02 Anomaly detection using Gaussian distribution.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Anomaly detection - Credit card transactions - using Gaussian distribution "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import libraries\n",
    "\n",
    "# Pandas for data frame\n",
    "import pandas as pd\n",
    "\n",
    "# Numpy for numerical calc\n",
    "import numpy as np\n",
    "\n",
    "# matplotlib for plots\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# seaborn for plots\n",
    "import seaborn as sns\n",
    "\n",
    "# For statistical test\n",
    "from scipy.stats import ttest_ind\n",
    "\n",
    "# Default dictionary\n",
    "from collections import defaultdict\n",
    "\n",
    "# For Gaussian estimation\n",
    "from scipy.stats import multivariate_normal\n",
    "\n",
    "# For metrics\n",
    "from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score, confusion_matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load train, dev and test data, saved from data preparation notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import data\n",
    "df_train = pd.read_csv('C:\\\\Myprojects\\\\Dev\\\\py3\\\\creditcard_train.csv')\n",
    "df_dev = pd.read_csv('C:\\\\Myprojects\\\\Dev\\\\py3\\\\creditcard_dev.csv')\n",
    "df_test = pd.read_csv('C:\\\\Myprojects\\\\Dev\\\\py3\\\\creditcard_test.csv')\n",
    "\n",
    "# split train & dev dataset into Fraud and Non Fraud\n",
    "df_traindevNF = pd.concat([df_train,df_dev[df_dev.Class==0]])\n",
    "df_traindevF  = df_dev[df_dev.Class==1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature importance\n",
    "We have seen from the historgram plot in data preparation notebook that the distribution of certain features differ significantly for Fraud and Non fraud data. This would be key to find the important features for training however we would use t-test to mathematically evaluate the values as most of the distribution looks like a normal ditribution. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Find feature importance using student-t test:\n",
    "\n",
    "Compare fraud data and Non fraud data disttibution using t test. Lower the p value , different are the distribution and critical is to find anomalies in the data.\n",
    "\n",
    "Here we have left out 'time' feature as we dont think it can be factor to identify the fraudulent transaction. The above distribution proves that too."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>columns</th>\n",
       "      <th>pval</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>V14</td>\n",
       "      <td>8.686017e-70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>V4</td>\n",
       "      <td>8.342289e-67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>V11</td>\n",
       "      <td>2.419406e-61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>V12</td>\n",
       "      <td>2.304468e-54</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>V10</td>\n",
       "      <td>2.037737e-46</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>V16</td>\n",
       "      <td>3.081924e-41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>V9</td>\n",
       "      <td>1.462521e-40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>V3</td>\n",
       "      <td>1.766157e-38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>V17</td>\n",
       "      <td>2.509562e-34</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>V2</td>\n",
       "      <td>7.046266e-33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>V6</td>\n",
       "      <td>7.395797e-27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>V7</td>\n",
       "      <td>1.293363e-26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>V18</td>\n",
       "      <td>4.668268e-26</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>V1</td>\n",
       "      <td>4.013193e-22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>V5</td>\n",
       "      <td>7.376158e-19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>V19</td>\n",
       "      <td>6.817791e-12</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>V20</td>\n",
       "      <td>5.780831e-07</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>V28</td>\n",
       "      <td>6.671765e-04</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>V21</td>\n",
       "      <td>6.907884e-03</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>Amount</td>\n",
       "      <td>7.688104e-02</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>V27</td>\n",
       "      <td>1.476592e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>V8</td>\n",
       "      <td>1.511907e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>V24</td>\n",
       "      <td>1.549644e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>V25</td>\n",
       "      <td>2.461138e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>V26</td>\n",
       "      <td>2.504349e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>V15</td>\n",
       "      <td>2.526804e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>V13</td>\n",
       "      <td>3.544793e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>V23</td>\n",
       "      <td>6.697651e-01</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>V22</td>\n",
       "      <td>9.652101e-01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   columns          pval\n",
       "0      V14  8.686017e-70\n",
       "1       V4  8.342289e-67\n",
       "2      V11  2.419406e-61\n",
       "3      V12  2.304468e-54\n",
       "4      V10  2.037737e-46\n",
       "5      V16  3.081924e-41\n",
       "6       V9  1.462521e-40\n",
       "7       V3  1.766157e-38\n",
       "8      V17  2.509562e-34\n",
       "9       V2  7.046266e-33\n",
       "10      V6  7.395797e-27\n",
       "11      V7  1.293363e-26\n",
       "12     V18  4.668268e-26\n",
       "13      V1  4.013193e-22\n",
       "14      V5  7.376158e-19\n",
       "15     V19  6.817791e-12\n",
       "16     V20  5.780831e-07\n",
       "17     V28  6.671765e-04\n",
       "18     V21  6.907884e-03\n",
       "19  Amount  7.688104e-02\n",
       "20     V27  1.476592e-01\n",
       "21      V8  1.511907e-01\n",
       "22     V24  1.549644e-01\n",
       "23     V25  2.461138e-01\n",
       "24     V26  2.504349e-01\n",
       "25     V15  2.526804e-01\n",
       "26     V13  3.544793e-01\n",
       "27     V23  6.697651e-01\n",
       "28     V22  9.652101e-01"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Feature importance using df_traindev and df_devfraud - using t-test\n",
    "features = df_traindevNF.columns[2:-1]\n",
    "val1,val2=ttest_ind(df_traindevNF[features], df_traindevF[features],axis=0, equal_var=False)\n",
    "Feat_imp = pd.DataFrame({'columns':features[np.argsort(val2)],'pval':val2[np.argsort(val2)]})\n",
