6YkpeDC0cuoWqOKeMrsRshNmFpVwI1ztzCx5zT82mMqNszZhJSkFEn372NZOljg2c0XSHiRAAB4FR6LqHvRsKljAwB4HfMabxJSUalWRXEdWXktmFczQ9T96HzHzXiTAW09mfjcrKopoKaGuyfuQS6XI/1NBu6feoBKrhWhUa7sHcqZmZm4f/ch6jaoLbapq6ujboPauH3jrvICUxHMT8GYn8IxR1SaVPoarqdPn2Ly5MlYvXq1UuN49SIWp3afRYtunmjVywtP7kVgy+IdKKepgS98GiAx7jUAwMBYT2E9A2M9JP3/shw7VuzByZ1nkJGWATsXWwz7fdB/24mMRVx0PK6cvIa+P/eEXC5g69KdWOm/FiPn+Uq/ox+oTqfayEjNxKZRW6Curga5XECD7vXh0LgaAOBNQioAQMdQR2E9HUMdpP7/svclRiXi5oFb+OK7L8Q2A3MDtPu1DYLnH8WpFWcgyAVYOJijjV8rifZMuRITkiDPlsO4grFCu3EFIzwNf6qkqFQH81Mw5qdwzBGVJpWeFoiLi8O6desK7JOeno6kpCSFR3p6eonGIQgCbKpXQsfv28KmeiU0avclvmr7BU7vOVvssVp+2wx+y8fgx1lDoa6ujnUzNkIQhLfbkQvIysxC3597olotezjUrobeY7/F/dCHiI6IKdF9KklhIY/w4MxDeI1oji4zv0ZzX09c23Md907c/6DxkuNSsO/3g6jqURUuXk5i+5uENzi5/BQcmlZHl4BO6ODfDurlNHB43hExh0RERKpIqTNcu3fvLnD5o0ePCh0jICAAU6ZMUWibPHky/P39PyY0BYYmBrCqYqHQZlnZAldPXf//5W8/IZcUnwzDCv994jApPhmVqlkrrKdnqAc9Qz1Y2JjD0tYC//t2Ch7ffoKqNarAsIIB1DXUYWFj/t92bN9uNy4mHhaVzaGKQjacR52Obqj2lT0AoEJlE7x++RpXd4bC0dMB5Y3ezmylJqZC1/i/DxqkJqaiQpUKCmOlxKVgz5S9sHQ0R9PBjRWW3Tx4G1rlteDRu6HY1uJHT2wY9jdiHsTAwkHxd/SpMzR6ezzEx8YrtMfHJsDY1ERJUakO5qdgzE/hmCMqTUotuDp16gQ1NbUCZydybiuQHz8/P4wePVqhTSaT5dP7w1StaYfop4ozTDHPYmBi8XYauoJVBRiY6OPelfuwqfb2OqXUlDSE33mCJh2+zHdcQf52v7My335Sr2oNO8iz5Xj5/JV4QX7Ms7cXnudsSxVlpWdBTV3x96Smri7+XvXN9VHeSAfPbzyH6f8XWBlvMhDz8CVqeLuI6yT/f7FlZmcGzx+a5hozKyMr1/Ggpv52krYsTnBpamrCwakarl68hq+avT2O5HI5rl4MRcdv2is5OuVjfgrG/BSOOaLSpNSCy8rKCkuXLkXHjh3zXB4aGgp3d/cCx5DJZCVeYL2vedemmPPjQhzcGIy6nrXx5G4Ezuw7h56jvwHwtihs3qUpDmwIhnlFM1SwMsGeNQdgaGoAt0auAIDHd57gyd0I2LtWRXk9Hbx6EYs9a/bDzNoUdi5VAABO7g6wqV4J62dvQjffThAEAZsWboOTu4PCrJeqsXWvjCvbQ6FnqgfjSsaIDX+F63tvwKnZ2w8RqKmpwbVNTVzefhWGVobQN9fHxU2XUN64PKrUtwXwttja7b8X+mZ6+KJPQ6QlpYnjlzd6OytmW9cG1/fdwKWtV1DtK3tkpmbiwt8XoWemB1O7CrkDKwO69O6MWZPnwcG5OhxrOmBH0C6kpabDp4Pq3petNDE/BWN+CsccUWlRasHl7u6Oy5cv51twFTb7VVqqOFXGkKkDsGvVPuz/6zAqWJmg6w+d0MDrv2KwZffmSE/LQNC8zXiTnAp7VzsMnzEEmlqaAAAtmSZCT1/HvnUHkZ6aAcMKBnCp74TWvVtCU+vtr0FdXR3Dfv8emxdtx7yRiyHT1oJLA2d0GdZBKftdVI0GfImL/1zG6VX/vj1taFIeLi2d4N61rtindkc3ZKVn4eTy08h48/bGp21/aYVy/7/vz64/R1JUEpKikrBhaJDC+EM3v/1gQcWaFeE1ojlCd19D6K5rKCcrB0sHC7T9pbU4Tlnj6d0UCfFJWBe4HvGx8bB3qIrpi6bmusj3c8X8FIz5KRxzRKVFTVBiRXP69GmkpKSgVau8P2WWkpKCS5cuoWnTph80/tHn+z8mvDKrRcU24s/zr81RYiSqaZTbWPHniOQwJUaimirr2Ys/Mz95Y44KxvwU7N38lAbXJaV3+vSG755S25aqUeq0QOPGjQtcrqur+8HFFhEREZGqUOnbQhARERGVBSy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhU0qlTp9C+fXtYW1tDTU0NO3fuFJdlZmZiwoQJcHV1ha6uLqytrdGnTx+8ePFCYYwqVapATU1N4TFjxgyFPtevX0fjxo2hra0NGxsbzJo1K1csW7ZsgZOTE7S1teHq6or9+4v3bTYsuIiIiEglpaSkwM3NDUuWLMm17M2bN7hy5QomTpyIK1euYPv27bh37x46dMj9/cNTp05FZGSk+Pjxxx/FZUlJSfD29oatrS0uX76M2bNnw9/fHytWrBD7nD17Fj169MDAgQNx9epVdOrUCZ06dcLNmzeLvC9l8xt/iYiI6JPXunVrtG7dOs9lhoaGCA4OVmhbvHgxGjRogIiICFSuXFls19fXh6WlZZ7jbNy4ERkZGVi9ejW0tLRQo0YNhIaGYt68eRg8eDAAYOHChWjVqhXGjRsHAJg2bRqCg4OxePFiBAYGFmlfOMNFREREpSI9PR1JSUkKj/T09BIbPzExEWpqajAyMlJonzFjBipUqIA6depg9uzZyMrKEpeFhISgSZMm0NLSEtt8fHxw7949xMfHi328vLwUxvTx8UFISEiRY+MMFxER0WfMxa5iqW0rICAAU6ZMUWibPHky/P39P3rstLQ0TJgwAT169ICBgYHYPmLECNStWxcmJiY4e/Ys/Pz8EBkZiXnz5gEAoqKiYGdnpzCWhYWFuMzY2BhRUVFi27t9oqKiihwfCy4iIiIqFX5+fhg9erRCm0wm++hxMzMz8c0330AQBCxbtkxh2bvbq1WrFrS0tDBkyBAEBASUyLaLigUXERERlQqZTFbiRU5OsfXkyRMcO3ZMYXYrLw0bNkRWVhbCw8Ph6OgIS0tLREdHK/TJeZ5z3Vd+ffK7LiwvvIaLiIiIPkk5xdaDBw9w5MgRVKhQodB1QkNDoa6uDnNzcwCAh4cHTp06hczMTLFPcHAwHB0dYWxsLPY5evSowjjBwcHw8PAocqyc4SIiIiKVlJycjIcPH4rPHz9+jNDQUJiYmMDKygpdu3bFlStXsHfvXmRnZ4vXVJmYmEBLSwshISE4f/48mjVrBn19fYSEhGDUqFHo3bu3WEz17NkTU6ZMwcCBAzFhwgTcvHkTCxcuxPz588Xt/vTTT2jatCnmzp2Ltm3bYtOmTbh06ZLCrSMKw4KLiIiIVNKlS5fQrFkz8XnO9Vh9+/aFv78/du/eDQCoXbu2wnrHjx+Hp6cnZDIZNm3aBH9/f6Snp8POzg6jRo1SuK7L0NAQhw8fhq+vL9zd3WFqaopJkyaJt4QAgC+//BJBQUH49ddf8csvv6B69erYuXMnatasWeR9YcFFREREKsnT0xOCIOS7vKBlAFC3bl2cO3eu0O3UqlULp0+fLrBPt27d0K1bt0LHyg+v4SIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIompCYXdF5+IiIjKrG/3Dy21bf3TJrDUtqVqOMNFREREJDEWXEREREQSK6fsAKQUkRym7BBUUmU9e/Fn5ii3d/NTmlPtn4p3Twnw+MkbX2MFY34K9m5+qOzgDBcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxMopO4CybtfmPdjy1zbExcbDvrodfMcPg1NNR2WHpTLKan6cjauhfVVv2BlWhom2EWZfXoZL0dcU+nSr3h4tbBpBV1MH9+LDsOrm34h6EwMAMNOpgK+rtUHNCo4wkhkgLi0RZ16cx/aHB5AtZOfankV5M8xs9D/IBTkGBI9WWPaFZV1849ABZjoVEPUmBhvv7kDoy5vS7XwpK6vHUElhfgrHHFFp4AyXhE4cPonl81ai9+CeWLZxEao6VIXf8ImIj0tQdmgqoSznR1ZOhievn2H1rU15Lu9Q1RutqzTDqptB+N/ZmUjLzsAvDX6Epvrbv4GsdS2grqaGlTc3YsypqfjrzhZ4VW6CHo6dco2loaaOn2oPxN24h7mWORhVxYjaA3H86b/4+czvuBgVinHuQ2GjZ12i+6ssZfkYKgnMT+GYIyotLLgktG3DDrTu3AqtOnjDtmpl/PTLcMi0ZTi067CyQ1MJZTk/oS9v4Z/7u3ExOjTP5W2qtMD2hwdwKeYaIl4/x5Jra2AsM0J9i9oAgGuvbmPZ9b9w/dUdxKS+wuWY69j7KBgNLGvnGutbh454nhKFkMjLuZa1rtIcoa9uYc/jYDxPicLmB3vwODECPlU8S25nlagsH0MlgfkpHHNEpYUFl0QyMzNx/+5D1G1QW2xTV1dH3Qa1cfvGXeUFpiI+5/yY65jCWNsQN17dEdtSs9LwMOExqhtVzXe98po6SM58o9BWo4IjvrCqm+9MmoNxVdx8pZjPa69uw6GA7XwqPudjqCiYn8IxR1SaWHBJJDEhCfJsOYwrGCu0G1cwQvyrOCVFpTo+5/wYyQwAAIkZSQrtiRmvxWXvsyhvhla2zXAk4rTYpqepix9q9cWya+uQmpWW77YS0t/bTvprGOaznU/J53wMFQXzUzjmiEqT0guu1NRUnDlzBrdv3861LC0tDX/99VeB66enpyMpKUnhkZ6eLlW4RKXOWGaEX+r/iHNRl3Hs6RmxfYhrb5x5cQF34nNfu0VERKpFqQXX/fv34ezsjCZNmsDV1RVNmzZFZGSkuDwxMRH9+/cvcIyAgAAYGhoqPAICAqQOvVCGRgZQ11BHfGy8Qnt8bAKMTU2UFJXq+JzzkzPjZKilOMtkqKWfazbKWGaISV+Mwv34R1hxY6PCshoVHNHeriWCWi1BUKslGFrrO+hqlkdQqyXwrPSluK33Z80MZfpIfG87n6LP+RgqCuancMwRlSalFlwTJkxAzZo1ERMTg3v37kFfXx9fffUVIiIiijyGn58fEhMTFR5+fn4SRl00mpqacHCqhqsX/7sVgFwux9WLoXBxdVJiZKrhc85PTOorxKclwtX0v/3UKaeNakZ2eJDwSGwzlhlh0hej8TgxAkuvr4MAQWGciWdnYcKZ38XH5vt78CYzFRP+/9OIAHA//hFqVlDMp6upM+6/s51P1ed8DBUF81M45ohKk1Lvw3X27FkcOXIEpqamMDU1xZ49e/DDDz+gcePGOH78OHR1dQsdQyaTQSaTlUK0xdeld2fMmjwPDs7V4VjTATuCdiEtNR0+HVoqOzSVUJbzI9OQwbK8mfjcXMcUtvqVkJyZgti0eOwPP4rO1VojMiUGMamv8G31DohPTxA/1WgsM8LkL0bjVWos1t/ZBgMtfXGsnGu/nqdEKWyzqqEtBAh4mvxCbDsQfgyTvxiDdnZeuBJzA19a14e9oS1Wvjdb9qkqy8dQSWB+CsccUWlRasGVmpqKcuX+C0FNTQ3Lli3D8OHD0bRpUwQFBSkxuo/n6d0UCfFJWBe4HvGx8bB3qIrpi6bmukDzc1WW82NvaIvJX/x3A9K+Lt0AACeehWDZ9XXY/egwZBoyDHbthfLlyuNe/EMEXFyETHkWAKCWqTOsdM1hpWuOwBYzFMb+dv/QIsdxP+ERFoX+iW8dOqC7Q0dEvYnB7MuBCkXZp6wsH0MlgfkpHHNEpUVNEASh8G7SaNCgAX788Ud89913uZYNHz4cGzduRFJSErKzc99ZuygiksM+NsQyqbKevfgzc5Tbu/kpTnHzufinTaD4M4+fvPE1VjDmp2Dv5qc0lOb73LvvH58bpV7D1blzZ/z99995Llu8eDF69OgBJdaDRERERCVCqQWXn58f9u/fn+/ypUuXQi6Xl2JERERERCVP6ffhIiIiIirrWHARERERSYwFFxEREZHEWHARERERSYwFFxEREZHEWHARERERSYwFFxEREZHEWHARERGRSjp16hTat28Pa2trqKmpYefOnQrLBUHApEmTYGVlBR0dHXh5eeHBgwcKfeLi4tCrVy8YGBjAyMgIAwcORHJyskKf69evo3HjxtDW1oaNjQ1mzZqVK5YtW7bAyckJ2tracHV1LfA+onlhwUVEREQqKSUlBW5ubliyZEmey2fNmoU//vgDgYGBOH/+PHR1deHj44O0tDSxT69evXDr1i0EBwdj7969OHXqFAYPHiwuT0pKgre3N2xtbXH58mXMnj0b/v7+WLFihdjn7Nmz6NGjBwYOHIirV6+iU6dO6NSpE27evFnkfVHql1cTERER5ad169Zo3bp1nssEQcCCBQvw66+/omPHjgCAv/76CxYWFti5cye6d++OO3fu4ODBg7h48SLq1asHAFi0aBHatGmDOXPmwNraGhs3bkRGRgZWr14NLS0t1KhRA6GhoZg3b55YmC1cuBCtWrXCuHHjAADTpk1DcHAwFi9ejMDAon0/ZJEKrt27dxdpMADo0KFDkfsSERHR5yM9PR3p6ekKbTKZDDKZrNhjPX78GFFRUfDy8hLbDA0N0bBhQ4SEhKB79+4ICQmBkZGRWGwBgJeXF9TV1XH+/Hl07twZISEhaNKkCbS0tMQ+Pj4+mDlzJuLj42FsbIyQkBCMHj1aYfs+Pj65TnEWpEgFV6dOnYo0mJqaGrKzs4u8cSIiIvp8BAQEYMqUKQptkydPhr+/f7HHioqKAgBYWFgotFtYWIjLoqKiYG5urrC8XLlyMDExUehjZ2eXa4ycZcbGxoiKiipwO0VRpIKLXyBNREREH8vPzy/XTNGHzG59ingNFxEREZWKDz19mBdLS0sAQHR0NKysrMT26Oho1K5dW+wTExOjsF5WVhbi4uLE9S0tLREdHa3QJ+d5YX1ylhfFBxVcKSkpOHnyJCIiIpCRkaGwbMSIER8yJBEREVGR2dnZwdLSEkePHhULrKSkJJw/fx7Dhg0DAHh4eCAhIQGXL1+Gu7s7AODYsWOQy+Vo2LCh2Od///sfMjMzoampCQAIDg6Go6MjjI2NxT5Hjx7FyJEjxe0HBwfDw8OjyPEWu+C6evUq2rRpgzdv3iAlJQUmJiZ49eoVypcvD3NzcxZcREREVCKSk5Px8OFD8fnjx48RGhoKExMTVK5cGSNHjsRvv/2G6tWrw87ODhMnToS1tbV47bmzszNatWqFQYMGITAwEJmZmRg+fDi6d+8Oa2trAEDPnj0xZcoUDBw4EBMmTMDNmzexcOFCzJ8/X9zuTz/9hKZNm2Lu3Llo27YtNm3ahEuXLincOqIwxb4P16hRo9C+fXvEx8dDR0cH586dw5MnT+Du7o45c+YUdzgiIiKiPF26dAl16tRBnTp1AACjR49GnTp1MGnSJADA+PHj8eOPP2Lw4MGoX78+kpOTcfDgQWhra4tjbNy4EU5OTmjRogXatGmDRo0aKRRKhoaGOHz4MB4/fgx3d3eMGTMGkyZNUrhX15dffomgoCCsWLECbm5u2Lp1K3bu3ImaNWsWeV/UBEEQirPzRkZGOH/+PBwdHWFkZISQkBA4Ozvj/Pnz6Nu3L+7evVuc4SQVkRym7BBUUmU9e/Fn5ii3d/Pz7f6hSoxENf3T5r97zvD4yRtfYwVjfgr2bn5KQ2m+z737/vG5KfYMl6amJtTV365mbm6OiIgIAG8rxKdPn5ZsdERERERlQLGv4apTpw4uXryI6tWro2nTppg0aRJevXqF9evXF2tqjYiIiOhzUewZrunTp4sfv/z9999hbGyMYcOG4eXLl8W6eIyIiIjoc1HsGa53b49vbm6OgwcPlmhARERERGVNsWe4iIiIiKh4ij3DZWdnBzU1tXyXP3r06KMCIiIiIipril1wvXuXVQDIzMzE1atXcfDgQYwbN66k4iIiIiIqM4pdcP300095ti9ZsgSXLl366ICIiIiIypoSu4ardevW2LZtW0kNR0RERFRmlFjBtXXrVpiYmJTUcERERERlxgfd+PTdi+YFQUBUVBRevnyJpUuXlmhwH6u0vx7hU8QcFexz/hqKouDxUzjmqGDMD30uil1wdezYUaHgUldXh5mZGTw9PeHk5FSiwREREZG0vqhYTdkhfBaKXXD5+/tLEAYRERFR2VXsgktDQwORkZEwNzdXaI+NjYW5uTmys7NLLLiPxW+hz9u7U/jMUW7MT8Hezc/R5/uVGInqalGxjfgzj6Hc+BorGE+zlk3FvmheEIQ829PT06GlpfXRARERERGVNUWe4frjjz8AAGpqali1ahX09PTEZdnZ2Th16hSv4SIiIiLKQ5ELrvnz5wN4O8MVGBgIDQ0NcZmWlhaqVKmCwEB+oouIiIjofUUuuB4/fgwAaNasGbZv3w5jY2PJgiIiIiIqS4p90fzx48eliIOIiIiozCr2RfNdunTBzJkzc7XPmjUL3bp1K5GgiIiIiMqSYhdcp06dQps2bXK1t27dGqdOnSqRoIiIiIjKkmIXXMnJyXne/kFTUxNJSUklEhQRERFRWVLsgsvV1RX//PNPrvZNmzbBxcWlRIIiIiIiKkuKfdH8xIkT8fXXXyMsLAzNmzcHABw9ehRBQUHYunVriQdIRERE9KkrdsHVvn177Ny5E9OnT8fWrVuho6MDNzc3HDt2DCYmJlLESERERPRJK3bBBQBt27ZF27ZtAQBJSUn4+++/MXbsWFy+fFmlvkuRiIiISBUU+xquHKdOnULfvn1hbW2NuXPnonnz5jh37lxJxkZERERUJhRrhisqKgpr167Fn3/+iaSkJHzzzTdIT0/Hzp07ecE8ERERUT6KPMPVvn17ODo64vr161iwYAFevHiBRYsWSRkbERERUZlQ5BmuAwcOYMSIERg2bBiqV68uZUxEREREZUqRZ7jOnDmD169fw93dHQ0bNsTixYvx6tUrKWMjIiIiKhOKXHB98cUXWLlyJSIjIzFkyBBs2rQJ1tbWkMvlCA4OxuvXr6WMk4iIiOiTVexPKerq6mLAgAE4c+YMbty4gTFjxmDGjBkwNzdHhw4dpIiRiIiI6JP2wbeFAABHR0fMmjULz549w99//11SMRERERGVKR9VcOXQ0NBAp06dsHv37pIYjoiIiKhMKZGCi4iIiIjyx4KLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkxoKLiIiISGIsuIiIiIgkVuQvr6YPs2vzHmz5axviYuNhX90OvuOHwammo7LDUhnMT+E+xxwdCjqCXav2odnXTdBteGcAQGZGJrYt24XLx68iKyMLzvWd0P2nrjAw0RfXu3vlPvasPoAXjyMh09ZCQ5/66DCwDTQ0NAAA0RExCFqwBVFPopCanAZDUwPUb+6Otn19oFFOQyn7WlTXr9zAlr+24f6dh4h7FQf/Ob/iq2Zfisv/Wr4BJw6dwsvolyinqYnqztXQ/4c+cHZ1Uhjn/OkL2LAyCI8ehkNLSwu16tbElHmTxOVLZgXi1rXbCA8Lh41dZSz/e3Gp7aOyfI6vMSp9nOGS0InDJ7F83kr0HtwTyzYuQlWHqvAbPhHxcQnKDk0lMD+F+xxzFH43Amf2hqBiVWuF9q1LduJGyC18P6kfRi0YjsTYRKyYvFpc/izsOZb6rUCNBk7wWzEGAyb1wfWzN7FzxV6xj0Y5dXzRsh5+nDUUk9f5odsPnfHv/hDsXXuw1PbvQ6WlpqGqgx1+nPBDnssrVa6I4ROGYcU/SzH/z9mwsDLHz76/IiE+Uexz+ugZzJw0B94dWmL534uxYPUcNG/lmWssn44t0dS7iVS7olI+x9cYKQcLLglt27ADrTu3QqsO3rCtWhk//TIcMm0ZDu06rOzQVALzU7jPLUdpqelYO30Deo35BuX1dcT21ORUnD1wHl2GdYRj3eqo7GCD78b3wKNb4Xh8OxwAcPl4KKyrWqNNHx+YVzSDg1s1dB7cHqd2/Yu0N2kAAFNrU3i0bohK9hVRwdIEtb6qifot3PHwRpgydrdYGnxVH/1/6ItGzb/Mc3nz1s1Qt2EdWFWyQhV7WwwdPRhvUt7g0YPHAIDsrGwsnbMcg34aiPZd26KSbSXYVq2cq7DyHT8UHb9pD6uKlpLvkyr43F5jpDwsuCSSmZmJ+3cfom6D2mKburo66jaojds37iovMBXB/BTuc8zRPwu3omZDZzi5K57Oibj/DNlZ2QrtlpUtYGJujEe3wgEAWZlZ0NRSvEpCS6aJzIxMRNx/luf2Yp6/xO2Ld1G9VrWS3REly8zMxP7tB6Crpwv76nYAgAd3H+JVTCzU1NUwtOdwfOvdC7/8OBGPH4YrN1gl+hxfY6Q8Sr+G686dOzh37hw8PDzg5OSEu3fvYuHChUhPT0fv3r3RvHnzAtdPT09Henq6QptMJoNMJpMy7EIlJiRBni2HcQVjhXbjCkZ4Gv5USVGpDuancJ9bji4du4KnD55jwrJRuZYlxSehnKYGyuvpKLTrG+sjKf41AMC5niOObTuJi0evwN2zNpLikrD/r7ezFImxSQrrzR6+EE8fPENWZhYatfNAu/6tJNqr0nXu1Hn8/stMpKelw8TUBDOX/g5DY0MAQOTzKADA+uUbMXT0IFhYW2Dr+u0YO/hnrNmxEgaG+gUNXSZ9bq8xUi6lznAdPHgQtWvXxtixY1GnTh0cPHgQTZo0wcOHD/HkyRN4e3vj2LFjBY4REBAAQ0NDhUdAQEAp7QERlYS4mHhsWbID/X7pDU0tzQ8aw6W+E74e0gF/L9iCET7j4N83ADUaOgMA1NTVFPoOnNQHfsvHoP//vsPNc7dxZPPxj94HVeBW3w2Bfy/GgjVzUf9Ld/z2c4B4LZIglwMAeg7sjsYtGsHBuTrG+o+Gmhpw6shpJUZN9HlQ6gzX1KlTMW7cOPz222/YtGkTevbsiWHDhuH3338HAPj5+WHGjBkFznL5+flh9OjRCm3Knt0CAEMjA6hrqCM+Nl6hPT42AcamJkqKSnUwP4X7nHIUcf8ZXscnY8aQuWKbXC7Hw+uPcHLnGQyfNQRZmdl4k5yqMMv1Ov41DIz/m5lp0c0Tzbs2RWJsEsrr6yA2Kh67Vu2DqVUFhe2ZmL+d0bCqYgm5XI6geZvh1a0Z1DU+7assdHS0UdHGGhVtrOHi6oS+nb7HwZ2H0GPAtzD5/2PG1q6y2F9LSxNWFS0RE/VSWSEr1ef0GiPlU+q7y61bt9CvXz8AwDfffIPXr1+ja9eu4vJevXrh+vXrBY4hk8lgYGCg8FCFgktTUxMOTtVw9eI1sU0ul+PqxVC4vPcx7c8R81O4zylHTnWr49c/x+OXlWPFR2VHG9RvURe/rBwLWwcbaJTTwL0r98V1oiNiEBcTj6o1qiiMpaamBiNTQ2jJtHDp2BUYmxuhcvVK+W5bEARkZ2VDEASpdk9pBLkcmZmZAIDqztWhqaWJp0/+u54tKzMLUZExsLAyV1aISvU5vcZI+ZR+DZea2tupfnV1dWhra8PQ0FBcpq+vj8TExPxWVXldenfGrMnz4OBcHY41HbAjaBfSUtPh06GlskNTCcxP4T6XHGmX14a1nZVCm0xbC7oGumL7l60bYtvSXSivXx46utr454/tsHOpAjuXKuI6wZuOwaWBE9TU1BB65gYO/30UAyf1FWeuLhy5DI1y6rC2s4ampgae3H+KXSv3wb1ZHZW/D1fqm1Q8f/pCfB71IhoP74XBwEAf+kYGCPpzEzyafoEKpsZITEjC7s178eplLJp4NQYA6OqVR7subfDX8g0wszCDhZU5Nv+1FQDQxKuROO7zpy+Q+iYVcbHxyEhPx8N7bz/BaVu1MjQ1P+x0ryr7XF5jpHxKLbiqVKmCBw8ewN7eHgAQEhKCypX/m+6OiIiAlZVVfqurPE/vpkiIT8K6wPWIj42HvUNVTF80NdcFmp8r5qdwzNF/uvp2gpq6Glb6r0VWZhac6zmi+8iuCn1uXbiDgxuDkZWZjYr21hg6baB4HRcAqGuo4/DfxxDz7CUgCDCxMEbTzo3QomvT0t6dYrt/+wHGDvlZfB44byUAoGU7L4z8ZTiehj9D8N7fkZSQCH1DAzjWcMD8VbNRxd5WXGfwTwOhoaGBmZPmICM9HU41HTE7MAD6Bv+dlp03bSGuX74hPh/W80cAwPo9a2BpbSH1bpY6vsaotKgJSpxHDwwMhI2NDdq2bZvn8l9++QUxMTFYtWrVB40fkaz699ZRhsp69uLPzFFuzE/B3s3P0ef7lRiJ6mpRsY34M4+h3PgaK9i7+SkN86/NKbVtjXIbW2rbUjVKneEaOnRogcunT59eSpEQERERSefT/kgOERER0SeABRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGpnCpVqkBNTS3Xw9fXFwDg6emZa9n79/eMiIhA27ZtUb58eZibm2PcuHHIyspS6HPixAnUrVsXMpkM1apVw9q1ayXZH6V/lyIRERHR+y5evIjs7Gzx+c2bN9GyZUt069ZNbBs0aBCmTp0qPi9fvrz4c3Z2Ntq2bQtLS0ucPXsWkZGR6NOnDzQ1NcUbqz9+/Bht27bF0KFDsXHjRhw9ehTff/89rKys4OPjU6L7w4KLiIiIVI6ZmZnC8xkzZsDe3h5Nm/733afly5eHpaVlnusfPnwYt2/fxpEjR2BhYYHatWtj2rRpmDBhAvz9/aGlpYXAwEDY2dlh7ty5AABnZ2ecOXMG8+fPL/GCi6cUiYiIqFSkp6cjKSlJ4ZGenl7oehkZGdiwYQMGDBgANTU1sX3jxo0wNTVFzZo14efnhzdv3ojLQkJC4OrqCguL/7503cfHB0lJSbh165bYx8vLS2FbPj4+CAkJ+dhdzYUFFxEREZWKgIAAGBoaKjwCAgIKXW/nzp1ISEhAv379xLaePXtiw4YNOH78OPz8/LB+/Xr07t1bXB4VFaVQbAEQn0dFRRXYJykpCampqR+6m3niKUUiIiIqFX5+fhg9erRCm0wmK3S9P//8E61bt4a1tbXYNnjwYPFnV1dXWFlZoUWLFggLC4O9vX3JBV1CWHARERFRqZDJZEUqsN715MkTHDlyBNu3by+wX8OGDQEADx8+hL29PSwtLXHhwgWFPtHR0QAgXvdlaWkptr3bx8DAADo6OsWKszA8pUhEREQqa82aNTA3N0fbtm0L7BcaGgoAsLKyAgB4eHjgxo0biImJEfsEBwfDwMAALi4uYp+jR48qjBMcHAwPD48S3IO3WHARERGRSpLL5VizZg369u2LcuX+OykXFhaGadOm4fLlywgPD8fu3bvRp08fNGnSBLVq1QIAeHt7w8XFBd999x2uXbuGQ4cO4ddff4Wvr684yzZ06FA8evQI48ePx927d7F06VJs3rwZo0aNKvF9YcFFREREKunIkSOIiIjAgAEDFNq1tLRw5MgReHt7w8nJCWPGjEGXLl2wZ88esY+Ghgb27t0LDQ0NeHh4oHfv3ujTp4/Cfbvs7Oywb98+BAcHw83NDXPnzsWqVatK/JYQAK/hIiIiIhXl7e0NQRBytdvY2ODkyZOFrm9ra4v9+/cX2MfT0xNXr1794BiLijNcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQksTL91T6V9eyVHYLKY44KxvwUrEXFNsoOQeXxGCoY80OfC85wEREREUmMBRcRERGRxMr0KcWI5DBlh6CS3p3CZ45yY34KxvwUjjkqGPNTMJ5mLZs4w0VEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQksXLKDoCIiIiUp5api7JD+CxwhouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGL6+W2K7Ne7Dlr22Ii42HfXU7+I4fBqeajsoOS2UwP4X7HHJ0/coNbPlrG+7feYi4V3Hwn/Mrvmr2pbh81uR5CN57RGGdeh7uCFg8TXw+cdQUhN17hIT4BOjr66FOw9r4fsQAmJpVEPsIgoCt67dj344DiImMgYGRIdp3a4teA7tLv5MfYc+WfdizdR+iI6MBALZVbdF7UA80+Ko+ACAjPQOB81fixOFTyMzIRD2Puhjxsy+MKxgDAMLuP8KmtVtwK/QWEhOSYGFlgXZdWuPrnp0UtrNr8x7s+mcvoiOjYW5php4DuqNluxaluq8l6VXMK6z6Yw0unL2E9LR0WFeywlj/UXB0cQAAnD72L/Zu3Y8Hdx/ideJrLAtahGqO9nmOJQgC/jdiEi6evZzr+CQqChZcEjpx+CSWz1uJEb8Mh3NNJ2wP2gm/4ROxevsKGJsYKTs8pWN+Cve55CgtNQ1VHezg08EbU8b9lmef+l+6Y+zkUeJzTS1NheW169VCjwHfooKpMV7FxGLFgj8xbfx0LFwzV+yzdPZyXD53BYNHfg+7alXwOuk1Xie+lmanSpCphSkG/tgfFStbA4KAw3uPYvLoaVgWtAhV7G2xbO4KnD9zERNn+EFXXxeLZy6D/7jfsHD1231/cOchjIwNMWHaOJhbmOLW9TtY8NsiqGtooNO37QG8LepWL16LUb+OgKOLA+7euo/5v/0BPQM9eDRpqMzd/yCvk15j5ICxcKtXC9P/mApDY0M8j3gBfX19sU9aahpq1q6Bpi0bY/5vfxQ43vagnYCamsRRU1mmcgWXIAhQKyMH9bYNO9C6cyu06uANAPjpl+E4f+YiDu06jO79v1FydMrH/BTuc8lRg6/qi7M1+dHU1ISJqUm+y7v06iz+bGFlgW/7dYP/mGnIysxCOc1yePI4Anu27sPKzctgU6USAMCqomXJ7IDE3i94Bvj2xd6t+3Dnxl2YmZvi4K7D8Pt9POo0qA0AGDt5FAZ2HYLbN+7CxdUJrTp6K6xvVckKt6/fwb/H/hULriP7j6Ht123g6d1U7HPv1n38s3bLJ1lw/bN2K8wszDDOf7TY9v7vu2Xbt7N3US+iCxzr4b0wbN2wHUvWL8S3Pr1LPlj6LKjcNVwymQx37txRdhgfLTMzE/fvPkTd/38DBAB1dXXUbVAbt2/cVV5gKoL5KRxzpOja5Rvo5tUD/b8ehIXTFyMpISnfvkmJr3HswHG41HJGOc23f1eeO3UeVpUsce70BXzXvj96t+uHuVMXIOkTmOF6V3Z2No4fOom01DS41HLG/TsPkJWVhboNa4t9KtvZwNzSDHeu5/9e+ib5DfQN/5vtyczIhNZ7s4YybS3cu3UfWZlZJb4fUgs5dQ4OLtUxdfx0dPPqgaE9h2P/9oPFHictNQ0B/5uFHyf8UGDBT1QYpc1wjR49Os/27OxszJgxAxUqvL3uYt68eaUZVolJTEiCPFsuXkORw7iCEZ6GP1VSVKqD+Skcc/Sf+l+6o1HzL2FlbYEXzyKxesk6/DJiEhaumQsNDQ2x38o/VmP3P3uQlpYOZ1cn/LbAX1wW+TwK0ZExOHXkNMZPHQN5thzL5q3AtPG/Y/byGUrYq+J5/OAxRvQfg4yMDOjo6GDynImwrVoZYffCoKlZDnr6egr9jSsYIy42Ps+xbl27jROHT+G3hVPENnePujiw8xC+bOaB6k7VcP/OAxzYeRhZWVlITEhCBbNPq9iIfB6FPVv3oUuvzug54Fvcu30fS+YEopxmOXi39yryOIHzVsKlljO+9PSQMFr6HCit4FqwYAHc3NxgZGSk0C4IAu7cuQNdXd0inVpMT09Henq6QptMJoNMJivJcIlIiZr5NBV/tqtuh6rV7dCn40Bcu3xDYQbwm++6oHVHb0RHxmD9iiDMnDQXvy30h5qaGgS5gMyMTEyYOgaVbN+eUhwzcSR+6D0CT8OfiacZVVWlKpUQ+PdipCSn4PSRM5g9eS7mrpxV7HEePwzH5NFT8d3gnqjnUVds7/19D8THxmNE39EQIMDYxBgt27XA5nVboa7+6V3mIcgFOLhUx8Dh/QAA1ZzsEf7wCfZu21/kguvsyXO4evEaAoMWSRgpfS6UVnBNnz4dK1aswNy5c9G8eXOxXVNTE2vXroWLi0uRxgkICMCUKVMU2iZPngx/f/+SDLfYDI0MoK6hjvj3/sKMj02AMaelmZ8iYI7yZ1XJCoZGBnjx9IVCwWVobAhDY0NUsq2EynaV0bNNH9y5cRcutZxhYmoCDQ0NsdgC3p56A4CYqBiVL7g0NTVR0cYaAODgXB33bj/Ajr93oWnLxsjMzELy62SFWa742HiYvDc7+uRRBMYP+wVtvm6NXt/3UFgm05Zh7ORRGPnLj4iPS4CJqTH2bz+I8ro6MDQ2lH4HS5iJqbH4+81R2c4Gp4/9W+QxQi9eQ+SzSHTy7KbQPnX8dNSsUwNzV8wskVjp86C0a7h+/vln/PPPPxg2bBjGjh2LzMzMDxrHz88PiYmJCg8/P78Sjrb4NDU14eBUDVcvXhPb5HI5rl4MhYurkxIjUw3MT+GYo/y9jH6FpMTXBV5TI8jlAN5emwQANdxckJ2djRdPI8U+zyKeAwAsrMwljFYaglyOjIxMODhXR7ly5XD1Qqi47Gn4M8REvYRzLWexLTzsCcYO+Rne7VpggG/ffMctp1kOZham0NDQwPHDJ9GwUQOoq6vc5b6FquHmgmdPniu0PYt4Xqzfdfd+3bB80xIEBi0WHwAwdPQghU/MEhWFUj+lWL9+fVy+fBm+vr6oV68eNm7cWOxPKKry6cMuvTtj1uR5cHCuDseaDtgRtAtpqenw6dBS2aGpBOancJ9LjlLfpOL50xfi86gX0Xh4LwwGBvrQN9TH+hVBaNTiK5hUMMaLZ5FYtXA1rG2sUM/DHQBw58Zd3Lv9ADVru0DfQA8vnkZibeB6WFeyEouOug1ro7pTNcyZOh8/jBkCuSDHohlLUbdhHYVZL1X056I1qP9VPZhbmiM15Q2OHTyBa5dvIGDxNOjq66JVR28EzlsJfQN9lNcrjyWzAuFSy1kszB8/DMf4oX5w96iLLr06I+5VHABAXUMDRv8/e/XsyTPcvXUfTjUdkZyUjG0bdyA87AnGTxmjtP3+GF16dcZP/ccgaPU/aNqyMe7dvIf92w9g5P9GiH2SEl8jJioGsS/f5uPZk2cAAJMKxjAxNREf7zO3NPtkPuFKqkPpt4XQ09PDunXrsGnTJnh5eSE7O1vZIZUYT++mSIhPwrrA9YiPjYe9Q1VMXzQ110XQnyvmp3CfS47u336AsUN+Fp8HzlsJAGjZzgs/+fni0YPHCN57BMmvU1DBzATuX9RFv2HfiZ+q09aW4d9j/+Kv5RuQlpqGCqYmqOfhjl4zuot91NXVMXXBZCyZtQyjB42Hto426n/pjiGjBpX+DhdTQnwiZk2ai7hXcdDV04VddTsELJ4G9y/eXoM1bMxgqKmrYer435GZkQl3D3eM+PkHcf3TR88gIT4RR/cfx9H9x8V2CytzbNi7FgCQLZdj64bteBb+HBrlNFC7Xi0sXD0XltYWpbqvJcWxhgP85/yKPxevxYaVQbC0tsSwMUPQok0zsU/IyXOYM2W++Px3v7enCL8b3BN9hvD2D1Sy1ARBEJQdRI5nz57h8uXL8PLygq6u7kePF5EcVgJRlT2V9f67kzJzlBvzUzDmp3DMUcGYn4K9m5/ScPT5/lLbVouKbUptW6pG6TNc76pUqRIqVVLtqX0iIiKi4vr0roQkIiIi+sSw4CIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4CIiIiKV4+/vDzU1NYWHk9N/3yOblpYGX19fVKhQAXp6eujSpQuio6MVxoiIiEDbtm1Rvnx5mJubY9y4ccjKylLoc+LECdStWxcymQzVqlXD2rVrJdkfFlxERESkkmrUqIHIyEjxcebMGXHZqFGjsGfPHmzZsgUnT57Eixcv8PXXX4vLs7Oz0bZtW2RkZODs2bNYt24d1q5di0mTJol9Hj9+jLZt26JZs2YIDQ3FyJEj8f333+PQoUMlvi8qdad5IiIiohzlypWDpWXuLwpPTEzEn3/+iaCgIDRv3hwAsGbNGjg7O+PcuXP44osvcPjwYdy+fRtHjhyBhYUFateujWnTpmHChAnw9/eHlpYWAgMDYWdnh7lz5wIAnJ2dcebMGcyfPx8+Pj4lui+c4SIiIqJSkZ6ejqSkJIVHenp6vv0fPHgAa2trVK1aFb169UJERAQA4PLly8jMzISXl5fY18nJCZUrV0ZISAgAICQkBK6urrCw+O8L2H18fJCUlIRbt26Jfd4dI6dPzhgliQUXERERlYqAgAAYGhoqPAICAvLs27BhQ6xduxYHDx7EsmXL8PjxYzRu3BivX79GVFQUtLS0YGRkpLCOhYUFoqKiAABRUVEKxVbO8pxlBfVJSkpCampqSeyyiKcUiYiIqFT4+flh9OjRCm0ymSzPvq1btxZ/rlWrFho2bAhbW1ts3rwZOjo6ksYpBc5wERERUamQyWQwMDBQeORXcL3PyMgIDg4OePjwISwtLZGRkYGEhASFPtHR0eI1X5aWlrk+tZjzvLA+BgYGJV7UseAiIiIilZecnIywsDBYWVnB3d0dmpqaOHr0qLj83r17iIiIgIeHBwDAw8MDN27cQExMjNgnODgYBgYGcHFxEfu8O0ZOn5wxShILLiIiIlI5Y8eOxcmTJxEeHo6zZ8+ic+fO0NDQQI8ePWBoaIiBAwdi9OjROH78OC5fvoz+/fvDw8MDX3zxBQDA29sbLi4u+O6773Dt2jUcOnQIv/76K3x9fcVZtaFDh+LRo0cYP3487t69i6VLl2Lz5s0YNWpUie8Pr+EiIiIilfPs2TP06NEDsbGxMDMzQ6NGjXDu3DmYmZkBAObPnw91dXV06dIF6enp8PHxwdKlS8X1NTQ0sHfvXgwbNgweHh7Q1dVF3759MXXqVLGPnZ0d9u3bh1GjRmHhwoWoVKkSVq1aVeK3hAAANUEQhBIfVUVEJIcpOwSVVFnPXvyZOcqN+SkY81M45qhgzE/B3s1PaTj6fH+pbatFxTalti1Vw1OKRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBJjwUVEREQkMRZcRERERBIr09+lSERERAXjdymWDs5wEREREUmMBRcRERGRxMopOwApRSSHKTsElVRZz178mTnKjfkpGPNTOOaoYO/mZ/61OUqMRDWNchur7BBIApzhIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpIYCy4iIiIiibHgIiIiIpJYOWUHUNbt2rwHW/7ahrjYeNhXt4Pv+GFwqumo7LBUBvNTOOaoYJ9rfnq364foyJhc7e27tcWIn32xb/sBHDt4Ag/vPsSblFTsOLEZevp6ufqfP30BG1YG4dHDcGhpaaFW3ZqYMm9SaexCidrg+zeSXybnaq/h7YLG338lPhcEAfsDDuJp6DP4jG0JuwZVxGVnVp9F1L1oxD2Ng3FFI3Sb3SXXeLFPYnH6z3/xMuwVtA20UbNVDdTp6CbJPlHZwoJLQicOn8TyeSsx4pfhcK7phO1BO+E3fCJWb18BYxMjZYendMxP4Zijgn3O+Vm8fiHk2dni8/CwJ5jww//Q1KsxACA9LR31PdxR38Mdfy5em+cYp4+ewfzf/kB/376oU98N2dlyhD8ML4XoS16XgE4Q5IL4PC4iHnt/24+qHnYK/a7vuwmoqeU7jlMzB8Q8jEHsk7hcyzLeZGDvbwdQybUimgxqhLiIeJxYdhIyXS24eDmX3M5QmcRTihLatmEHWnduhVYdvGFbtTJ++mU4ZNoyHNp1WNmhqQTmp3DMUcE+5/wYGRvCxNREfJw7fQHWlaxQy90VAPB1z07o3v8bOLs65bl+dlY2ls5ZjkE/DUT7rm1RybYSbKtWRlPvJqW5GyVGx0AH5Y3Ki48nVyJgYGEAaxcrsc+r8Fhc33sDzYblvY+NBnyJmq1qQN/cIM/lD848hDxLDs8fmsDExgTVvrJHzdY1cX3vDUn2icoWFlwSyczMxP27D1G3QW2xTV1dHXUb1MbtG3eVF5iKYH4KxxwVjPn5T2ZmJo7uPw6fjt5QK2D25l0P7j7Eq5hYqKmrYWjP4fjWuxd++XEiHn+iM1zvys7KxoPTD+DUzEHMR2Z6Fo4uPIZGA79EeaPyHzRu9P1oWDlbQqOchthm41YJCS8SkZ6cXiKxU9nFgksiiQlJkGfLYVzBWKHduIIR4l/lnqr+3DA/hWOOCsb8/Ofs8RAkJyfDu71XkdeJfB4FAFi/fCN6DeyOaQv9oaevh7GDf0ZS4mupQi0Vjy+EIz0lA46eDmLb2XUhsHC0gF39Kh887puEVOgY6Si05Tx/k/Dmg8elz4NKXcOVkpKCzZs34+HDh7CyskKPHj1QoUKFAtdJT09HerriXxYymQwymUzKUImIVMaBXYfR4Mt6MDUr+P3yXYJcDgDoObA7GrdoBAAY6z8aPVt/h1NHTqNdlzaSxFoa7h6/h8q1baBrogsACL/0BM9vvkC3WV8rOTLVVN2w7H/IRBUodYbLxcUFcXFv/xJ9+vQpatasiVGjRiE4OBiTJ0+Gi4sLHj9+XOAYAQEBMDQ0VHgEBASURvgFMjQygLqGOuJj4xXa42MTYGxqoqSoVAfzUzjmqGDMz1vRkdG4eiEUrTv5FGs9k//Pka1dZbFNS0sTVhUtERP1skRjLE2vX77G8+sv4NTivyLi+c0XSIpOwup+67C8+yos774KAHB47hHs8t9b5LHLG+kgNSFVoS3n+YeepqTPh1ILrrt37yIrKwsA4OfnB2trazx58gQXLlzAkydPUKtWLfzvf/8rcAw/Pz8kJiYqPPz8/Eoj/AJpamrCwakarl68JrbJ5XJcvRgKl3wuYv2cMD+FY44Kxvy8dWh3MIyMDdGwUYNirVfduTo0tTTx9MkzsS0rMwtRkTGwsDIv6TBLzd3j96FjqA3buv8VknU6ueGb2V3QbdbX4gMAvuz7BZr9UPQPCVg4WCDyThSys+Ri27Prz2FkbQiZHs+qUMFU5pRiSEgIAgMDYWhoCADQ09PDlClT0L179wLXU+XTh116d8asyfPg4FwdjjUdsCNoF9JS0+HToaWyQ1MJzE/hmKOCfe75kcvlOLQ7GC3beSlcyA0Aca/iEBcbj+dPXwAAHj8Mh055HZhbmsPAUB+6euXRrksb/LV8A8wszGBhZY7Nf20FADTxalTq+1ISBLmAeyfuw6GpA9Q1/ptPyPnk4vv0TPVg8M4nEhOjEpGZloXUhDfIysjGq/BYAIBxJSNolNNAtUbVcGnLFZwMPInaHd0Q9zQeNw7cxJd9v5B+5+iTp/SCK+cTJGlpabCyslJYVrFiRbx8+elObXt6N0VCfBLWBa5HfGw87B2qYvqiqbku8v1cMT+FY44K9rnn58r5UMREvUSrjrkLzL3b9mP9iiDx+ejvxwMAxk4eJRakg38aCA0NDcycNAcZ6elwqumI2YEB0DfQL50dKGHPbjxH8qtkODVzKLxzHk4Enkbk7Ujx+dbx2wEAPRd3h4G5PmTltdDu19Y4/ee/2PbzTmjra8O9Sx3eg4uKRE0QBKHwbtJQV1dHzZo1Ua5cOTx48ABr165Fly7/3dn31KlT6NmzJ549e1bAKPmLSA4rqVDLlMp69uLPzFFuzE/BmJ/CMUcFezc/86/NUWIkqmmU29hS3V5pHqPv/u4/N0qd4Zo8ebLCcz09xa+d2LNnDxo3blyaIRERERGVOJUquN43e/bsUoqEiIiISDq88SkRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGRxFhwEREREUmMBRcRERGpnICAANSvXx/6+vowNzdHp06dcO/ePYU+np6eUFNTU3gMHTpUoU9ERATatm2L8uXLw9zcHOPGjRO/xznHiRMnULduXchkMlSrVg1r164t8f1hwUVEREQq5+TJk/D19cW5c+cQHByMzMxMeHt7IyUlRaHfoEGDEBkZKT5mzZolLsvOzkbbtm2RkZGBs2fPYt26dVi7di0mTZok9nn8+DHatm2LZs