    "Feat_imp"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The above table arranges the features in the order of its importance. We can now find p-value by comparing dev data distribution against the multivariate gaussian formed with mean and variance/covariance of train data.\n",
    "\n",
    "Note: We've used covariance matrix to evaluate Gaussian (multivariate normal pdf) however it is worth trying variance array first when the data has linearly independent data (like output of PCA here). The advantage is variance array is computationally less expensive however doesnt consider any covariance between different features if there exist which may impact the metrics. Here we have used covariance matrix as we have found considerable improvement in F1-score in our test. Use train_var instead of train_cov in the below function if you would like to compare the result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Find best F1 scores, other scores for given number of features\n",
    "def find_bestF1(no_feat,min_prob=0.0,max_prob=0.0,step=1000,log=True):\n",
    "    impcol = Feat_imp[:no_feat]['columns']\n",
    "    train_mean,train_var,train_cov = df_train[impcol].mean(),df_train[impcol].var(),df_train[impcol].cov()\n",
    "    pval = multivariate_normal(mean=train_mean, cov=train_cov).pdf(df_dev[impcol])\n",
    "    if min_prob == 0.0 and max_prob == 0.0:\n",
    "        start = np.min(pval[(df_dev.Class==1) & (pval>0)])\n",
    "        end = np.max(pval[df_dev.Class==1])        \n",
    "    else :\n",
    "        start = min_prob\n",
    "        end = max_prob\n",
    "    step = 1000.0\n",
    "    if log==True:\n",
    "        x = np.geomspace(start,end,num=step)\n",
    "    else :\n",
    "        x = np.linspace(start,end,num=step)\n",
    "    y = np.array([f1_score(df_dev['Class'],\n",
    "                 [1 if (x<eps) else 0 for x in pval]) \n",
    "        for eps in x])\n",
    "    epsilon=x[np.argmax(y)]\n",
    "    bestF1=np.max(y)\n",
    "    recall=recall_score(df_dev['Class'],[1 if (x<epsilon) else 0 for x in pval])\n",
    "    precision=precision_score(df_dev['Class'],[1 if (x<epsilon) else 0 for x in pval])\n",
    "    accuracy=accuracy_score(df_dev['Class'],[1 if (x<epsilon) else 0 for x in pval])\n",
    "    return(epsilon,bestF1,recall,precision,accuracy,x,y,pval)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Training\n",
    "Let's pick first 12 important features and find epsilon & other scores against best F1-score.Epsilon is a value to help determine whether a transaction is an outlier i.e. anomaly or potential fraud. Any transaction in dev/train dataset with p-value less than epsilon means the transaction is significantly different from non fraud transactions found in training data.\n",
    "We will use log scale to search epsilon between min and max p-value and best F1-score.\n",
    "\n",
    "Note:\n",
    "1) We've chosen 12 as arbitrary to prove we can find different scores quickly. Once proven we would look at what different numbers of features would give us\n",
    "2) Since the value of epsilon can be sensitive to scores, we would use log scale to find the range that give the best output."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epsilson: 7.343955470097046e-85 \n",
      "F1-score: 0.7682403433476395 \n",
      "Recall: 0.7276422764227642 \n",
      "Precision: 0.8136363636363636 \n",
      "Accuracy: 0.9981088795111104\n"
     ]
    }
   ],
   "source": [
    "# Search within min and max with log scale\n",
    "epsilon,bestF1,recall,precision,accuracy,x,y,dev_pval = find_bestF1(12,log=True)\n",
    "\n",
    "# Scores\n",
    "print('Epsilson: {} \\nF1-score: {} \\nRecall: {} \\nPrecision: {} \\nAccuracy: {}'.format(epsilon,bestF1,recall,precision,accuracy))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the p value and f1-score vs epsilon\n",
    "\n",
    "def plot_score(dev_pval,x,y):\n",
    "    fig = plt.figure(figsize=(20,8))\n",
    "    ax = fig.add_subplot(1,2,1)\n",
    "    sns.distplot(dev_pval[df_dev.Class==0], bins=50,ax=ax)\n",
    "    sns.distplot(dev_pval[df_dev.Class==1], bins=50,ax=ax)\n",
    "    ax.legend(['No Fraud', 'Fraud'])\n",
    "    ax.set_xlabel('P-value')\n",
    "    ax.set_ylabel('Frequency')\n",
    "    ax.set_title('Histogram of P- value')\n",
    "\n",
    "    ax = fig.add_subplot(1,2,2)\n",
    "    ax.plot(x,y,label='f1-score',color='red')\n",
    "    ax.axvline(x[np.argmax(y)],ls='--',color='blue')\n",
    "    ax.axhline(np.max(y),ls='--',color='blue')\n",
    "    #ax.axvline(x[np.argmax(y)-10],ls='--',color='lightblue')\n",
    "    #ax.axvline(x[np.argmax(y)+30],ls='--',color='lightblue')\n",
    "    ax.legend()\n",
    "    ax.set_xlabel('Epsilon - on log scale')\n",
    "    ax.set_xscale('log')\n",
    "    ax.set_ylabel('F1-score')\n",
    "    ax.set_title('F1-score vs epsilon')\n",
    "\n",
    "    plt.grid()\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "    \n",
    "plot_score(dev_pval,x,y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The p-value of Fraud data is concentrated near 0 which is a good sign. We have found the epsilon value against best F1-score.\n",
    "\n",
    "Lets narrow the search window and check if it improves the F1-score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABZgAAAI4CAYAAADnFoykAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nOzde5gcdZXw8e+ZBCEBkUtiVAIkooCAGDBcVyUIXl8uKmpAwI2sAgqysOAqiC7rBXcVxSssuEgEQdBwEVhYQXBA16hclTsicgkgQiRAIAFCzvtH9wyTSWYYpru6uqu/n+fpZ7qrqqtOVWo6VWdOn19kJpIkSZIkSZIkvVg9ZQcgSZIkSZIkSepMJpglSZIkSZIkSaNiglmSJEmSJEmSNCommCVJkiRJkiRJo2KCWZIkSZIkSZI0KiaYJUmSJEmSJEmjYoJZUqVExM0RMaPsOMoUEe+NiPsiYmFEbFF2PH0iIiPiNWXHIUmSJJWhfn3+6vrz2RHxpbJjkqRmMMEsqWNExN0RsfOgabMi4td9rzNz08zsfYH1TKknO8cWFGrZjgMOzszVMvP6wTPr+/5k/QL3/oj4RkSMKSFOSZIklaB+Xb2ofj3Y93hVfd7JEXF7RCyNiFklh1op9evzu8qOQ5KazQSzJDVZGySu1wdufoFl3pCZqwE7AR8CPlZ4VJIkSWonu9YTnn2PB+rT/wB8AriuxNiAtriuliSNgAlmSZUysMo5IraOiGsi4vGIeCgivlFf7Kr6zwX1ao3tIqInIo6OiHsi4m8RcVpEvGzAej9cnzc/Ij43aDvHRMSciPhRRDwOzKpve25ELIiIByPiuxHxkgHry4j4RET8KSKeiIgvRsQG9fc8HhE/Gbj8oH1cYawRsXJELATGAH+IiD+/0PHKzNuAXwGbjeDYbhsRfx1Y7Vxvx/HHAcd7yH0etK7eiPjogNfLVKJHxMYRcVlE/L1eQfPBF4pPkiRJjcvM72Xm5cDiF1o2IlapXwPPr18DXh0Rk+rz1oqIUyPigYh4NCLOH/C+j0XEnfVrvQv6qqfr8zIiDoqIPwF/qk8b0bVhROwZEdcMmnZYRFxQf/7uiLilfv19f0QcMcy+7RcRt9Zj/3lErD8oxkMi4q6IeCQivhYRPfV5r4mIKyPisfq8swe9b4Ut40ZwTA6s3zs8GhHfi4gYKnZJajUTzJKq7FvAtzJzdWAD4Cf16W+p/1yjXq0xF5hVf+wIvBpYDfguQERsApwA7A28EngZsM6gbe0OzAHWAM4AngMOAyYA21GrFP7EoPe8E3gjsC3wr8DJ9W2sSy3hu9cQ+7XCWDPz6XpVMtQqlDcY+tDU1PftzcByrTQGy8zfAk8Cbx0w+UPAmfXnI9nnFxQRqwKX1df7cmrH4YSI2PTFrkuSJEmF+kdq18brAmsDBwKL6vNOB8YDm1K7pjseICLeCnwF+CC1a+t7gLMGrfc9wDbAJi/y2vACYKOIeO2AaQOvV08BDsjMl1K73r5iRTsVEe8BjgLeB0ykVpDx40GLvReYDmxJ7V5gv/r0LwKXAmsCk4HvrGgbg7Y3kmOyC7AV8Ib6cu