2aITQ0FCNHjsT333+PQ4cOlej+KP27FImIiIjed/DgQYXna9euhbm5OS5fvowmTZqI7eXLl4elpWWeYxw+fBi3b9/GkSNHYGFhgdq1a2PatGmYMGEC/P39oaWlhcDAQNjZ2WHu3LkAAGdnZ5w5cwbz58+Hj49Pie0PZ7iIiIioVKSnpyMpKUnhkZ6eXqR1ExMTAQAmJiYK7Rs3boSpqSlq1qwJPz8/vHnzRlwWEhICV1dXWFhYiG0+Pj5ISkrCrVu3xD5eXl4KY/r4+CAkJOSD9jE/LLiIiIioVAQEBMDQ0FDhERAQUOh6crkcI0eOxFdffYWaNWuK7T179sSGDRtw/Phx+Pn5Yf369ejdu7e4PCoqSqHYAiA+j4qKKrBPUlISUlNTP3hf38dTikRERFQq/Pz8MHr0aIU2mUxW6Hq+vr64efMmzpw5o9A+ePBg8WdXV1dYWVmhRYsWCAsLg729fckEXUI4w0VERESlQiaTwcDAQOFRWME1fPhw7N27F8ePH0elSpUK7NuwYUMAwMOHDwEAlpaWiI6OVuiT8zznuq/8+hgYGEBHR6foO1cIFlxERESkcgRBwPDhw7Fjxw4cO3YMdnZ2ha4TGhoKALCysgIAeHh44MaNG4iJiRH7BAcHw8DAAC4uLmKfo0ePKowTHBwMDw+PEtqTt1hwERERkcrx9fXFhg0bEBQUBH19fURFRSEqKkq8riosLAzTpk3D5cuXER4ejt27d6NPnz5o0qQJatWqBQDw9vaGi4sLvvvuO1y7dg2HDh3Cr7/+Cl9fX3FmbejQoXj06BHGjx+Pu3fvYunSpdi8eTNGjRpVovvDgouIiIhUzrJly5CYmAhPT09YWVmJj3/++QcAoKWlhSNHjsDb2xtOTk4YM2YMunTpgj179ohjaGhoYO/evdDQ0ICHhwd69+6NPn36YOrUqWIfOzs77Nu3D8HBwXBzc8PcuXOxatWqEr0lBMCL5omIiEgFCYJQ4HIbGxucPHmy0HFsbW2xf//+Avt4enri6tWrxYqvuDjDRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxFlxEREREEmPBRURERCQxNaGwLysiIiKiMisiOazUtlVZz77UtqVqOMMlsfT0dPj7+yM9PV3Zoags5qhgzE/BmJ/CMUcFY36oNHCGS2JJSUkwNDREYmIiDAwMlB2OSmKOCsb8FIz5KRxzVLDPPT+c4SodnOEiIiIikhgLLiIiIiKJseAiIiIikhgLLonJZDJMnjwZMplM2aGoLOaoYMxPwZifwjFHBWN+qDTwonkiIqLPGC+aLx2c4SIiIiKSGAsuIiIiIomx4CIiIiKSGAsuIiIiIomx4JLYkiVLUKVKFWhra6Nhw4a4cOGCskNSGadOnUL79u1hbW0NNTU17Ny5U9khqZSAgADUr18f+vr6MDc3R6dOnXDv3j1lh6Uyli1bhlq1asHAwAAGBgbw8PDAgQMHlB2WypoxYwbU1NQwcuRIZYeiMvz9/aGmpqbwcHJyUnZYVEax4JLQP//8g9GjR2Py5Mm4cuUK3Nzc4OPjg5iYGGWHphJSUlLg5uaGJUuWKDsUlXTy5En4+vri3LlzCA4ORmZmJry9vZGSkqLs0FRCpUqVMGPGDFy+fBmXLl1C8+bN0bFjR9y6dUvZoamcixcvYvny5ahVq5ayQ1E5NWrUQGRkpPg4c+aMskOiMoq3hZBQw4YNUb9+fSxevBgAIJfLYWNjgx9//BE///yzkqNTLWpqatixYwc6deqk7FBU1suXL2Fubo6TJ0+iSZMmyg5HJZmYmGD27NkYOHCgskNRGcnJyahbty6WLl2K3377DbVr18aCBQuUHZZK8Pf3x86dOxEaGqrsUJSKt4UoHZzhkkhGRgYuX74MLy8vsU1dXR1eXl4ICQlRYmT0qUpMTATwtqggRdnZ2di0aRNSUlLg4eGh7HBUiq+vL9q2bavwXkT/efDgAaytrVG1alX06tULERERyg6Jyqhyyg6grHr16hWys7NhYWGh0G5hYYG7d+8qKSr6VMnlcowcORJfffUVatasqexwVMaNGzfg4eGBtLQ06OnpYceOHXBxcVF2WCpj06ZNuHLlCi5evKjsUFRSw4YNsXbtWjg6OiIyMhJTpkxB48aNcfPmTejr6ys7PCpjWHARfQJ8fX1x8+ZNXl/yHkdHR4SGhiIxMRFbt25F3759cfLkSRZdAJ4+fYqffvoJwcHB0NbWVnY4Kql169biz7Vq1ULDhg1ha2uLzZs387Q0lTgWXBIxNTWFhoYGoqOjFdqjo6NhaWmppKjoUzR8+HDs3bsXp06dQqVKlZQdjkrR0tJCtWrVAADu7u64ePEiFi5ciOXLlys5MuW7fPkyYmJiULduXbEtOzsbp06dwuLFi5Geng4NDQ0lRqh6jIyM4ODggIcPHyo7FCqDeA2XRLS0tODu7o6jR4+KbXK5HEePHuU1JlQkgiBg+PDh2LFjB44dOwY7Oztlh6Ty5HI50tPTlR2GSmjRogVu3LiB0NBQ8VGvXj306tULoaGhLLbykJycjLCwMFhZWSk7FCqDOMMlodGjR6Nv376oV68eGjRogAULFiAlJQX9+/dXdmgqITk5WeEvycePHyM0NBQmJiaoXLmyEiNTDb6+vggKCsKuXbugr6+PqKgoAIChoSF0dHSUHJ3y+fn5oXXr1qhcuTJev36NoKAgnDhxAocOHVJ2aCpBX18/1/V+urq6qFChAq8D/H9jx45F+/btYWtrixcvXmDy5MnQ0NBAjx49lB0alUEsuCT07bff4uXLl5g0aRKioqJQu3ZtHDx4MNeF9J+rS5cuoVmzZuLz0aNHAwD69u2LtWvXKikq1bFs2TIAgKenp0L7mjVr0K9fv9IPSMXExMSgT58+iIyMhKGhIWrVqoVDhw6hZcuWyg6NPhHPnj1Djx49EBsbCzMzMzRq1Ajnzp2DmZmZskOjMoj34SIiIvqM8T5cpYPXcBERERFJjAUXERERkcRYcBERERFJjAUXERERkcRYcBERERFJjAUXERERkcRYcBERERFJjAUXERERkcRYcBFRgfr164dOnTqJzz09PTFy5MhSj+PEiRNQU1NDQkJCqW+biOhjseAi+kT169cPampqUFNTg5aWFqpVq4apU6ciKytL0u1u374d06ZNK1JfFklERG/xuxSJPmGtWrXCmjVrkJ6ejv3798PX1xeamprw8/NT6JeRkQEtLa0S2aaJiUmJjENE9DnhDBfRJ0wmk8HS0hK2trYYNmwYvLy8sHv3bvE04O+//w5ra2s4OjoCAJ4+fYpvvvkGRkZGMDExQceOHREeHi6Ol52djdGjR8PIyAgVKlTA+PHj8f7Xrb5/SjE9PR0TJkyAjY0NZDIZqlWrhj///BPh4eHil5MbGxtDTU1N/NJtuVyOgIAA2NnZQUdHB25ubti6davCdvbv3w8HBwfo6OigWbNmCnESEX1qWHARlSE6OjrIyMgAABw9ehT37t1DcHAw9u7di8zMTPj4+EBfXx+nT5/Gv//+Cz09PbRq1UpcZ+7cuVi7di1Wr16NM2fOIC4uDjt27Chwm3369MHff/+NP/74A3fu3MHy5cuhp6cHGxsbbNu2DQBw7949REZGYuHChQCAgIAA/PXXXwgMDMStW7cwatQo9O7dGydPngTwtjD8+uuv0b59e4SGhuL777/Hzz//LFXaiIgkx1OKRGWAIAg4evQoDh06hB9//BEvX76Erq4uVq1aJZ5K3LBhA+RyOVatWgU1NTUAwJo1a2BkZIQTJ07A29sbCxYsgJ+fH77++msAQGBgIA4dOpTvdu/fv4/NmzcjODgYXl5eAICqVauKy3NOP5qbm8PIyAjA2xmx6dOn48iRI/Dw8BDXOXPmDJYvX46mTZti2bJlsLe3x9y5cwEAjo6OuHHjBmbOnFmCWSMiKj0suIg+YXv37oWenh4yMzMhl8vRs2dP+Pv7w9fXF66urgrXbV27dg0PHz6Evr6+whhpaWkICwtDYmIiIiMj0bBhQ3FZuXLlUK9evVynFXOEhoZCQ0MDTZs2LXLMDx8+xJs3b9CyZUuF9oyMDNSpUwcAcOfOHYU4AIjFGRHRp4gFF9EnrFmzZli2bBm0tLRgbW2NcuX+e0nr6uoq9E1OToa7uzs2btyYaxwzM7MP2r6Ojk6x10lOTgYA7Nu3DxUrVlRYJpPJPigOIiJVx4KL6BOmq6uLatWqFalv3bp18c8//8Dc3BwGBgZ59rGyssL58+fRpEkTAEBWVhYuX76MunXr5tnf1dUVcrkcJ0+eFE8pvitnhi07O1tsc3FxgUwmQ0RERL4zY87Ozti9e7dC27lz5wrfSSIiFcWL5ok+E7169YKpqSk6duyI06dP4/Hjxzhx4gRGjBiBZ8+eAQB++uknzJgxAzt37sTdu3fxww8/FHgPrSpVqqBv374YMGAAdu7cKY65efNmAICtrS3U1NSwd+9evHz5EsnJydDX18fYsWMxatQorFu3DmFhYbhy5QoWLVqEdevWAQCGDh2KBw8eYNy4cbh37x6CgoKwdu1aqVNERCQZFlxEn4ny5cvj1KlTqFy5Mr7++ms4Oztj4MCBSEtLE2e8xowZg++++w59+/aFh4cH9PX10blz5wLHXbZsGbp27YoffvgBTk5OGDRoEFJSUgAAFStWxJQpU/Dzzz/DwsICw4cPBwBMmzYNEydOREBAAJydndGqVSvs27cPdnZ2AIDKlStj27Zt2LlzJ9zc3BAYGIjp06dLmB0iImmpCfldDUtERERlXkRyWKltq7KefaltS9VwhouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhIYiy4iIiIiCTGgouIiIhU1pIlS1ClShVoa2ujYcOGuHDhgrJD+iAsuIiIiEgl/fPPPxg9ejQmT56MK1euwM3NDT4+PoiJiVF2aMXGG58SERF9xlT5xqcNGzZE/fr1sXjxYgCAXC6HjY0NfvzxR/z8889ShCgZznARERGRysnIyMDly5fh5eUltqmrq8PLywshISFKjOzDlFN2AERERP/X3h2jKAxEARh+IDidlRCPISiCJ7EUr5DORuMNPI5N7uERDNpapFC2WwjuFgs72V32+yDNI8Urf4Yh4X9o2zbatu3MUkqRUnp593a7xePxiKIoOvOiKOJ8PmfdMwfBBQD/WJ//N6yqKg6HQ2e23++jqqredvgpggsA6MV2u42yLDuzj063IiLG43EMBoNomqYzb5omJpNJth1zcYcLAOhFSilGo1Hn+Sy4hsNhzGazqOv6ffZ8PqOu61gul32t/G2ccAEAv1JZlrFer2M+n8disYjj8Rj3+z02m81Pr/ZlggsA+JVWq1Vcr9fY7XZxuVxiOp3G6XR6uUj/F/gOFwBAZu5wAQBkJrgAADITXAAAmQkuAIDMBBcAQGaCCwAgM8EFAJCZ4AIAyExwAQBkJrgAADITXAAAmQkuAIDM3gCKuVDSiQCdWAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 700x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create and train decision tree with default model\n",
    "clf = tree.DecisionTreeClassifier()\n",
    "clf = clf.fit(X_train, y_train)\n",
    "\n",
    "# test\n",
    "y_pred = clf.predict(X_test).round()\n",
    "print(\"Default Decision Tree Model Description\")\n",
    "print(\"Depth: %d\" % clf.get_depth())\n",
    "print(\"Number of Leaves: %d\" % clf.get_n_leaves())\n",
    "print(\"*******************************************\")\n",
    "print(\"Classification Report of the default model\")\n",
    "print(classification_report(y_test, y_pred))\n",
    "\n",
    "#vizualizing classification report\n",
    "cmap =sns.color_palette(\"Greens\")\n",
    "cm = confusion_matrix(y_test, y_pred)\n",
    "plt.figure(figsize=(7,7))\n",
    "sns.heatmap(cm, annot=True, fmt='g', cmap=cmap, \n",
    "            annot_kws={'fontsize':10}, linewidth=1, square=True)\n",
    "plt.xlabel(\"Predicted\")\n",
    "plt.ylabel(\"Actual\")\n",
    "plt.title('Decision Tree Default Parameter Model Classification')\n",
    "plt.show()\n",
    "\n",
    "# saving the values in variables to export them later\n",
    "report_under_default = classification_report(y_test, y_pred,output_dict=True)\n",
    "macro_f1_under_default = report_under_default['macro avg']['f1-score']\n",
    "weighted_f1_under_default = report_under_default['weighted avg']['f1-score']\n",
    "class_0_under_default = report_under_default['0']['f1-score']\n",
    "class_1_under_default = report_under_default['1']['f1-score']\n",
    "class_2_under_default = report_under_default['2']['f1-score']\n",
    "class_3_under_default = report_under_default['3']['f1-score']\n",
    "class_4_under_default = report_under_default['4']['f1-score']\n",
    "class_5_under_default = report_under_default['5']['f1-score']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "eee7d6cc",
   "metadata": {},
   "source": [
    "##### 7.3 Undersampled Dataset: Hyper Parameter Tuning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "b3b7b7b6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best hyperparameters for decision tree classification:  {'criterion': 'gini', 'max_depth': 53, 'min_samples_split': 40}\n"
     ]
    }
   ],
   "source": [
    "#specify the value of parameters of interest\n",
    "tree_params = {'criterion':['gini','entropy'],'max_depth': list(range(45, 56)),'min_samples_split':list(range(2, 10))+np.arange(10, 100, 10).tolist()}\n",
    "\n",
    "#determine the best hyper parameter\n",
    "dtc_under_top = GridSearchCV(tree.DecisionTreeClassifier(), tree_params, cv=5)\n",
    "\n",
    "# Training the model for classification with the same dataset \n",
    "model2 = dtc_under_top.fit(X_train, y_train)\n",
    "\n",
    "print(\"Best hyperparameters for decision tree classification: \", dtc_under_top.best_params_)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "7c2e2f05",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification Report of the default model\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.77      0.83      0.80     22940\n",
      "           1       0.68      0.60      0.64     14064\n",
      "           2       1.00      1.00      1.00     10240\n",
      "           3       0.71      0.79      0.75      5454\n",
      "           4       0.62      0.52      0.56      5179\n",
      "           5       0.91      0.93      0.92      8126\n",
      "\n",
      "    accuracy                           0.79     66003\n",
      "   macro avg       0.78      0.78      0.78     66003\n",
      "weighted avg       0.79      0.79      0.79     66003\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 700x700 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#saving the model with the best performing hyperparameters\n",
    "filename = \"models/DT_undersample_model.pickle\"\n",
    "\n",
    "#train the model with the best hyperparameters\n",
    "model_under_final = tree.DecisionTreeClassifier(criterion = \"gini\", max_depth = 53, min_samples_split = 40)\n",
    "model_under_final.fit(X_train, y_train)\n",
    "\n",
    "# save model\n",
    "pickle.dump(model_under_final, open(filename, \"wb\"))\n",
    "\n",
    "# Measure performance of the model with the best hyper parameter \n",
    "y_top_pred = model_under_final.predict(X_test)\n",
    "print(\"Classification Report of the default model\")\n",
    "print(classification_report(y_test, y_top_pred))\n",
    "\n",
    "#vizualizing classification report\n",
    "cmap =sns.color_palette(\"Greens\")\n",
    "cm = confusion_matrix(y_test, y_top_pred)\n",
    "plt.figure(figsize=(7,7))\n",
    "sns.heatmap(cm, annot=True, fmt='g', cmap=cmap, \n",
    "            annot_kws={'fontsize':10}, linewidth=1, square=True)\n",
    "plt.xlabel(\"Predicted\")\n",
    "plt.ylabel(\"Actual\")\n",
    "plt.title('Decision Tree Parameter Tuning Model Classification')\n",
    "plt.show()\n",
    "\n",
    "# saving the values in variables to export them later\n",
    "report_under_top = classification_report(y_test, y_top_pred,output_dict=True)\n",
    "macro_f1_under_top = report_under_top['macro avg']['f1-score']\n",
    "weighted_f1_under_top = report_under_top['weighted avg']['f1-score']\n",
    "class_0_under_top = report_under_top['0']['f1-score']\n",
    "class_1_under_top = report_under_top['1']['f1-score']\n",
    "class_2_under_top = report_under_top['2']['f1-score']\n",
    "class_3_under_top = report_under_top['3']['f1-score']\n",
    "class_4_under_top = report_under_top['4']['f1-score']\n",
    "class_5_under_top = report_under_top['5']['f1-score']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "e6c43ea7",
   "metadata": {},
   "source": [
    "#### 8. Cross Validation\n",
    "\n",
    "Cross validation helps us better understand the performance of our model.\n",
    "<br>In this section, we use <b><i>k-fold cross validation</i></b> method to evaluate our model.\n",
    "<br>K-fold cross validation splits the dataset into a given <i>k</i> number of folds.\n",
    "<br>Each of these folds are used as a test set whilst other k-1 folds are used to train the model.\n",
    "<br>scikit-learn's <i>KFold</i> library is used to implement this algorithm.\n",
    "<br>Here, we test the validity of the model's performance with 8 and 10 fold cross validations."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "5d77ceaa",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Testing with 8 fold:\n",
      "Accuracy: 0.784 (0.001)\n",
      "Testing with 10 fold:\n",
      "Accuracy: 0.784 (0.001)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from numpy import mean\n",
    "from numpy import std\n",
    "\n",
    "# load dataset\n",
    "df = pd.read_csv('preprocessing/data/london_clean.csv')\n",
    "\n",
    "# train and test split\n",
    "X = df[['DateOfCall', 'PropertyType', 'NumPumpsAttending', 'PumpHoursRoundUp', 'mean_temp']]\n",
    "y = df[['CostCat']]\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)\n",
    "\n",
    "#load model\n",
    "loaded_model = pickle.load(open(\"models/DT_model.pickle\", 'rb'))\n",
    "\n",
    "folds = [8, 10]\n",
    "for i in folds:\n",
    "    cross_val = KFold(n_splits=i, random_state=42, shuffle=True)\n",
    "    scores = cross_val_score(loaded_model, X, y, scoring='accuracy', cv=cross_val, n_jobs=4)\n",
    "    print(\"Testing with {} fold:\".format(i))\n",
    "    print('Accuracy: %.3f (%.3f)' % (mean(scores), std(scores)))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "7918f1eb",
   "metadata": {},
   "source": [
    "#### 9. Export Result for Comparison\n",
    "<br> The performance of all decision tree models (default parameter, hyper parameter tuned, undersampled with default parameter, undersampled with hyper parameter tuned) is recorded and arranged into pandas' dataframes.\n",
    "<ol>\n",
    "<li> Dataframe 1: weighted f1-score of each model\n",
    "<li> Dataframe 2: f1-score of each class for each model\n",
    "</ol>\n",
    "<br>The dataframes are exported and saved into csv files for easier access when comparing the performance between different models implemented in this project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "7fa071e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "filename_model ='test_results/DT_model_result.csv'\n",
    "filename_class = 'test_results/DT_class_result.csv'\n",
    "\n",
    "#overall f1-score result exporting\n",
    "model_d = {'model':['DT_default', 'DT_top', 'DT_under_default', 'dtc_under_top'], 'macro_f1-score': [macro_f1_default, macro_f1_top, macro_f1_under_default, macro_f1_under_top], \n",
    "           'weighted_f1-score': [weighted_f1_default, weighted_f1_top, weighted_f1_under_default, weighted_f1_under_top]}\n",
    "df_model_summary = pd.DataFrame(data=model_d)\n",
    "df_model_summary.to_csv(filename_model)\n",
    "\n",
    "#f1-score by class result exporting\n",
    "class_d = {'class':['0', '1', '2', '3', '4', '5', '0', '1', '2', '3', '4', '5', '0', '1', '2', '3', '4', '5', '0', '1', '2', '3', '4', '5'], \n",
    "           'model':['DT_default','DT_default','DT_default','DT_default','DT_default','DT_default','DT_top','DT_top','DT_top','DT_top','DT_top','DT_top',\n",
    "                    'DT_under_default','DT_under_default','DT_under_default','DT_under_default','DT_under_default','DT_under_default',\n",
    "                    'DT_under_top','DT_under_top','DT_under_top','DT_under_top','DT_under_top','DT_under_top'],\n",
    "           'f1-score': [class_0_default, class_1_default, class_2_default, class_3_default, class_4_default, class_5_default,\n",
    "                        class_0_top, class_1_top, class_2_top, class_3_top, class_4_top, class_5_top,\n",
    "                        class_0_under_default, class_1_under_default, class_2_under_default, class_3_under_default, class_4_under_default, class_5_under_default,\n",
    "                        class_0_under_top, class_1_under_top, class_2_under_top, class_3_under_top, class_4_under_top, class_5_under_top]}\n",
    "df_class_summary = pd.DataFrame(data=class_d)\n",
    "df_class_summary.to_csv(filename_class)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "2912d884",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>macro_f1-score</th>\n",
       "      <th>weighted_f1-score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.782033</td>\n",
       "      <td>0.774581</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.778997</td>\n",
       "      <td>0.777630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.773774</td>\n",
       "      <td>0.778843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>dtc_under_top</td>\n",
       "      <td>0.779106</td>\n",
       "      <td>0.788770</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              model  macro_f1-score  weighted_f1-score\n",
       "0        DT_default        0.782033           0.774581\n",
       "1            DT_top        0.778997           0.777630\n",
       "2  DT_under_default        0.773774           0.778843\n",
       "3     dtc_under_top        0.779106           0.788770"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_model_summary"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "68c8e6ff",
   "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>class</th>\n",
       "      <th>model</th>\n",
       "      <th>f1-score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.812891</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.656805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.734690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.569415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>DT_default</td>\n",
       "      <td>0.918394</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.815727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.662191</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>2</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>3</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.747551</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>4</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.527591</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>5</td>\n",
       "      <td>DT_top</td>\n",
       "      <td>0.920919</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.786117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.611261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>2</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>3</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.738609</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>4</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.589013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>5</td>\n",
       "      <td>DT_under_default</td>\n",
       "      <td>0.917647</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>0.798779</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>1</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>0.639112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>2</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>3</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>0.750000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>4</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>0.564334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>5</td>\n",
       "      <td>DT_under_top</td>\n",
       "      <td>0.922413</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   class             model  f1-score\n",
       "0      0        DT_default  0.812891\n",
       "1      1        DT_default  0.656805\n",
       "2      2        DT_default  1.000000\n",
       "3      3        DT_default  0.734690\n",
       "4      4        DT_default  0.569415\n",
       "5      5        DT_default  0.918394\n",
       "6      0            DT_top  0.815727\n",
       "7      1            DT_top  0.662191\n",
       "8      2            DT_top  1.000000\n",
       "9      3            DT_top  0.747551\n",
       "10     4            DT_top  0.527591\n",
       "11     5            DT_top  0.920919\n",
       "12     0  DT_under_default  0.786117\n",
       "13     1  DT_under_default  0.611261\n",
       "14     2  DT_under_default  1.000000\n",
       "15     3  DT_under_default  0.738609\n",
       "16     4  DT_under_default  0.589013\n",
       "17     5  DT_under_default  0.917647\n",
       "18     0      DT_under_top  0.798779\n",
       "19     1      DT_under_top  0.639112\n",
       "20     2      DT_under_top  1.000000\n",
       "21     3      DT_under_top  0.750000\n",
       "22     4      DT_under_top  0.564334\n",
       "23     5      DT_under_top  0.922413"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_class_summary"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "c5dc88d7",
   "metadata": {},
   "source": [
    "#### 10. Conclusion\n",
    "\n",
    "Decision tree model with default parameter performed fairly well with weighted f1-score of 0.77.\n",
    "<br>After hyper parameter tuning, we were able to slightly improve its performance, with weighted f1-score of 0.78.\n",
    "<br>An interesting aspect of the decision tree model was the inconsistent performance between each class. \n",
    "<br>For example, class 1 and class 4 are highly misclassified whereas class 2 is perfectly classified.\n",
    "<br>\n",
    "<br>We explored different methods to ensire the validity of the decision tree model's performace.\n",
    "<br>First, we re-trained model with undersampled dataset as the original dataset was highly imbalanced with more than half of the observations classified in class 0 and 1.\n",
    "<br>This method was able to improve the model's performance very slightly again, with the weighted f1-score of 0.79.\n",
    "<br>However, the inconsistent performance between each classes did not improve.\n",
    "<br>Next, we validated the model's performance with k-fold cross-validation method.\n",
    "<br>The tests with 8 folds and 10 folds validation both resulted with 0.78 accuracy.\n",
    "<br>\n",
    "<br>Considering the goal of this project being to estimate the cost of operation of each fire extinguishing operation, we find this scoring acceptable.\n",
    "<br>Nonetheless, there are still room for improvement. \n",
    "<br>Thus, it would be best to explore other models and algorithms to solve this problem."
   ]
  }
 ],
 "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": "3.10.8"
  },
  "vscode": {
   "interpreter": {
    "hash": "41a25e4fc04ae1ac97e89ed19feb20c86178bc3aba5c26b45b17fff7270b24df"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}