94ofVKUqu0XYI5In5Qr8i7aQTLHh8RN9Qfd0TEglbEKKlU59erIxbUf+dPGGbZZ4HXRMSEzFxYT5AOZW/gG5l5V2YuBI4E9oza1/LeD1yYmb/OzGeAzwM56P1zM/P8zFyamYsy89rM/G1mLsnMu4GTgB0Gvec/M/PxzLwZuAm4tL79x4BLgKEG6Bsu1pG6LiIeBS4E/hs4dYTv+zH1xHdEvBR4d30aI9znkdgFuDszT62v6zrgHGr/DpIkSWqOgdfV57/w4iv0LLXE8msy87n69eDjEfFK4F3AgZn5aGY+m5lX1t+zN/CDzLwuM5+mdi27XURMGbDer2Tm3zNzES/i2jAznwJ+xvPXq68FNqaWeO6Ld5OIWL0e11BtQA6ox3BrZi4BjgWmDaxipnYt//fMvBf4Js8XhzxLrWXdqzJzcWb+mhc2kmPyH5m5oL69XwLTRrBeSWqJtkswA7OpVfW9oMw8LDOnZeY0an8VPLfIwCS1hfdk5hp9D4avkP0nYEPgtvrX9XYZZtlXUasU6HMPMBaYVJ93X9+M+oXr/EHvv2/gi4jYMCIuqreUeJzaRemEQe95aMDzRSt4vRorNlysI7VlZq6ZmRtk5tGZuTQi9o7nB3m5ZIj3nQm8LyJWplbRcV1m3gMj3ueRWB/YZtAfEvYGXjGKdUmSJGnFBl5Xv2ckb4hlBwVcj1qV8s+Bs+qtML4aEStRq2j+e2Y+uoLVLHMtWy+YmM+y3xAceG39Yq8Nz+T5ZO+HgPPr1+8Ae1ArkLin3sZiuyHWsT7wrQHb+zsQw8R4T32/oPbNxAB+HxE3R8R+vLCRHJO/Dnj+FEPfK0hSy7Vdgjkzr6L24d0van1J/zciro2IX0XExit4614s/5UVSV0sM/+UmXtR+yrdfwJz6l+xG1x9DPAAtQvJPusBS6glfR+k9vU2ACJiHLVKjWU2N+j1icBtwGvrLTqOonah2QzDxTpqmXnGgEFe3jXEMrdQu/h9F8t+3RBe3D4/Se0rk30G3iDcB1w58A8J9Zg+Pro9kyRJUjMMGhTw3npl8r9n5ibA9tSqjT9M7XpurYhYYwWrWeZatn59vjZw/8BNDXj+Yq8NLwUmRMQ0anmC/uvVzLw6M3endn9wPs+30BvsPmqtNAZuc1xm/mbAMusOeL5efb/IzL9m5scy81XUKqFPGKrv8gAjOSaS1LbaLsE8hJOBT2bmG4EjGPSV+PrXVKYyRP8kSd0pIvaJiImZuRToa6HzHPAwsJRa/+I+PwYOi4ipEbEaterbs+tfiZsD7BoR20dt0Lp/54WTxS8FHgcW1v8o1szk6HCxtsKZwCHUeln/dMD0F7PPN1CrhB5fv+D+pwHzLgI2jIh9I2Kl+mOriHhdc3dDkiRJg0XESyJiFWrXuytFbSC/FeYOImLHiHh91AaBfpxae4jnMvNBai3fToiINevXc33joJwJfCQiptW/FXcs8Lt6i7UVeVHXhgOu378GrEWtf3Pffu0dES/LzGfr8T43xDb/Cziyr89z1AbU/sCgZT5V37d1gX8Gzq4v+4GI6CtOeZRasnyo7fR5scdEktpK2yeY68mT7YGfRsQN1Hp6vnLQYnsCczLzhT60JXWXdwI3R8RCagP+7Vnvg/YU8GXg/+pfe9sW+AG1r/hdBfyF2qjZnwSo90j+JLWBNh4EngD+Bjw9zLaPoFbh+wTwfeoXnE0yZKwt8mNgBnBFZj4yYPqL2efjgWeoVV3/kNrAiABk5hPA26l9tj9A7euA/wms3LQ9kCRJ0lAupdaubXtqxV6LeH6Q7MFeQS2Z+zhwK3Al8KP6vH2pJZxvo3btfChAZl4OfI5aH+UHqQ3GvedQwYzy2vBMYGfgp4OKMPYF7q63czsQ2GeIbZ5X38ZZ9WVvovYNvoF+BlxLrXDif6gNIAi1gfh+V78HuQD458z8yzCxvuhjIkntJjJX9E3xctUb2V+UmZtFxOrA7Zk5OKk8cPnrgYMGfV1FkgpR/8PXAmqtIIa9WJQkSZJULRGR1O4F7iw7FklqB21fwZyZjwN/6fs6StS8oW9+RGwErAnMLSlESV0gInatt3NYFTgOuBG4u9yoJEmSJEmSytV2CeaI+DG1ZPFGETEvIv6J2gix/xQRfwBuBnYf8Ja9gLOyHUuxJVXJ7tS+kvcA8Fpq7Tb83JEkSZIkSV2tLVtkSJIkSZIkSZLaX9tVMEuSJEmSJEmSOsPYsgMYaMKECTllypSyw5AkSVKHufbaax/JzIllx9Eq3XbdfPvttZ8bbdS8dT755JOsuuqqzVuhXjT/Dcrl8S+Xx79cHv9yefzL9eSTT3Lbbbc19dq5rRLMU6ZM4Zprrik7DEmSJHWYiLin7Bhaqduum2fMqP3s7W3eOnt7e5nRt2KVwn+Dcnn8y+XxL5fHv1we/3L19vay4447NvXa2RYZkiRJkiRJkqRRaasKZkmSJEka7PTTy45AkiRJQzHBLEmSJKmtrbtu2RFIkiRpKCaYJUmSWuTZZ59l3rx5LF68uOxQOtYqq6zC5MmTWWmllcoORS109tm1nzNnlhuHJEmSlmeCWZIkqUXmzZvHS1/6UqZMmUJElB1Ox8lM5s+fz7x585g6dWrZ4aiFTjyx9tMEsyRJUvtxkD9JkqQWWbx4MWuvvbbJ5VGKCNZee20rwCVJkqQ2YoJZkiSphUwuN8bjJ0mSJLUXE8ySJEmSJEmSpFGxB7MkSVJJzvzdvU1d34e2We8Fl4kI/uVf/oWvf/3rABx33HEsXLiQY445ZkTbmD17Np/61KdYZ511ANh888057bTTRh3zUI455hhWW201jjjiiKavW5IkSVLzmGCWJEnqIiuvvDLnnnsuRx55JBMmTBjVOmbOnMl3v/vdIecvWbKEsWO9zFTzzJlTdgSSJEkaii0yJEmSusjYsWPZf//9Of7445ebd88997DTTjux+eabs9NOO3HvvSOvsJ4xYwZHHXUUO+ywA9/61re48MIL2Wabbdhiiy3Yeeedeeihh4BaZfJxxx3X/77NNtuMu+++G4Avf/nLbLTRRuy8887cfvvtje2oKmXChNpDkiRJ7ccEsyRJUpc56KCDOOOMM3jssceWmX7wwQfz4Q9/mD/+8Y/svffeHHLIISt8/9lnn820adOYNm0ap556av/0BQsWcOWVV3L44Yfzpje9id/+9rdcf/317Lnnnnz1q18dNqZrr72Ws846i+uvv55zzz2Xq6++uvEdVWXMnl17SJIkqf343UVJkqQus/rqq/PhD3+Yb3/724wbN65/+ty5czn33HMB2HffffnXf/3XFb5/qBYZM2fO7H8+b948Zs6cyYMPPsgzzzzD1KlTh43pV7/6Fe9973sZP348ALvtttuL3i9VV19yedasMqOQJEnSiljBLEmS1IUOPfRQTjnlFJ588skhl4mIF7XOVVddtf/5Jz/5SQ4++GBuvPFGTjrpJBYvXgzUWnQsXbq0f7m+6aPZniRJkqTymWCWJEnqQmuttRYf/OAHOeWUU/qnbb/99px11lkAnHHGGbzpTW8a9fofe+wx1llnHQB++MMf9k+fMmUK1113HQDXXXcdf/nLXwB4y1vewnnnnceiRYt44oknuPDCC0e9bUmSJEmtY4sMSZKkknxom/VK3f7hhx++TKuLb3/72+y333587WtfY+LEicv0V36xjjnmGD7wgQ+wzjrrsO222/YnkvfYYw9OO+00pk2bxlZbbcWGG24IwJZbbsnMmTOZNm0a66+/Pm9+85sb2zlJkiRJLWGCWZIkqYssXLiw//mkSZN46qmn+l9PmTKFK664Ytj3z5o1i1kraITb29u7zOvdd9+d3Xfffbnlxo0bx6WXXrrCdX/2s5/ls5/97LDblyRJktReTDBLkiRJamsXX1x2BJIkSRqKCWZJkiRJbW38+LIjkCRJ0lAc5E+SJElSWzvhhNpDkiRJ7ccKZkmSJKnD3H47zJix7LQPfhA+8Ql46il497uXf8+sWbXHI4/A+9+//PyPfxxmzoT77oN9911+/uGHw6671rZ9wAHLzz/6aNh5Z7jhBjj00OXnH3ssbL89/OY3cNRRy8//5jdh2jT4xS/gS19adt4NVz/Dhms9wicW/ogLb9mAr/9qq+Xef/rM/2HdNZ7g7D9szIm/nbbc/Dn7/IwJqy5i9jWbMfvazVi06JWMG3dv//yLPzKH8S9Zwglzp/GTP2683Pt7DzgLgOOu3IqLbttgmXnjxi7hkn+aA8AXf7Edl/95/WXmrz1+Eefs+zMAjrzkLcy991XLzJ/8sif40Z7/A8ChF7yVGx58+TLzN5zwKCfv8XMA9j/nHdzxyJrLzJ/2yr/xzd1q/dP3Oev/Me+xly4zf7v1HuAr77oKenrY44pPMP+pZUvCd9oJPve52vN3vQsWLVp233fZBY44ovZ88HkHoz/3FiyYxhprtPe5B3DSSbDRRnDhhfD1ry8///TTYd114eyz4cQTl58/Zw5MmACzZ9ceg118ca1K/4QT4Cc/WX5+X4v7446Diy5adt64cXDJJbXnX/wiXH75svPXXhvOOaf2/MgjYe7c5+ctWDCNzTaDH/2o9vrQQ2vHcKANN4STT649339/uOOOZedPm1Y7fgD77APz5i07f7vt4CtfqT3fYw+YP3/Z+WWde33KPPcWLJjG7Nndee4BTJ5c7rn36lev339Oddu5B+V/7n3mM7V6124899rhc2/SpOWnN8oEsyRJkqT2lUtrd1FP3Qef/jSwC7D28st95VhgHvBBYPXl53/h34H5wD8C4xk3eP7njgYWAR+vr2OQT3+6/uTwegwDLRow/2hgp0Hz5w+Yfyyw3aD58+D6vvnHA4MS5HfdAb/vm78GsOGg+TfA//XNfxUwedD8udBbzzBs9E54xWZIkiQ1S2Rm2TH0mz59el5zzTVlhzF615wK0z9SdhSSJKlN3Xrrrbzuda8rO4yOt6LjGBHXZub0kkJquY6/bn4xFi9mxrjfwpQp9N788hdefoSuuuoq3vKWtzRtfR1h4sRa6dJxx5UdCQC9vb3MWFF5llrC418uj3+5PP7l8viXq7e3lx133LGp185WMEuSJJXlmlObu74R/KF7zJgxvP71r+9/ff755zNlypSmhnH33Xezyy67cNNNNzV1vepSfQUx0dPU0f6WrrJK940e2NPz/PGUJElqEhPMkiRJXWTcuHHcMLjZ3ABLlixh7FgvEdVGTIg2TwQsXVp2FJIkqWJ6yg5AkiRJ5Zo9ezYf+MAH2HXXXXn729/OwoUL2Wmnndhyyy15/etfz89+Vhuc7O6772azzZ7v3XrcccdxzDHHAHDttdfyhje8ge22247vfe97ZeyGqiqTXnak98Czyo6k81nBLEmSCmB5iiRJUhdZtGgR06bVBhCbOnUq5513HgBz587lj3/8I2uttRZLlizhvPPOY/XVV+eRRx5h2223Zbfddht2vR/5yEf4zne+ww477MCnPvWpwvdDXaS/RUaUG0cVWMEsSZIKYIJZkiSpiwzVIuNtb3sba621FgCZyVFHHcVVV11FT08P999/Pw899NCQ63zsscdYsGABO+ywAwD77rsvl1xySTE7oO6TyXEcDlduxRH+7aIxVjBLkqQCmGCWJEkSq666av/zM844g4cffphrr72WlVZaiSlTprB48WLGjh3L0gHVj4sXLwZqCemwulRFyeQidoFbN+CIsmPpdFYwS5KkAtiDWZIkSct47LHHePnLX85KK63EL3/5S+655x4AJk2axN/+9jfmz5/P008/zUUXXQTAGmuswcte9jJ+/etfA7UEtdQ0/S0yyg2jEnp6TDBLkqSms4JZkiSpLNM/UnYEK7T33nuz6667Mn36dKZNm8bGG28MwEorrcTnP/95ttlmG6ZOndo/HeDUU09lv/32Y/z48bzjHe8oK3RVkS0dmscWGZIkqQAmmCVJkrrIwoULl5s2a9YsZs2a1f96woQJzJ07d4XvP+SQQzjkkEOWm/7GN76RP/zhD/2vjznmmIZjlYABCVFLmBtmiwxJklQAW2RIkiRJIxQR74yI2yPizoj4zArmHx8RN9Qfd0TEggHz1ouISyPi1oi4JSKm1KfvFBHX1d/z64h4Tev2qANkMo5FjFtpSdmRdD4rmCVJUgFMMEuSJEkjEBFjgO8B7wI2AfaKiE0GLpOZh2XmtMycBnwHOHfA7NOAr2Xm64Ctgb/Vp58I7F1/z5nA0cXuSYfJ5BLezSUfv6DsSDqfFcySJKkAJpglSZJaKK0ebEjJx29r4M7MvCsznwHOAnYfZvm9gB8D1BPRYzPzMoDMXJiZT9WXS2D1+vOXAQ8UEXzH6h/kzxYZDXOQP0mSVAATzJIkSS2yyiqrMH/+/LKTpB0rM5k/fz6rrLJKWSGsA9w34PW8+rTlRMT6wFTgivqkDYEFEXFuRFwfEV+rV0QDfBS4OCLmAfsC/1FI9J0qky9yNF/8+VZlR9L5ImyRIUmSms5B/iRJklpk8uTJzJs3j4cffrjsUDrWKquswuTJk8va/IpKaIfK1u0JzMnM5+qvxwJvBrYA7gXOBmYBpwCHAe/OzN9FxKeAb1BLOi+78Yj9gf0BJk2aRG9v76h3pJOs9OijXM5OLLphbd7cxH1euHBh1xzDPts88wyPPfAAt7XJfnfjv0E78fiXy+NfLo9/uTz+5VrRoN+NMsEsSZLUIiuttBJTp04tOwyN3jxg3QGvJzN0O4s9gYMGvff6zLwLICLOB7aNiAuAN2Tm7+rLnQ3874pWmJknAycDTJ8+PWfMmDHK3egwDz0E3Mq48eNp5j739vY2dX0dYfx4xk2axCvaZL+78t+gjXj8y+XxL5fHv1we/3IVkdy3RYYkSZI0MlcDr42IqRHxEmpJ5OVGnouIjYA1gbmD3rtmREysv34rcAvwKPCyiNiwPv1twK0Fxd+ZbOnQPA7yJ0mSCmAFsyRJkjQCmbkkIg4Gfg6MAX6QmTdHxBeAazKzL9m8F3BWDmi2nZnPRcQRwOUREcC1wPfr6/wYcE5ELKWWcN6vlfvV9voPo4P8Naynx4S9JElqOhPMkiRJ0ghl5sXAxYOmfX7Q62OGeO9lwOYrmH4ecF7zoqyYTNZmPqy2uOxIOp8VzJIkqQC2yJAkSZLUvjI5h/dzzoG/KDuSztfTY4JZkiQ1nQlmSZIkSe2rr6VD2CKjYRG2yJAkSU1nglmSJElS+8rkSI7lyPO2KjuSzmcFsyRJKoA9mCVJkiS1r0zmsh38eVLZkXQ+B/mTJEkFsIJZkiRJUvvqb5FRbhiV4CB/kiSpACaYJUmSJLUvK26bxwpmSZJUABPMkiRJktpXf0LUEuaGWcEsSZIKYA9mSZIkSe0rk8nMg7XWLTuSzucgf5IkqQAmmCVJkiS1r0x+xL5w4A+BDcqOprNF2CJDkiQ1nS0yJEmSJLWv/kH+bJHRMCuYJUlSAUwwS5IkSWpfmRzK8Rx6xvSyI+l8DvInSZIKYIsMSZIkSe0rkxuYBvesVXYknc9B/iRJUgGsYJYkSZLUvvpbZJQbRiVYwSxJkgpgglmSJElS+zIh2jxWMEuSpAKYYJYkSZLUvvoTzJYwN8xB/iRJUgFMMEuSJElqX5lsyB1s+Monyo6k89kiQ5IkFcBB/iRJkiS1r0xO5gA44Gxgg7Kj6Wy2yJAkSQWwglmSJElS+wtbZDTMCmZJklQAE8ySJEmS2lcm+3MS+5/0xrIj6XxWMEuSpALYIkOSJElS+8rkDjaEB19adiSdzwpmSZJUACuYJUmSJLWvvoSoHTIaZwWzJEkqgAlmSZIkSe3Litvm6ekxwSxJkprOBLMkSZKk9tWfYLaEuWG2yJAkSQWwB7MkSZKk9pXJNG6AqesCE8qOprPZIkOSJBXABLMkSZKk9pXJNzkMPvZqYIOyo+lsVjBLkqQC2CJDkiRJUvvqH+TPFhkNs4JZkiQVwASzJEmSpPaVyT6czj7f2KLsSDqfFcySJKkAtsiQJEmS1L4ymcdkeGRc2ZF0PiuYJUlSAaxgliRJktS++ltklBtGJfT0mGCWJElNZ4JZkiRJUvuypUPz2CJDkiQVwASzJEmSpPbVnxC1hLlhtsiQJEkFsAezJEmSpPaVyXbMhdetB6xVdjSdzQpmSZJUABPMkiRJktpXJl/hKNhvS+DVZUfT2axgliRJBbBFhiRJkqT21T/Iny0yGmYFsyRJKoAJZkmSJEntK5M9mMMeX3xD2ZF0PiuYJUlSAWyRIUmSJKl9ZTKfteHxl5QdSefr6THBLEmSms4KZkmSJEnty5YOzWOLDEmSVAATzJIkSZLaV38P5nLDqARbZEiSpAKYYJYkSZLUvqy4bR4rmCVJUgHswSxJkiSpfWWyE1fAlusDa5YdTWezglmSJBXABLMkSZKk9pXJ5/gS/OPOwNSyo+lsVjBLkqQC2CJDkiRJUvvq78FsE+aGWcEsSZIKUGiCOSIOi4ibI+KmiPhxRKxS5PYkSZIkVUwm7+Ji3vWvm5UdSefr6THBLEmSmq6wBHNErAMcAkzPzM2AMcCeRW1PkiRJUgVlsohxLHp6TNmRdD5bZEiSpAIU3SJjLDAuIsYC44EHCt6eJEmSpCoxIdo8tsiQJEkFKCzBnJn3A8cB9wIPAo9l5qWDl4uI/SPimoi45uGHHy4qHEmSJEmdqL8Hc7lhVIIVzJIkqQBFtshYE9id2lDPrwJWjYh9Bi+XmSdn5vTMnD5x4sSiwpEkSZLUiUyINo8VzJIkqQBFtsjYGfhLZj6cmc8C5wLbF7g9SZIkSVWTyS5cxC7/sKDsSDqfFcySJKkAYwtc973AthExHlgE7ARcU+D2JEmSJFVNJkfwddjnA8D6ZUfT2axgliRJBSiyB/PvgDnAdcCN9W2dXNT2JEmSJFVQfw9mmzA3rKfHBLMkSWq6IltkkJn/lpkbZ+ZmmblvZj5d5PYkSZKkIkXEOyPi9oi4MyI+s4L5x0fEDfXHHRGxYMC89SLi0oi4NSJuiYgp9em/GvCeByLi/NbtUQfIZAa/ZMbHNy47ks5niwxJklSAIltkSJIkSZUREWOA7wFvA+YBV0fEBZl5S98ymXnYgOU/CWwxYBWnAV/OzMsiYjVgaf09bx7wnnOAnxW6I52mPyFqBXPDbJEhSZIKUGgFsyRJklQhWwN3ZuZdmfkMcBaw+zDL7wX8GCAiNgHGZuZlAJm5MDOfGrhwRLwUeCtgBfNAVtw2jxXMkiSpACaYJUmSpJFZB7hvwOt59WnLiYj1ganAFfVJGwILIuLciLg+Ir5Wr4ge6L3A5Zn5eJPj7mz9PZjLDaMSrGCWJEkFsEWGJEmSNDIrSnEOVQ66JzAnM5+rvx4LvJlay4x7gbOBWcApA96zF/DfQ248Yn9gf4BJkybR29v7IkLvXBNuvBF4E0888QS9vdc2bb0LFy7smmPYZ8q997J+Jle2yX53479BO/H4l8vjXy6Pf7k8/uVauHBh09dpglmSJEkamXnAugNeTwYeGGLZPYGDBr33+sy8C6A+kN+21BPMEbE2tePaPFgAACAASURBVBYc7x1q45l5MnAywPTp03PGjBmj2omO8+ijfJCfwB5H0sx97u3tber6OsIvf1kbNHGHHWrVzCXryn+DNuLxL5fHv1we/3J5/MtVRHLfFhmSJEnSyFwNvDYipkbES6glkS8YvFBEbASsCcwd9N41I2Ji/fVbgVsGzP8AcFFmLi4k8k6WySc4kU/M/HvZkXS+nvrtn32YJUlSE5lgliRJkkYgM5cABwM/B24FfpKZN0fEFyJitwGL7gWclfl8Fq/eKuMI4PKIuJFau43vD3jPntQHBNQgmTzFOJ5a7K1Lw0wwS5KkAtgiQ5IkSRqhzLwYuHjQtM8Pen3MEO+9DNh8iHkzmhNhBWXybi6Gg6fSe3XZwXS4vrYYS5fCmMFjTEqSJI2OZQCSJEmS2ld/tW35PYM7nhXMkiSpACaYJUmSJLUvk6HNM7CCWZIkqUlMMEuSJElqX30JZguYG2cFsyRJKoAJZkmSJEnty2Ro81jBLEmSCuAgf5IkSZLaVyazmA3v3QhYtexoOltfBbMJZkmS1ERWMEuSJElqX5nM4ofMet/jZUfS+WyRIUmSCmCCWZIkSVL7yuQR1uaRR8eUHUnns0WGJEkqgAlmSZIkSe0rk/czh/f/86vKjqTzWcEsSZIKYIJZkiRJUvvqT4ZGqWFUghXMkiSpACaYJUmSJKkbOMifJEkqgAlmSZIkSe2rr4LZAubG9VUw2yJDkiQ1kQlmSZIkSe3LZGjzWMEsSZIKMLbsACRJkiRpSJl8nBNhn02AcWVH09kc5E+SJBXABLMkSZKk9pXJTH4Cu3617Eg6n4P8SZJUnrvvhquvXnbattvCuuuWEk4zmWCWJEmS1L4yuY/J8OBY1l2/7GA6nBXMkiSV54AD4NJLl5121lkwc2Y58TSRCWZJkiRJ7SuTfTkdDlub3rllB9PhrGCWJKk8Tz0FW28NP/jB89MmTy4vniYywSxJkiSpffVX20apYVSCg/xJklSeTFhtNdh007IjaToTzJIkSZLal+0cmqevgvnBB2GllVq//Ve8AsaMaf12JUlqB5nP/19cMSaYJUmSJLWvvgRzNe/HWmvcuNrP7bcvZ/sHHggnnljOtiVJKpsJZkmSJEkqgRXMzbPbbnDaabB4ceu3/YUvwAMPtH67kiS1CxPMkiRJklSCTA7n63DgacDKZUfT2caNg333LWfbJ55o72dJUnczwSxJkiRJJchkVy6CdzxTdiRqRE+P1eiSpO5W4QRzT9kBSJIkSdKQMrmdDbn9z9bGdLQIK5glSd3NBLMkSZIklSCTAziJAz710rIjUSOsYJYkdTsTzJIkSZJUApOS1WAFsySp25lgliRJkqQS9CeYq3lD1jV6ekwwS5K6mwlmSZIkSSqBFczVYIsMSVK3M8EsSZIkSSXoS0pW836se9giQ5LU7SqcYHYoZkmSJEntK5Oj+RIcfh6wUtnRaLSsYJYkdTsTzJIkSZJUgkx25nLY8bmyI1EjrGCWJHW7CieYbZEhSZIkqX1lcgNv4IYbx5QdiRphBbMkqduZYJYkSZKkEmRyKN/k0KPGlR2JGmEFsySp25lgliRJkqQSWPVaDT09JpglSd3NBLMkSZIklaA/wVzNG7KuYYsMSVK3y6z9f1hB1dwrSZIkSdVgUrIabJEhSep2VjBLkiRJUgn6EszVvB/rHlYwS5K6XYUTzGPLDkCSJEmShpTJsRwF//5LYEzZ0Wi0rGCWJHU7E8ySJEmSVIJMtmcubGf1a0dzkD9JUrercILZFhmSJEmS2lcmv2E7fvNbb106WoQtMiRJ3c0EsyRJkiSVIJOjOJaj/s0vX3Y0K5glSd3OBLMkSZIklcCq12pwkD9JUrczwSxJkiRJJehPSlbzhqxrOMifJKnbmWCWJEmSpBJY9VoNVjBLkrqdCWZJkiRJKkFfUrKa92PdwwpmSVK3q3CC2ZEyJEmSJLWvTL7JoXD89WVHokY4yJ8kqduZYJYkSZKkEmQyjT/AFtW8IesatsiQJHW7CieYbZEhSZIkqX1l8gt24he/KDsQNcQWGZKkbmeCWZIkSZJKkMmXOJovfansQNQQK5glSd3OBLMkSZIklcCkZDVYwSxJ6nYmmCVJkiSpBJlANW/GuooVzJKkbmeCWZIkSZJKYFKyGqxgliR1OxPMkiRJkiLinRFxe0TcGRGfWcH84yPihvrjjohYMGDeehFxaUTcGhG3RMSU+vSIiC/Xl781Ig5p3R51gEwLmKugp8cEsySpu1U4wTy27AAkSZKkThARY4DvAW8D5gFXR8QFmXlL3zKZediA5T8JbDFgFacBX87MyyJiNaAv2zYLWBfYODOXRsTLi92TDpPJSWMPhpNuLDsSNcIWGZKkbmeCWZIkSep6WwN3ZuZdABFxFrA7cMsQy+8F/Ft92U2AsZl5GUBmLhyw3MeBD2Xm0vq8vxUTfofKZKOeP8FGZQeihtgiQ5LU7UwwS5IkSV1vHeC+Aa/nAdusaMGIWB+YClxRn7QhsCAizq1P/wXwmcx8DtgAmBkR7wUeBg7JzD+tYJ37A/sDTJo0id7e3mbsU9t79T33cN1z7+aWY29k++3nN229Cxcu7Jpj2A42/OtfmfD00/xmwDH336BcHv9yefzL5fEvV7ce/+2feYaHH3iAP5W87wsXLnzhhV4kE8ySJEnSyKyo5GSo7/zvCcypJ5Chdt39ZmotM+4FzqbWGuMUYGVgcWZOj4j3AT+oL7vshjJPBk4GmD59es6YMWPUO9JR/ud/+PDSXeHS13PUUc1bbW9vL11zDNvB2WfD2LHLHPOu/De49lq49dbmrGvCBHjnO0f99q48/m3E418uj3+5uvb4jx3LOpMns07J+15Ect8EsyRJkjQy86j1Su4zGXhgiGX3BA4a9N7rB7TXOB/YllqCeR5wTn2584BTmxhz57NvbzU4yF/NbrvBA0N9bIzCPffAeus1b32SpOLYIkOSJEnqelcDr42IqcD91JLIHxq8UERsBKwJzB303jUjYmJmPgy8FbimPu/8+usfADsAdxS2B50okxUXj6ujOMhfzaJFsM8+8G//1th6LrwQ/uVfauuTJHUGE8ySJElSd8vMJRFxMPBzYAzwg8y8OSK+AFyTmRfUF90LOCvz+WxaZj4XEUcAl0dEANcC36/P/g/gjIg4DFgIfLRFu9QZTEpWg4P81WTCGmvAa17T2Hpe+craT4+pJHUOE8ySJEmSMvNi4OJB0z4/6PUxQ7z3MmDzFUxfAPy/5kVZMZkWMFeBFcw1zUou9PTUfppglqTOYYJZkiRJkkqQyenjDoDTbyk7EjXCCuaaZieYTdpLUucwwSxJkiRJ5Vh3zAPLDq+ozmMFc02zkgt96zBpL0mdY+nSyiaYe8oOQJIkSZKGlMnZz76Ps88uOxA1xArmGiuYJal7VbiC2QSzJEmSpPaVyYnP7MeJJ5YdiBrS02OCGaxglqRuZoJZkiRJkkpghWY12CKjxkH+JKl7mWCWJEmSpBJkAtW8GesqtsiosUWGJHUvE8ySJEmSVAITaNVgBXONLTIkqXuZYJYkSZKkEmRawFwFVjDXWMEsSd2rwgnmsWUHIEmSJElDymTOyz4Kc24tOxI1wgrmGiuYJal7mWCWJEmSpBJkMmHMozCh7EDUkIjajXWFb65HxEH+JKl7Vfj/QFtkSJIkSWpfmcxeNJPZs8sORA2xpUONLTIkqXtVOMFsBbMkSZKk9pXJ7MUzYTbMmlV2MBo1E6I1tsiQqmXBAvja12DRorIjacwee8A//EPZUVSfCWZJkiRJKkG3JySrYmBCdMyYcmMpkxXMUrVccQUceyyMH9+5n20LF8Kf/ww/+1nZkVSfCWZJkiRJKkEmUM2bsa5iQrTGCmapWp57rvbz97+HTTctN5bReuMbn98PFavCCWZ7MEuSJElqX92ekKwKE6I1DvInVUvf/1GdnDTs6fH/2lYxwSxJkiRJJci0gLkKTIg+zxYZUnVUIcEc4WdzK3XyuTIMW2RIkiRJal+ZXDzhH+HiO8qORI0wIdrcRJQV4VJ7qEKCuafHz5JWqMK5MgwTzJIkSZKK8/3vw4UXjv7911/P+DE9ML55IakEJkSbm1wwYS+1hyokDW2R0RpVOFeGYYJZkiRJUnH+67/gzjthgw1G9/6JEzlhraPhBPjEJ5obmlrIhKgVzFIVVSFpaIuM1qjCuTIME8ySJEmSipMJO+wAF1ww6lX8ZAbwExPMHc2EqBXMUhVVIWloi4zWqMK5MgwH+ZMkSZJUnAqPmK4XwYSoFcxSFVUhaWiLjNaowrkyDBPMkiRJkopjgllgQhSKqWDu5uMptYMqJA1tkdEaVThXhmGCWZIkSVJxTDALTIiCLTKkKqpC0tAK5taowrkyDBPMkiRJkopjgllgQhRskSFVURWShlYwt0YVzpVhOMifJEmSpOI0IcHc29ucUFQiE6JWMEtVVIWkoYP8tUYVzpVhWMEsSZIkqThWMAtMiIIVzFIVVSFpaIuM1qjCuTIME8ySJEmSitOEBPNxx9Ue6mAmRB3kT6qiKiQNbZHRGlU4V4ZRaII5ItaIiDkRcVtE3BoR2xW5PUmSJEltpgkJ5osuqj3UwUyI2iJDqqIqJA2tYG6NKpwrwyi6B/O3gP/NzPdHxEuA8QVvT5IkSVI78aZV8PwNteeDLTKkKqlC0tAezK1RhXNlGIUlmCNideAtwCyAzHwGeKao7UmSJElqQ/ZgFljBDFYwS1VUhaShLTJaowrnyjCKbJHxauBh4NSIuD4i/jsiVh28UETsHxHXRMQ1Dz/8cIHhSJIkSWo5E8wCE6LgIH9SFVUhaWiLjNaowrkyjCITzGOBLYETM3ML4EngM4MXysyTM3N6Zk6fOHFigeFIkiRJarkmJJjHjas91MFMiDrIn1RFVUgaWsHcGlU4V4ZRZA/mecC8zPxd/fUcVpBgliRJklRhTUgwX3JJk2JReaxgtkVGMzz2GCxe3Ph6Xv7yyiZ51GJVSBrag7k1qnCuDKOwBHNm/jUi7ouIjTLzdmAn4JaitidJkiSpDdkiQ2AFM9gio1G33AKbbw7PPdf4uj77WfjSlxpfj1SFpKEtMlqjCufKMIqsYAb4JHBGRLwEuAv4SMHbkyRJktROmpBg/uIXaz8/97kmxKNy2NLBCuZGPfhgLbl82GHw2teOfj2f/SzMm9e8uNTdqpA0tEVGa1ThXBlGoQnmzLwBmF7kNiRJkiS1sSYkmC+/vPbTBHMH68aE6GBWMDem7/i95z3wlreMfj1f/Wp3HTcVqwpJQyuYW6MK58owihzkT5IkSVK3s0WGoDsTooM5yF9jmnX8TKapmaqQNLSCuTWqcK4MwwSzJEmSpOKYYBZYwQy2yGhUs46fyTQ1UxWShg7y1xpVOFeGYYJZkiRJUnFMMAusYAZbZDTKCma1oyokDf2daI0qnCvDKHqQP0mSJEndrAkJ5rXXblIsKk83VtwOZgVzY6xgVjuqQtLQ34nWqMK5MgwTzJIkSZKK04QE8znnNCkWlacbK24Hs4K5Mc2sYO6m46ZiVSFp6O9Ea1ThXBmGLTIkSZIkFccWGYLuHJRuMAf5a4wtMtSOqpA09HeiNapwrgzDBLMkSZKk4jQhwXzkkbWHOlg3tnQYrIgK5m46nrbIUDuqQtLQ34nWqMK5MgxbZEiSJEkqThMSzHPnNikWlacbWzoMZgVzY6xgVjuqQtLQ34nWqMK5MgwrmCVJkiQVxxYZAiuYwUH+GmUFs9pRFZKG/k60RhXOlWGYYJYkSZJUHBPMAiuYwUH+GuUgf2pHVUga+jvRGlU4V4ZhglmSJElScUwwC7qzpcNgVjA3xhYZakdVSBr6O9EaVThXhmEPZkmSJEnFaUKCefLkJsWi8nRjQnQwK5gbY4sMtaMqJA39nWiNKpwrwzDBLEmSJKk4TUgw/+hHTYpF5enGhOhgDvLXGFtkqB1VIWloBXNrVOFcGYYtMiRJkiQVxxYZAiuYwRYZjWpmBXM3HTcVqwpJQ//o0hpVOFeGYYJZkiRJGqGIeGdE3B4Rd0bEZ1Yw//iIuKH+uCMiFgyYt15EXBoRt0bELRExpT59dkT8ZcD7prVuj1qgCQnmQw+tPdTBrGC2RUajrGBWO6pC0tAWGa1RhXNlGLbIkCRJkkYgIsYA3wPeBswDro6ICzLzlr5lMvOwAct/EthiwCpOA76cmZdFxGrAwLu5T2XmnEJ3oCxNSDDfcEOTYlF5urHidjArmBvjIH9qR1VIGvo70RpVOFeGYQWzJEmSNDJbA3dm5l2Z+QxwFrD7MMvvBfwYICI2AcZm5mUAmbkwM58qOuC2YIsMQXdW3A5mBXNjHORP7agKSUN/J1qjCufKMKxgliRJkkZmHeC+Aa/nAdusaMGIWB+YClxRn7QhsCAizq1P/wXwmcx8rj7/yxHxeeDy+vSnV7DO/YH9ASZNmkRvb2/DO9QKb3r2Wf56//3c2UC8CxbUuob09javlHnhwoUdcwyrYPUbb2RL4A/XX8+jY2u3od32bzBu3jy2AW697TYeanS/M5kB3H3XXdw9ynV12vGfeNNNbApcfc01PDl//qjXM+2JJ8gnn+QPJe97px3/qmnW8V//rruYCvReeSWMGdPw+sqwwf3386olS/hVC8/Hbjz/x913H9sAt9x6K39rg8+fZjPBLEmSJI3MikpOhvpO6Z7AnAEJ5LHAm6m1zLgXOBuYBZwCHAn8FXgJcDLwaeALy20o8+T6fKZPn54zZswY5W602JgxTF53XSY3EO8aa9R+NnOfe3t7m7o+vYCVVwbgDauvDhMnAvD7v/yFradPh1e8oszIWueOOwB43Sab8LpmnHsRTFl/faaMcl0d9zvw0EMAbLX11rDppqNfz5prAs39PBmNjjv+FdO043/VVQDM2HHH51vXdJqLL4aIlp6PXXn+3347AJtsuimbtMHnT7OZYJYkSZJGZh6w7oDXk4EHhlh2T+CgQe+9PjPvAoiI84FtgVMy88H6Mk9HxKnAEU2NumxNaJGx4YZNikXlGT++9vPAA/snbQ21xPP8+bDqqqWE1VLN/np0t32tvZktMp577oWXk0aiCm0Puu2zpCxVOFeGYYJZkiRJGpmrgddGxFTgfmpJ5A8NXigiNgLWBOYOeu+aETExMx8G3gpcU1/+lZn5YEQE8B7gpmJ3o8WakGA++eQmxaLybL45/Pzn8Nhj/ZMePPVUXnnJJfDUUyaYR6PbBuZq5iB/S5Y0Ho8E1UgadttnSVmqcK4MwwSzJEmSNAKZuSQiDgZ+DowBfpCZN0fEF4BrMvOC+qJ7AWdlPn+3lpnPRcQRwOX1RPK1wPfrs8+IiInUWnDcADxf4lkFDvInqJ0Db3/7MpOe+L//qyWYu6VyzgrmxjjIn9pRFRKz/k60hglmSZIkSQCZeTFw8aBpnx/0+pgh3nsZsPkKpr+1iSG2pwZvpvbfv/bTSuaK6etXWoUEzUgUUcHcTUmhZlYwd9NxU7Gq8EdUfydawwSzJEmSJI1SE26+62OjqWL608rdktiwRUZzNCPB3I3HTcWoSoLZ34niVTzB3KFDXEqSJEnqCFW4+VYx+iqYTTCPTrd9rd0WGWpHVfg/LqK2HyaZi2WCWZIkSZJGqQo33ypE9p0X3ZLUsIK5Mc1skdFNx03FWrq08/+P67Z2RWUxwSxJkiRJo2SCWUPpOy+6pZrUCubGWMGsdlSF/+NMMLdGxRPM9mCWJEmSVJwm3HxPm9akWNReui2p4SB/jXGQP7WjKiSYB/6xb8yYcmOpMhPMkiRJkjRKTbj5/uY3mxSL2oqD/DWo21o92CJD7agKCeZu+2NfWSqeYLZFhiRJkqTiVOHmW8XotqRGs/ez21o92CJD7agK/8d1W7uisphghojYrOhAJEmSJFVQE26+99mn9lC1ZLclNaxgbowVzGpHVUgw9/2xr1s+i8tS8QTzSFtk/FdEvASYDZyZmQuKC0mSJElSZTTh5nvevCbFovZigrkx3VaJawWz2lGVEsz+4aVYFU8wj6iCOTPfBOwNrAtcExFnRsTbCo1MkiRJUuerws23itFtSQ0rmBvjIH9qR1X4P67b/thXFhPMNZn5J+Bo4NPADsC3I+K2iHhfUcFJkiRJ6nBVuPlWIRzkr0HdVolriwy1oyr8H2eLjNYwwQwRsXlEHA/cCrwV2DUzX1d/fnyB8UmSJEmSqsgK5sZ0WyWuLTLUjqqUYO6Wz+KyVDzBPNIezN8Fvg8clZmL+iZm5gMRcXQhkUmSJEkFi4hxwHqZeXvZsVRSk26mttuuCbGo7TjIX4O6rRLXCma1oyokmLvts7gsJpgBeDewKDOfA4iIHmCVzHwqM08vLDpJkiSpIBGxK3Ac8BJgakRMA76QmbuVG1mFNOlm6itfaUIsaj/dltSwRUZjrGBWO6pCgtkK5taoeIJ5pD2YfwGMG/B6fH2aJEmS1KmOAbYGFgBk5g3AlBLjqZ6K30ypQd2W1LCCuTEO8qd2VIUEc7f9sa8sFb8mGmmCeZXMXNj3ov58fDEhSZIkSS2xJDMfKzuISmvSzdQee9QeqhYH+WtQt1Xi2iJD7agKCWYH+WsNE8wAPBkRW/a9iIg3AouGWV6SJElqdzdFxIeAMRHx2oj4DvCbsoOqlCbdTM2fX3uoYqxgbky3VeLaIkPtqEoJ5m75LC5LxRPMI+3BfCjw04h4oP76lcDMYkKSJEmSWuKTwGeBp4EzgZ8DXyo1oqqp+M2UGuMgfw3qtkpcK5jVjqqQYO62z+KyVPyaaEQJ5sy8OiI2BjYCArgtM58tNDJJkiSpIBExBvj3zPwUtSSzXsg3vgFz5ox8+Y9+FPbeu/a8ojdTalC3JTVskdEYK5jVjqqQYLaCuTVMMPfbitqgJ2OBLSKCzDytkKgkSZKkAmXmc/W2bxqplVeG1VYb2bK//W0tGf2hD9VeV/RmSg3qtqSGFcyNcZA/taMqJJi77Y99ZTHBDBFxOrABcAPwXH1yAiaYJUmS1Kmuj4gLgJ8CT/ZNzMxzywupjR10UO0xEttsU7uRatLN1E47NfR2tSkH+WtQt1Xi2iJD7agKCWYH+WsNE8wATAc2yfRTWJIkSZWxFjAfeOuAaQmYYG5UX+KrSTdTn/tcE2JS++m2CuY+DvI3OrbIUDuqUoK52z6LW80EMwA3Aa8AHiwwFkmSJKllMvMjZcdQWX0VghW/mVJjHOSvQd1WiWsFs9pRFRLM3fZZXJaKXxONNME8AbglIn5PbZRtADJzt0KikiRJkgoWEZOB7wD/QK1y+dfAP2fmvFIDq4ImVzC/6121n5dc0mBcai9950W3JPtskdEYK5jVjqqQYLZFRmuYYAbgmCKDkCRJkkpwKnAm8IH6633q095WWkRV0eQK5kWLmhCT2k+3JTWsYG6Mg/ypHVUpwdxNnydlqHiCuWckC2XmlcDdwEr151cD1xUYlyRJklS0iZl5amYuqT9mAxPLDqoSmlzBrGpykL8GdVslri0y1I6qkGC2RUZr9H3u9IwoFdtxRrRXEfExYA5wUn3SOsD5RQUlSZIktcAjEbFPRIypP/ahNuifGtVXIWiCWcPptqq5IiqYuykhZIsMtaMqJJi77bO4LBW/Jhpp2vwgar3pHgfIzD8BLy8qKEmSJKkF9gM+CPyV2mDW769PU6MiHORPL6zbquZskdEYK5jVjqqQYO62z+KyVPyaaKQ9mJ/OzGeifhAiYiwDvtEkSZIkdZrMvBdw0Ooi9PTAs8827WZql12aEJPaTjrIX2O6rRLXCma1oyokmLutH35ZKp5gHmkF85URcRQwLiLeBvwUuLC4sCRJkqRiRcQPI2KNAa/XjIgflBlTZTR5kL8jjqg9VDHdltSwgrkxDvKndlSlBHM3fZ6UwQQzAJ8BHgZuBA4ALgaOLiooSZIkqQU2z8wFfS8y81FgixLjqQ4H+dMIOMhfg7qtEtcWGWpHVUgw2yKjNSp+TTSiFhmZuRT4fv0hSZIkVUFPRKxZTywTEWsx8hZyGk6TK5hnzKj97O1taDVqN91WNecgf42xRYba0f9n787j5KjLxI9/nskN4ZJwiOGIcomCQcMlIPkporAiKiiHuuLuigjoerCK/lw8Vv2p64EgiuABogiIyKIGFdTx4j6CSDgWwhURhEAIIRcz8/390TXJZDKT6Zmu7qrq/rxfr3p1d3V11VPfmp6eeeaZ59sOCeZO+2+Sophghoi4jyF6LqeUnp97RJIkSVJrfBm4OiIuyR6/GfhsgfG0DyuYVY9Oq5prRgXz0qXwwANjevmkRx4Z82ubauONYaON1l5vBbPKqJ0SzL4vmqvNfyaqt0Jj1oD7k6n98P2c/MORJEmSWiOl9P2IuBF4JRDAm1JK8woOqz30V1a2+S9TaoyT/DVoyhS46irYbrsxvXyffKLI3wYbwGOPwaRJa663glll1A4J5k77Y19R2vxnonpbZCwctOq0iPgTcGr+IUmSJEnNFxEvAO5NKc2LiNnAgRHx8MC+zBqjiFxbZKhNddq/Zef9fvj61+Hqq8f88jvvvJOdd945n1jycuWV8KMfwbJlzUswd1prETVXOySYrWBujTb/majeFhkvHfCwi1pF8wZNiUiSJElqjZ8AsyJie+DbwM+AC4BDCo2qHVjBrDqsSmV0SlIj7/fDTjvVljF6pLubnfsbnJfF00/XEsxDJYDzTDD378/vTWpUO3wdWcHcGm3+M1G9LTK+POB+D3A/8Jbco5EkSZJapy+l1BMRbwK+llI6IyJuKTqotpDzJH9v8TeP9mQFswZbVyVlni0y+vfntVCj2uHrqNO+FxelzT8D6m2R8X+aHYgkSZLUYs9GxNHAPwOHZusmFBhP+8h5kr8TTsghJpVPp1XNtXlyIRfr+ppoRgWz1Kh2SjD7nmiuNv8MqLdFxgfX9XxK6Sv5hCNJkiS1zDuB44HP6UMlFAAAIABJREFUppTui4gZwA8Kjqk95FzBvHRp7Xa99RqMS6XiJH9aSysrmPv6YNy4xvYltUOCudP+2FeUNv8MqLdFxixgD+Dy7PGhwB+Ah5oRlCRJktRsKaV5wPugNudISulm4PPFRtUmcq5gPiTrit3d3VhYKplO+7fsNk8u5KKeCuZGddrXnZqrHRLMVjC3Rpt/BtSbYJ4GvDSl9DRARHwS+HFK6d+aFZgkSZLUQt8GXjriVqqPk/ypDk7yp7WsK/mbVyLPZJry1A4JZiuYW6PNPwO66txuG2DlgMcrge1yj0aSJEkqRnv+tF+UiDWTN236y5Qa1GmVpG2eXMjFSC0y8hg7k2nKUzskmDvte3FR2vwzoN4K5vOB6yPip9T+0PxG4PtNi0qSJElqrU8VHUBbsYJZ9ei0RJ/vh5GN1CLDCmaVTTslmH1PNFebfwbUlWBOKX02Iq4A9s9WvTOldEvzwpIkSZJaJ6V0GUBE7JxSurPoeCov50n+1J6c5E9rsYJZVdMOCWbfE63R5p8B9VYwA6wHLE4pfS8iNouIGSml+5oVmCRJklSAX1NrD6dG5DzJ37HHNh6SSqjT/i27zZMLuWhlBXOnfN2pudohwex7ojXa/DOgrgRzRHwCmAXsBHwPmAD8ANi3eaFJkiRJ+YuI04d7Cti4lbG0rZwrmE0wtycn+dNanORPVdNOCWbfE83V5p8B9VYwvxHYHbgZIKX0cERs0LSoJEmSpOZ5J/AhYMUQzx3d4ljaU84VzI8/XrudNq3BuFQunVY11+bJhVzYIkNV0w4JZt8TrdHmnwH1JphXppRSRCSAiFi/iTFJkiRJzXQD8NeU0tWDn4iIT7Y+nDaU8yR/RxxRu+3ubiwslUynJTXaPLmQCyf5U9W0Q4J5fJYafMMbVt9vsv37+la/FztFT0/tdty4YuNoknq/ci6OiG8BG0fEu4B/Ac5pXliSJElS0xwBLB/qiZTSjBbH0p4inORPI3KSP63FCmZVTTskmF/yEviv/4KnnmrZIf/20ENss/XWLTteaUybBjPa80fNuhLMKaUvRcSrgcXU+jCfmlK6sqmRSZIkSc0xNaX0RNFBtLWcK5jVpmyRocGc5E9V0w4J5gkT4OMfb+kh53d3s83s2S09ppprxHr0iBgXEVellK5MKf1HSulkk8uSJEmqsMv670TET0bzwoh4bUTcFRH3RMQpQzz/1YiYmy13R8SiAc9tExG/jog7ImJeRGw36LVnRMSS0Z9OCeU8yZ/ak5P8aS1O8qeqaYcEs5SDESuYU0q9EbE0IjZKKbWuXl6SJElqjoG/CT6/7hdFjAPOBF4NLABuiIjLU0rz+rdJKX1gwPbvpTZRdr/vA59NKV0ZEVOBvgHbzgI2Hu2JlFbOk/ypTXVaJanvh5HZIkNVY4JZAurvwbwcuC0irgSe6V+ZUnpfU6KSJEmSmicNc38kewL3pJTmA0TEhcBhwLxhtj8a+ES27S7A+P7/BEwprapUzhLX/w0cA7xxFPGUV84VzO95Tw4xqXzswazBbJGhqjHBLAH1J5h/kS2SJElS1b0kIhZTq2Sekt0ne5xSShsO87rnAQ8NeLwA2GuoDSNiW2AG8Nts1Y7Aooi4NFt/FXBKSqkXOAm4PKX091jHL6kRcRxwHMAWW2xBd3f3SOdZmB0efpjNVqxg7nXXsSdw+7x5PNZAvFtsUbvN85SXLFlS6jHsBCuWLgXgf++6i791wLXYYt48Xghce911LH/ooRG3b7YyvgemzZvHi4Ebrr+eZxYuXOO57R96iC37+vhTgzFveffd7Axcc/XVrNh884b21Ygyjn8nyWv8d1u4kPHPPMPNXstR8eu/WEuW5N+RbZ0J5ojYJqX0YErpvNyPLEmSJBUgpTRujC8dKvs7XOnlUcAlWQIZaj9370+tZcaDwEXAsRFxBfBmYPZIB08pnQ2cDTBr1qw0u8yT41xyCYwbx5577AHAi170Imgg3v5cXJ4Tznd3d1PqMewAf/r5zwHY4QUvYIdOuBb33w/A3vvsAzNmFBsLJX0PLKq1rd/jZS+DmTPXfO6nP4Xx4xuPef58APbZay/YdtvG9tWAUo5/B8lt/DfZJJ+vyw7j13+xmpHcH2mSvzFPgCJJkiS1mQXAwBTndODhYbY9CvjRoNfeklKan1LqofZz9kupJZy3B+6JiPuB9SLinrwDb7mIXFtkvP3ttUXtxUn+tBYn+VPV2CJDAkZukTGmCVAkSZKkNnQDsENEzAD+Ri2JfMzgjSJiJ2AT4JpBr90kIjZLKT0GvBK4MaX0C2DLAa9dklLavonn0BpdXU7yp5F1Wi9c3w8jc5I/VY0JZgkYuYJ5rBOgSJIkSW0lqzw+CfgVcAdwcUrp9oj4dES8fsCmRwMXppQGFGimXuBk4DcRcRu1Qo5zWhd9i+U8yZ/alJP8aTAn+VPVmGCWgJErmMc6AYokSZLUdlJKc4A5g9adOujxJ4d57ZXAbiPsf2qDIZZDhBXMGlHqtEpS3w8ja0UFsy0ylCcTzBIwQoK5gQlQJEmSJHUqK5hVj06rJPX9MLJWVDB32h821FwmmCVg5ApmSZIkSRqdnCuYP/ShHGJS6TjJn9biJH+qGhPMEmCCWZIkSVLecp7k79BDc4hJ5WMFswZzkj9VjQlmCRh5kj9JkiRJGp2IXFtk3HVXbVGbcZI/DTZSi4w8dNofNtRcJpglwApmSZIkSXnLuYL53e+u3XZ3NxaWysVJ/rQWJ/lT1ZhglgArmCVJkiTlzUn+VA8TzBrMSf5UNSaYJcAEsyRJkqS85TzJn9qULTI0mJP8qWpMMEuACWZJkiRJebOCWfXqb6fSCXw/jMxJ/lQ1JpglwASzJEmSpLwNTuD4y7eG0z8hZCcwwTyyVrTIcJI/5ckEswQ4yZ8kSZKkvA1O4DT4y/fHP95gPCovK5g1kJP8qWpMMEuACWZJkiRJeev/Zbu3d83HY3TggQ3Go/Lqb6fSCUwwj6yVk/x97Wuw1VbDb/O2t8GuuzZ+PLU3E8wSYIJZkiRJUt5yrmCeO7d2O3NmQ7tRGfVPCNkJTDCPrBWT/L3gBTBtGlx66fDbLFsGTz4JZ5/d+PHU3kwwS4AJZkmSJEl5608S5VTB/P731267uxvajcrIFhkaqBUtMnbaCR57bN3bTJ+++vuXtC4mmCWgBZP8RcS4iLglIn7e7GNJkiRJKgEn+VO9nORPA7WiRUY9Oql1ixpjglkCWpBgBv4duKMFx5EkSZJUBjlXMKuNdVIFcz/fD8NrRQVzPTqpdYsaY4JZApqcYI6I6cA/Ad9u5nEkSZIklUjOk/ypjXVSpagVzCMrUwWzCWbVwwSzBDS/gvk04MPAsN+ZI+K4iLgxIm58bKQ+SJIkSZLKL+dJ/tTGOqlS1ATzyNY1yR/YIkPlY4JZApo4yV9EvA74R0rppoiYPdx2KaWzgbMBZs2a5XdwSZIkqepyrmD+3OcajEfl1UmVoiaYR2aLDFWNCWYJaGKCGdgXeH1EHAJMBjaMiB+klN7WxGNKkiRJKlrOFcwvf3mD8ai8nORPA5WpRUanfF2qMSaYJaCJLTJSSh9NKU1PKW0HHAX81uSyJEmS1AFynuTv6qtri9qQFcwayApmVY0JZglobgWzJEmSpE40uAqxwV++P/ax2m13d0O7URl1UqWoCeaRlamC2QSz6mGCWQJalGBOKXUD3a04liRJkqSC5VzBrDbWSZWiJphHtq5J/myRoTIywSwBTWyRIUmSJKlD5VzBrDbWSYk8E8wjs0WGqsYEswSYYJYkSZKUNyuYVa9OSuSZYB5ZmVpkdMofPtQYE8wSYIJZkiRJUt76f9k2wayRdFKvWxPMI7OCWVVjglkCnORPkiRJUt4G91Ft8Jfv005rMB6VVydVippgHlmZKphNMKseJpglwASzJEmSpLzlnGCeObPBeFRenVQp2imJ9EY4yZ+qxgSzBNgiQ5IkSVLecp7k76qraovaUCcl8qxgHpktMlQ1JpglwApmSZIkSXnLeZK/z3ymdnvggQ3tRmXUSYk8E8wjs0WGqsYEswRYwSxJkiQpbzlXMKuNdVIizwTzyMpSwdxJlfVqjAlmCTDBLEmSJClvOVcwq411UiKvU86zEWWpYO6kyno1xgSzBJhgliRJkpS3/l+2TTBrJJ2UyDMRNTIn+VPV+L6WABPMkiRJkvI2OEnkL98aTicl8kxEjawsLTI66Q8faozvawlwkj9JkiRJecu5Rca3vtVgPCqvCLj2WnjrW4uOpPluvdVE1EjK0iKjk3qDqzEmmCXABLMkSZKkvOU8yd9OOzUYj8rrkEPgpz+F668vOpLWOOywoiMot7JUMHdSZb0aY4JZAkwwS5IkScpbzhXMP/tZ7fbQQxvajcroi1+sLRKUp4I5YvX3L2ldTDBLgAlmSZIkSXnLuYL5y1+u3ZpgltpcmSb56+lpzbFUbSaYJcBJ/iRJkiTlbXAFsyTVoywtMpzkT/UywSwBJpglSZIk5a3/l+2cWmRI6hBlaZHhJH+qlwlmCTDBLEmSJClvg//N3V++JdWjLBXMTvKneplglgATzJIkSZLylvMkf5I6RFkqmG2RoXqZYJYAJ/mTJEmSlLecJ/k7//wG45FUDWWa5M8KZtXDBLMEmGCWJEmSlLecK5i33rrBeCRVQ1laZFjBrHqZYJYAW2RIkiRJylvOFcwXXVRbJLW5srTIcJI/1csEswRYwSxJkiQpbzlXMH/zm7XbI49saDeSyq4sFcy2yFC9TDBLgBXMkiRJkvLW/8u2k/xJGo2yVDDbIkP1MsEsAVYwS5IkScrb4Im6/OVbUr2GS+7aIqN5nn4aLrwQVq4sOpKW2eruu+H22xvf0bJlfsZJmGCWJEmSlLecW2RI6iDDtaewRUbz/OQncNxxRUfRUjvmuTNnopVMMEuSJEnKWc6T/EnqIGWoYO60FhkrVtRu//IX2HLLYmNpkT//+c/su+++je8oAjbdtPH9SBVnglmSJElSvnKuYL7kkgbjkVQdVjC3Xv+5TpsGm21WbCwt8uzGG3fMuUqt4CR/kiRJUp0i4rURcVdE3BMRpwzx/FcjYm623B0RiwY8t01E/Doi7oiIeRGxXbb+OxFxa0T8JSIuiYiprTujJsm5gnnatNoiqQNYwdx6/Qlm/9tE0hhZwSxJkiTVISLGAWcCrwYWADdExOUppXn926SUPjBg+/cCuw/YxfeBz6aUrsySyP3Ziw+klBZnr/kKcBLw+aaeTLPlXMF87rm122OPbWg3kqpguAn2nOSveUwwS2qQFcySJElSffYE7kkpzU8prQQuBA5bx/ZHAz8CiIhdgPEppSsBUkpLUkpLs/v9yeUApgDV/7/s/iRFjgnm/iSzpDZni4zWM8EsqUFWMEuSJEn1eR7w0IDHC4C9htowIrYFZgC/zVbtCCyKiEuz9VcBp6SUerPtvwccAswDPjTMPo8DjgPYYost6O7ubvB0mmfD22/npcA/HnmEzYFrrr2WFffdN+b9LVo0E4Du7rn5BAgsWbKk1GPYCbwGxSrr+O/X18ffH3iAewfF9tLFi3kWuK0FMe/86KNstHQp1zXxWGUa/+fdfTc7AH+++upab+IOUKbx70SOf7GWLFmS+z5NMEuSJEn1Gaq0a7gSt6OAS/oTyNR+7t6fWsuMB4GLgGOB7wCklN6ZteA4AzgS+N5aB0rpbOBsgFmzZqXZs2eP9Tyab8oUADbfdFMA9nn5y2Hrrce8u/58R57n3N3dnev+NHpeg2KVdvwnTGDr6dPZenBsU6fCppu2JuZzz4W7727qsUo1/rfdBsC+++3XMQ3vSzX+HcjxL1Yzkvu2yJAkSZLqswAYmCWdDjw8zLZHkbXHGPDaW7L2Gj3AZcBLB74gS0ZfBByeW8RFyXmSP0kdxEn+Ws8WGZIaZIJZkiRJqs8NwA4RMSMiJlJLIl8+eKOI2AnYBLhm0Gs3iYjNssevBOZFzfbZ6wI4FLiziefQGv2T/D36aO3WpIWkejnJX+uZYJbUIFtkSJIkSXVIKfVExEnAr4BxwHdTSrdHxKeBG1NK/cnmo4ELU1o9Q1RKqTciTgZ+kyWSbwLOodZ247yI2DC7fyvwntadVZNstFHt9ppraoma9ddvaHdz5uQQk6RqcJK/1jPBLKlBJpglSZKkOqWU5gBzBq07ddDjTw7z2iuB3YZ4at+84iuNHXaAW2+FRYtg881XN1Eeo/XWyykuSeVni4zWM8EsqUEmmCVJkiTlb7ehculj841v1G5POCG3XUoqKyuYW88Es6QG2YNZkiRJUqldfHFtkdQBrGBuPRPMkhpkglmSJEmSJJWDk/y1nglmSQ0ywSxJkiRJksrBFhmtZ4JZUoNMMEuSJEmSpHKwRUbrmWCW1CATzJIkSZIkqRzKUsFsglmS6ja+6AAkSZIkaV26u4uOQFLLlKGC2RYZkjQqVjBLkiRJkqRyKMMkf7bIkKRRMcEsSZIkqdS+9KXaIqkDlKVFhhXMklQ3E8ySJEmSSu3nP68tkjpAGVpkWMEsSaNiD2ZJkiRJklQOXV2wYAHMmbPm+iVLnOSvWUwwS2qQCWZJkiRJklQOG28Mf/hDbRnquVawRYYkjYoJZkmSJEmSVA5z5sC99w793K67tiYGW2RI0qiYYJYkSZJUalOmFB2BpJbZdNPaUiQrmCVpVEwwS5IkSSq1K64oOgJJHcUKZkkala6iA5AkSZIkSSqNrixV0ilVzCaYJTXIBLMkSZKkUvuv/6otktQSnZpglqQxMsEsSZIkqdR+85vaIkkt0V/J2yltMlKyellSQ0wwS5IkSZIk9evECmYTzJIaYIJZkiRJkiSpnxXMkjQqJpglSZIkSZL69Vcwm2CWpLqMLzoASZIkSVqXTTctOgJJHcUWGZI0KiaYJUmSJJXaT35SdASSOootMiRpVGyRIUmSJEmS1M8KZkkaFRPMkiRJkkrtox+tLZLUElYwS9Ko2CJDkiRJUqldc03REUjqKE7yJ0mjYgWzJEmSJElSP1tkSNKomGCWJEmSJEnqZ4sMSRoVE8ySJEmSJEn9bJEhSaNiD2ZJkiRJpTZ9etERSOootsiQpFExwSxJkiSp1H7wg6IjkNRRbJEhSaNiiwxJkiRJkqR+VjBL0qiYYJYkSZJUau9/f22RpJawglmSRsUWGZIkSZJKbe7coiOQ1FGc5E+SRsUKZkmSJEmSpH62yJCkUTHBLEmSJEmS1M8WGZI0KiaYJUmSJEmS+lnBLEmjYg9mSZIkSaW2445FRyCpo1jBLEmjYoJZkiRJUqmdfXbREUjqKE7yJ0mjYosMSZIkSZKkfrbIkKRRMcEsSZIkqdSOO662SFJL2CJDkkbFFhmSJEmSSu3uu4uOQFJHsYJZkkbFCmZJkiRJkqR+VjBL0qiYYJYkSZIkSernJH+SNCommCVJkiRJkvrZIkOSRsUezJIkSZJKbebMoiOQ1FFskSFJo2KCWZIkSVKpnXZa0RFI6ij9FcyXXgo33dSUQzz3rrvgnntG3nC//WDnnZsSwyommCU1yASzJEmSJElSvy23rN1+5jNNO8RO9W746lfDr3/dtDgAE8ySGmaCWZIkSVKpve1ttdsf/KDYOCR1iD33hMceg+XLm3aIa665hn322WfdG73lLU2NYRUTzJIaZIJZkiRJUqktWFB0BJI6zrRpTd39is02g+nT173RlCmwcmVT4wBMMEtqWFfRAUiSJElVERGvjYi7IuKeiDhliOe/GhFzs+XuiFg04LltIuLXEXFHRMyLiO2y9T/M9vnXiPhuRExo3RlJkkorojUTDZpgltQgE8ySJElSHSJiHHAmcDCwC3B0ROwycJuU0gdSSjNTSjOBM4BLBzz9feC/U0ovBPYE/pGt/yGwM7ArMAX4t6aeiCSpGrq6asnfZjPBLKlBJpglSZKk+uwJ3JNSmp9SWglcCBy2ju2PBn4EkCWix6eUrgRIKS1JKS3N7s9JGeB6YIT/mZYkdQQrmCVVhAlmSZIkqT7PAx4a8HhBtm4tEbEtMAP4bbZqR2BRRFwaEbdExH9nFdEDXzMBeDvwy9wjr7h99qktktRRurpMMEuqBCf5kyRJkuoz1G/fw/3v8lHAJSml3uzxeGB/YHfgQeAi4FjgOwNe8w3gDymlPw558IjjgOMAtthiC7q7u0cZfnW95jW12zxPecmSJR01hmXkNSiW41+sesZ/1yefZMLixdzc5Ou0y6OPMnX5cq7voK8Hv/6L5fgXa8mSJbnv0wSzJEmSVJ8FwNYDHk8HHh5m26OAEwe99paU0nyAiLgM2JsswRwRnwA2A9493MFTSmcDZwPMmjUrzZ49e0wnoZru7m4cw2J5DYrl+BerrvGfNg2efbb51+mb34SHH+6orwe//ovl+BerGcl9W2RIkiRJ9bkB2CEiZkTERGpJ5MsHbxQROwGbANcMeu0mEbFZ9viVwLxs+38DXgMcnVJqwf9CV8/hh9cWSeooTvInqSJMMEuSJEl1SCn1ACcBvwLuAC5OKd0eEZ+OiNcP2PRo4MJs0r7+1/YCJwO/iYjbqLXbOCd7+ixgC+CaiJgbEae24HQqZeHC2iJJHcVJ/iRVRNNaZETE1sD3gS2BPuDslNLXmnU8SZIkqdlSSnOAOYPWnTro8SeHee2VwG5DrLdtnSRpbU7yJ6kimvnDbA/woZTSzRGxAXBTRFyZUprXxGNKkiRJkiRVny0yJFVE01pkpJT+nlK6Obv/NLV/I3xes44nSZIkSZLUNmyRIakiWvLveBGxHbA7cN0Qzx0HHAewzTbbtCIcSZIkSRXyqlcVHYEkFcAWGZIqoukJ5oiYCvwEeH9KafHg51NKZwNnA8yaNasF//shSZIkqUr+8z+LjkCSCmCLDEkV0bQWGQARMYFacvmHKaVLm3ksSZIkSZKktmGLDEkV0bQEc0QE8B3gjpTSV5p1HEmSJEnt7eCDa4skdRQrmCVVRDMrmPcF3g68MiLmZsshTTyeJEmSpDa0bFltkaSOYgWzpIpoWg/mlNKfAL9DSZIkSZIkjZaT/EmqiKb2YJYkSZIkSdIY2CJDUkWYYJYkSZIkSSobW2RIqoimtciQJEmSpDy87nVFRyBJBbCCWVJFmGCWJEmSVGonn1x0BJJUAHswS6oIW2RIkiRJkiSVjS0yJFWECWZJkiRJpTZ7dm2RpI5iiwxJFWGCWZIkSZIkqWysYJZUESaYJUmSJEmSysYezJIqwgSzJEmSJElS2dgiQ1JFmGCWJEmSJEkqG1tkSKqI8UUHIEmSJEnr8pa3FB2BJBXACmZJFWGCWZIkSVKpnXBC0RFIUgGsYJZUEbbIkCRJklRqS5fWFknqKE7yJ6kirGCWJEmSVGqHHFK77e4uNAxJaq1Wtsjosv5Q0tj5HUSSJEmSJKlsbJEhqSJMMEuSJEmSJJWNLTIkVYQJZkmSJEmSpLJpZYsME8ySGmCCWZIkSZIkqWxskSGpIpzkT5IkSVKpHXts0RFIUgGsYJZUESaYJUmSJJWaCWZJHckKZkkVYYsMSZIkSaX2+OO1RZI6SleWsml2FbMJZkkNsoJZkiRJUqkdcUTttru70DAkqbUGJpibmQA2wSypQVYwS5IkSZIklU1/0rfZbTJMMEtqkAlmSZIkSZKksrFFhqSKMMEsSZIkSZJUNlYwS6oIE8ySJEmSJEll01/BbIJZUsk5yZ8kSZKkUnvPe4qOQJIKYIsMSRVhglmSJElSqR15ZNERSFIBbJEhqSJskSFJkiSp1B56qLZIUkexRYakirCCWZIkSVKpvf3ttdvu7kLDkKTWskWGpIqwglmSJEmSJKlsbJEhqSJMMEuSJEmSJJWNFcySKsIEsyRJkiRJUtlYwSypIkwwS5IkSZIklY2T/EmqCCf5kyRJklRqH/pQ0RFIUgFskSGpIkwwS5IkSSq1Qw8tOgJJKoAtMiRVhC0yJEmSJJXaXXfVFknqKFYwS6oIK5glSZIkldq731277e4uNAxJai0rmCVVhBXMkiRJkiRJZeMkf5IqwgSzJEmSJElS2dgiQ1JFmGCWJEmSJEkqG1tkSKoIE8ySJEmSJEllY4sMSRXhJH+SJEmSSu3jHy86AkkqgC0yJFWEFcySJElSnSLitRFxV0TcExGnDPH8VyNibrbcHRGLBjy3TUT8OiLuiIh5EbFdtv6kbH8pIqa17myq48ADa4skdRRbZEiqCCuYJUmSpDpExDjgTODVwALghoi4PKU0r3+blNIHBmz/XmD3Abv4PvDZlNKVETEV6M8Y/Bn4OdDd3DOorrlza7czZxYbhyS1lBXMkirCBLMkSZJUnz2Be1JK8wEi4kLgMGDeMNsfDXwi23YXYHxK6UqAlNKS/o1SSrdk2zQv8op7//trt93dhYYhSa1lBbOkijDBLEmSJNXnecBDAx4vAPYaasOI2BaYAfw2W7UjsCgiLs3WXwWcklLqrffgEXEccBzAFltsQXcHZVsXLaqVLnd3z81tn0uWLOmoMSwjr0GxHP9i1TP+m91xBy8Crr/2WpY++mjTYtl7+XKefOQR7uqgrwe//ovl+BdryZIlI280SiaYJUmSpPoMVd413P8tHwVcMiCBPB7Yn1rLjAeBi4Bjge/Ue/CU0tnA2QCzZs1Ks2fPrvellbfxxrXbPM+5u7s71/1p9LwGxXL8i1XX+D/+OAB77rEHvPjFzQtm4kSeu9VWPLeDvh78+i+W41+sZiT3neRPkiRJqs8CYOsBj6cDDw+z7VHAjwa99paU0vyUUg9wGfDSpkQpSWoPtsiQVBEmmCVJkqT63ADsEBEzImIitSTy5YM3ioidgE2Aawa9dpOI2Cx7/EqG790sSZKT/EmqDFtkSJIkSXVIKfVExEnAr4BxwHdTSrdHxKeBG1NK/cnmo4ELU1qdEUgp9UbEycBvojab303AOQAR8T7gw8CWwF8iYk5K6d9ad2aq/GZhAAAdMUlEQVTl97nPFR2BJBWgP8FsBbOkkjPBLEmSJNUppTQHmDNo3amDHn9ymNdeCew2xPrTgdPzi7L9vPzlRUcgSQWwRYakirBFhiRJkqRSu/rq2iJJHcUWGZIqwgpmSZIkSaX2sY/Vbpsw6bkklZcVzJIqwgpmSZIkSZKksrEHs6SKMMEsSZIkSZJUNrbIkFQRJpglSZIkSZLKxhYZkirCBLMkSZIkSVLZWMEsqSKc5E+SJElSqZ12WtERSFIBrGCWVBEmmCVJkiSV2syZRUcgSQVwkj9JFWGLDEmSJEmldtVVtUWSOootMiRVhBXMkiRJkkrtM5+p3R54YLFxSFJL2SJDUkVYwSxJkiRJklQ2tsiQVBEmmCVJkiRJksrGFhmSKsIEsyRJkiRJUtnYIkNSRZhgliRJkiRJKhsrmCVVhJP8SZIkSSq1b32r6AgkqQBWMEuqCBPMkiRJkkptp52KjkCSCuAkf5IqwhYZkiRJkkrtZz+rLZLUUWyRIakirGCWJEmSVGpf/nLt9tBDi41DklrKFhmSKsIKZkmSJEmSpLKxgllSRZhgliRJkiRJKhsrmCVVhAlmSZIkSZKksnGSP0kVYYJZkiRJkiSpbGyRIakinORPkiRJUqmdf37REUhSAfqTvqefDpdf3rzjmGCW1CATzDm67r4nuLf3wbq2PWavbZocjSRJktQett666AgkqQBbbw177QWPPFJbmmXHHWGffZq3fylHzz77LAsWLGD58uVFh1IJkydPZvr06UyYMKGpxzHBLEmSJKnULrqodnvkkcXGIUktNXUqXHtt0VFIpbJgwQI22GADtttuO8LK+3VKKbFw4UIWLFjAjBkzmnosezBLkiRJKrVvfrO2SJKkzrZ8+XI23XRTk8t1iAg23XTTllR7m2CWJEmSJEmSVAkml+vXqrEywSxJkiRJkiRJGhMTzJIkSZIkSZJUh9NPP50XvvCFHH744eyzzz5MmjSJL33pS0WHVSgn+ZMkSZIkSZKkOnzjG9/giiuuYP311+eBBx7gsssua+nxe3p6GD++XCldK5hz8te/PcWZ9z2XK/76d66/7wnufWwJi5aupC+lokOTJEmSKu2SS2qLJElSkY4//njmz5/P61//en74wx+yxx57MGHChHW+5ve//z0zZ85k5syZ7L777jz99NMAfPGLX2TXXXflJS95CaeccgoAc+fOZe+992a33XbjjW98I08++SQAs2fP5mMf+xgHHHAAX/va13jsscc4/PDD2WOPPdhjjz3485//3NwTH0G50t0Vdf/jz/CO717PkmUb0LNoIb19q5PKz91oMu94+XZsOHndX2ySJEmShjZtWtERSJKk0nn/+2Hu3Hz3OXMmnHbasE+fddZZ/PKXv+R3v/sd0+r8AeVLX/oSZ555Jvvuuy9Llixh8uTJXHHFFVx22WVcd911rLfeejzxxBMA/PM//zNnnHEGBxxwAKeeeiqf+tSnOC2LZ9GiRfz+978H4JhjjuEDH/gA++23Hw8++CCvec1ruOOOOxo8+bEzwdyghUtWcOz3rqcvJT7/wvt4ZofDWLzsWRY+s5JHnlrOlfMe5ew/zOdf9p3Bc9afWHS4kiRJUuWce27t9thji4xCkiRp9Pbdd18++MEP8ta3vpU3velNTJ8+nauuuop3vvOdrLfeegA85znP4amnnmLRokUccMABALzjHe/gzW9+86r9HHnkkavuX3XVVcybN2/V48WLF/P000+zwQYbtOis1mSCuQFLV/bwL+fdyN+fWs4F79qbnutv494INl5vIhuvN5EXbDaVbZ6zHudefT9n/+Fe3rnvDLbYcHLRYUuSJEmVYoJZkiStZR2VxkU688wzOeeccwCYM2cOp5xyCv/0T//EnDlz2HvvvbnqqqtIKRERo9rv+uuvv+p+X18f11xzDVOmTMk19rGyB/MY9fT28d4LbuG2BYs44+jdedm2mwy53dbPWY93veL5JOCcP85nwZNLWxuoJEmSJEmSpJY48cQTmTt3LnPnzmWrrbbi3nvvZdddd+UjH/kIs2bN4s477+Sggw7iu9/9LkuX1vKETzzxBBtttBGbbLIJf/zjHwE4//zzV1UzD3bQQQfx9a9/fdXjuXm3ChklK5jHIKXEqZffzm/u/Af/9YYXc9CLtlzn9ltuOJnj9n8+3/3zfXz7T/dxzJ7btChSSZIkSZIkSXl75JFHmDVrFosXL6arq4vTTjuNefPmseGGG66x3Wmnncbvfvc7xo0bxy677MLBBx/MpEmTmDt3LrNmzWLixIkccsghfO5zn+O8887j+OOPZ+nSpTz/+c/ne9/73pDHPv300znxxBPZbbfd6Onp4RWveAVnnXVWK057SCaYx+D8ax/gguse5D2zX8Db9962rtdsOnUSx73iBZx39f2cd/X9bDhlAv/+qh0Y1zW6cvjCLH0CogumbFx0JJIkSZIkSVIh7r///lX3FyxYMOL2Z5xxxpDrTznlFE455ZQ11s2cOZNrr712rW27u7vXeDxt2jQuuuiikYNtEVtkjNJNDzzBp382j1ftvDn/cdBOo3rtRlMmcPwBL2D3bTbh9N/8L2//znU89vSKJkW6Dnf/Cp64b+Ttlj5RWwB+8UH49oHNjUuSJEmSJElSpZhgHoXHnl7BCT+8medtMoWvHDmTrnVUH094djEbPHM/Xb0r11g/cXwXR7xsOl88YjdufvBJDjn9j1xz78Jmh76mC94C39h75O2+OKO2AKx4GiZNbW5ckiRJ0hDmzKktkiRJKh9bZNSpp7ePky64maeWPcv3jt2TjaZMGHK7bR++gg2fuY9d7/kmAPc/92CunvnFtbZ7y6yt2W36Rpzww5s55tvXst/20zjiZdN57dR7mbTwDlh/M3jxm/I/kd5nsxNavvZzZ7wM+nrg329d+7kVS2DSBvnHI0mSJI1gvfWKjkCSJJVFSomIirScLVhKqSXHsYK5Tl/81V1cd98TfPYNu7LLVhsOu92+t354VXIZYLu/XwEDLuZW//gD43pryd2dt9yQy0/aj39/1Q7Mf+wZzrvoYib94FC44sNwyTt58okmVDaveHr45xbeA0/eP/zrJppgliRJUut94xu1RZIkdbbJkyezcOHCliVOqyylxMKFC5k8eXLTj2UFcx1+duvDnP2H+bx97205/GXTh91uk8V3DLn+mF/uxq/3Pp/ergnMvulE7tr2aHh5bWbHqZPG8/4Dd+R9m/+Frks/ucbrjvrvi1m4/vZsv/lUtt98Ks/daAobTZmwxrJh/+3k8YwfV8ffC1Ysrvu817DyaSuYJUmSVIiLL67dnnBCsXFIkqRiTZ8+nQULFvDYY48VHUolTJ48menTh89l5sUE8zo8vmQFX7jiTn580wJ232Zj/vN1uwy/cc8Kdnzox2usWjZpGlNWPA7Alo9fw6INdgRgg2ce5ILrHlxj22Ou+Ne1dnnwjPH8qXd9Vvb0cfnch1m8vGed8U6dNJ4NJ49n/Un9yzjWnzieqZPGs96kcaw/aTzbrpzPMdn2P7v14VXbrD9pPC/O1j+8aBkTuxLTssd9vX10rTDBLEmSJEmSpOJMmDCBGTNmFB2GBmlqgjkiXgt8DRgHfDul9PlmHi8vPb19/ODaB/jylXez/Nle3n3A83nfK3dg4vh1VAgvemitVcsmrk4w93VNYP3lf6/tf1x9TeT22KyXzbda/VeGZ3v7WPZsL8tW9rI8u132bG1Zmt1fvrKXlb19PLOihyeeWcnKnj5W9PSyoqePlT197MHtHDOxtr/3/uiWVfseTw/3ZBXzsz//Kyazkr9kj1/4f/+H2yYt5rzr/8GZN/2aCeO6mDiui0nju5jYv4yr3U5YdRuM7+pifFcwflwwflwXE7pqt6vWdWXb9a/Lnl/12gG3Q64b8Ny47Fjjsv3UbrsYN672uCuy23VMzChJkiRJkiRpdJqWYI6IccCZwKuBBcANEXF5Smles445Vs/29vHQE0u5f+Ez3P/4Un580wLu+Pti9t9hGp849EVsv/nUkXey6P61Vi2f9BzIWh5v+tRtrJi4CQCbPH0npD6IWsJ6XM/SIXc5eeUTazyeMK6WwN1w8tATDK5L9PWw8dP/y8FXf3bVul9s8yOu3eIoVqRx7PO3C+DJ2vp37tLFjGfugEdrj7+8xa+Z+FQvG234HHaeuiG9fX309CZ6+hK9fYnlz/ayZEUPvb2J3pToyW77+hJ9qbZNb4K+vjXX9xXQLieCNRPQaySkI0tI19aPi2z9uCES1wMfD0pwj4tYldgOICKIgCDoiloMXbUVdEVtm65V26zeftVzWVK8a9X62r4ism3J1kVt/2S3/dsM3n6NGNY49urnYlBcg+ONNe4PiDc7Hqv2Nej8+o/TNfTxYlDsa+538HgO3O+gMcsew1Cvx8kAJEmSJEmSctLMCuY9gXtSSvMBIuJC4DCgNAnmG+5/gpN/fCsLnlxG74Bs5/RNpvDNt76U1754y/oSUT0r4bqz11rdGxNX3d/60d+uur/B0oc45pcv4ckNdiTRxYbP3Dfkbl92xxeY8bfL6e2aSIouaim1mkQQJLr6eujrGk9X6iFSH6Q+gkSkXrr6ehjXt5xJK59kQu+ytfb/on/8jBf942drrf/o/Lev8fh1T10AwEs2WUnPLluteyxGoZZkTvT1MSARXUtA9/bVEtBrPh6wfo3kdSIN3E+WvB6YyO5bldgeeNzBj4d+/Ypn0+r1fQO2X+vx6tf0J9Rr1yqtmucxsXoGz5SGfqzWGJiUD/oT6AMT3wMS2V2DE9drJuqH+i5Rz/eOwZsM9ZKh9r7W6+o4/pDR5LSfteMZOeYhwxnDscZ6/KHHY2zHL5OShwdU4w88ZY+w7EO4zwum8cFX71h0GJIkSZJaJJo162JEHAG8NqX0b9njtwN7pZROGrTdccBx2cOdgLuaElBrTAMeLzoIjYnXrrq8dtXkdasur111tfu12zaltFnRQbRKRDwGPFB0HBXX7u+JKvAaFMvxL5bjXyzHv1iOf7GmAevn+bNzMyuYh6qvWSubnVI6G1i7/LeCIuLGlNKsouPQ6HntqstrV01et+ry2lWX1669dFIyvVl8TxTPa1Asx79Yjn+xHP9iOf7FysZ/uzz3uY5Z6xq2ANh6wOPpwMNNPJ4kSZIkSZIkqYWamWC+AdghImZExETgKODyJh5PkiRJkiRJktRCTWuRkVLqiYiTgF8B44DvppRub9bxSqItWn10KK9ddXntqsnrVl1eu+ry2klr8j1RPK9BsRz/Yjn+xXL8i+X4Fyv38W/aJH+SJEmSJEmSpPbWzBYZkiRJkiRJkqQ2ZoJZkiRJkiRJkjQmHZlgjojXRsRdEXFPRJwyxPOTIuKi7PnrImK7Ac99NFt/V0S8ZqR9ZpMcXhcR/5vtc+JYj6FyX7uI2DQifhcRSyLi680bhWoq+bV7dUTcFBG3ZbevbN5IVEvJr9ueETE3W26NiDc2bySqp8zXbsDrtsm+Z56c/whUV5mvXURsFxHLBrz3zmreSEiSJEmqhJRSRy3UJhy8F3g+MBG4Fdhl0DYnAGdl948CLsru75JtPwmYke1n3Lr2CVwMHJXdPwt4z1iOUfS4lWGpwLVbH9gPOB74etHjVaalAtdud2Cr7P6Lgb8VPWZlWCpw3dYDxmf3nwv8o/9xpy9lv3YDYvgJ8GPg5KLHrCxL2a8dsB3w16LHycXFxcXFxcXFxcWlPEsnVjDvCdyTUpqfUloJXAgcNmibw4DzsvuXAK+KiMjWX5hSWpFSug+4J9vfkPvMXvPKbB9k+3zDGI+hkl+7lNIzKaU/AcvzPe22UPZrd0tK6eFs/e3A5IiYlNvZV1fZr9vSlFJPtn4y4Ky1q5X62gFExBuA+dTec1qt9NdOamfZf1ZcHhHfHVTtv37U/svpdUXG1+6GGv+IeENEnBMR/xMRBxUdYzsbZvzXj4jzsmvw1qJjbGcRsX9EnBUR346Iq7N1Q35PUv6GGf+uiPhsRJwREe8oOsZ2NtT4Z+v9/G2RYd4DdX8Gd2KC+XnAQwMeL8jWDblNlrx4Cth0Ha8dbv2mwKIBCZCBxxrtMVT+a6fhVenaHQ7cklJaMaozbE+lv24RsVdE3A7cBhw/4PWdrtTXLiLWBz4CfGrMZ9i+Sn3tsudmRMQtEfH7iNh/bKcp5S9LwPwjIv46aP06284MsiPwi5TSv1D7r4B+H6FW8a9hNGv8U0qXpZTeBRwLHNmM2NtBE7/+3wRckl2D1+cfeXvIY/xTSn9MKR0P/JzVf+Qd7nuSBmji+B9G7WeiZ6n9nKQhNHH8wc/fujTrGozmM7gTE8xDVd8Mrnobbpu81o/lGCr/tdPwKnHtIuJFwBeAdw+xXScq/XVLKV2XUnoRsAfw0YiYPMS2najs1+5TwFdTSkuGeL7Tlf3a/R3YJqW0O/BB4IKI2HCIbaUinAu8duCKiBgHnAkcTC05c3RE7BIRu0bEzwctmwO3AEdFxG+B32X7OBCYBzzawnOponNpwvgP8PFsXxrauTRn/Kez+o+Uva04kYo6l8bHv98xwI+y++t6T2i1c2nO+O8EXJNS+iDwnqafRXWdSxPG38/fUTmX5rwH+o34GTy+0TOooAXA1gMeTwceHmabBRExHtgIeGKE1w61/nFg44gYn1X/DNx+LMfodGW/dhpe6a9dREwHfgr8c0rp3rGfalsp/XXrl1K6IyKeodZD+8bRn2rbKfu12ws4IiK+CGwM9EXE8pSSE6SW/NqllBKwAiCldFNE3Eutusr3nQqXUvpDDJpMlAEtYgAi4kLgsJTS/wPW+nfbqE06+olsX5cA3wP+D7W5NnYBlkXEnJRSX/POpJqaNf4REcDngStSSjc38xyqrIlf/wuofT7MpTML1OqSx/hn22wDPJVSWpyteidrXxMN0sTxXwCszO77B5ZhNHH8/fytU7OuwWg+gzvxA+IGYIeozZo+kdrENZcP2uZyoL+/zhHAb7NfqC6n9tfDSRExA9gBuH64fWav+V22D7J9/s8Yj6HyXzsNr9TXLiI2Bn4BfDSl9Odcz7zayn7dZmSJLyJiW2oVBvfnd/qVVuprl1LaP6W0XUppO+A04HMml1cp9bWLiM2yaggi4vnZMebneP5S3kbbfu6XwPsi4iyyz5SU0v9NKb0fuAA4x19uR6Xh8QfeCxxI7Q+TxzcjyDaWx/hfChweEd8EftaMINvYWNpf/itrJpGHuiaqTx7jfynwmog4A/hDvuG1vYbH38/fhuXxHqj7M7jjKphTSj0RcRLwK2qzqn83pXR7RHwauDGldDnwHeD8iLiHWkXQUdlrb4+Ii6mV6PcAJ6aUegGG2md2yI8AF0bEZ6j9e8t3svWjPkanK/u1y/Z1P7AhMDFqE1gdlFKa15wRqY4KXLuTgO2B/4yI/8zWHZRS+kczxqMqKnDd9gNOiYhngT7ghJTS480ajyqpwLXTMCpw7V4BfDoieqhV8hyfUvK/eFRmo2ptllL6K6v/6DL4uXNziqmTNDz+KaXTgdNzjqtT5DH+z1CrotXojbq1YkrpE4MeD/s9SSPKY/yXUku4afQaHv8B68/NI6AOlMd7oO7P4LDwUpIkSVI7yP499OcppRdnj/cBPplSek32+KMA2b+HKmeOf7Ec/2I5/sVy/Ivl+Bev6GvQiS0yJEmSJHWGetrOqHkc/2I5/sVy/Ivl+BfL8S9eS6+BCWZJkiRJlRcRPwKuAXaKiAUR8a+pNoFlf4uYO4CLB7SIUY4c/2I5/sVy/Ivl+BfL8S9eGa6BLTIkSZIkSZIkSWNiBbMkSZIkSZIkaUxMMEuSJEmSJEmSxsQEsyQ1WUT0RsTciPhrRPw4ItbLYZ/HRsTX84hPkiRJkiRprEwwS1LzLUspzUwpvRhYCRxfdECSJEmSJEl5MMEsSa31R2D7gSsioisi7o+IjQesuycitoiIQyPiuoi4JSKuiogtBu8wIs6NiCMGPF4y4P5/RMQNEfGXiPhUk85JkiRJkiR1KBPMktQiETEeOBi4beD6lFIf8D/AG7Pt9gLuTyk9CvwJ2DultDtwIfDhURzvIGAHYE9gJvCyiHhFDqciSZIkSZIEmGCWpFaYEhFzgRuBB4HvDLHNRcCR2f2jsscA04FfRcRtwH8ALxrFcQ/KlluAm4GdqSWcJUmSpDUMmDekfzllDPuYFRGnZ/dLP2dIRHRHxKyi4wCIiE9GxMlFxyFJYzG+6AAkqQMsSynNHLgiIk4E3pU9PAS4Btg+IjYD3gB8JnvuDOArKaXLI2I28Mkh9t9D9gfDiAhgYv9hgP+XUvpWfqciSZKkNrXWz6yjlVK6kVpRhSSpg1jBLEkFSCmdmU38NzOl9HBKKQE/Bb4C3JFSWphtuhHwt+z+O4bZ3f3Ay7L7hwETsvu/Av4lIqYCRMTzImLznE9FkiRJbSybK+QLEXF9tmyfrX9zRPw1Im6NiD9k62ZHxM+H2Me2EfGbbF6Q30TENtn6cyPi9Ii4OiLmD5xXpM7YJkfE9yLitmzOkv+TrT82Ii6NiF9GxP9GxBfr2NfR2X7+GhFfGLD+XyPi7qza+ZyhqrIj4oABld+3RMQG2foPZ/u8NSI+n617VzZHyq0R8ZOIWG+I/b0gi/2miPhjROw8mnGRpFYzwSxJ5XER8DZWt8eAWsXyjyPij8Djw7zuHOCAiLge2At4BiCl9GvgAuCarMXGJcAGzQldkiRJFTdlUIuMIwc8tziltCfwdeC0bN2pwGtSSi8BXj/Cvr8OfD+ltBvwQ+D0Ac89F9gPeB3w+VHGfCJASmlX4GjgvIiYnD03k1oLul2BIyNi6+F2EhFbAV8AXpm9bo+IeEO2/j+BvYFXU2s5N5STgROzCvD9gWURcTC1/0zcKxuj/iT3pSmlPbJ1dwD/OsT+zgbem1J6Wbbvb4w8FJJUHFtkSFKTpZSm1rndjdTaWgxc9z/UJgAcvO25wLnZ/Uep/dDb76MDtvsa8LXRxixJkqSOs64WGT8acPvV7P6fgXMj4mLg0hH2vQ/wpuz++axOtgJclk16PS8ithhlzPtRaylHSunOiHgA2DF77jcppacAImIesC3w/9u7nxebwjCA49/Hj8ZolF8rxY4oZRZsbGz8BxKyoSwUyUYhiZJIWWI1bJRSxgopMfJrZEGkrIbILKaEpEb0WJx3dNzuvd1uMhPfz+ae+573POc9Z/n2nOd52yLOGuBOZo6V+ReBiebYQ5n5oYxfrsWvuw+cLtddycx3EbEeOJ+ZX8v6PpS5KyPiGDAX6KP66vCX8vXhWqokk4nhnk5ehiRNFjOYJUmSJElSO9l4nJk7gUPAYuBpRCzoMt547TgaJ0bErlpW9aLG023uUY/7g/YJdq3itIv/S2aeAHYAvcCjUtIi+P05J1wAdpes66PArIbz04CPtXJ6/Zm5opN1SNJkcYNZkiRJkiS1s6n2+xCqOsGZOZyZh6lKubUsQQE8ADaX463AvU5v3Ni7pOH03RKPiFgGLAFedRq7Zpiq5NzCiJhOVW5jCHhcxudFxAxgQ7OLy7t4npknqZocLgduUvVDmV3mzC/T5wCjETFzYu0Nz/sZGImIjeW6iIhVXTyTJP01lsiQJEmSJEm9EfG09v9GZu4vxz0RMUyVpLaljJ2KiKVUmbq3gGfAuhax9wADEbEPGAO2/6E1nwHOlX4j34FtmTleKy3RkcwcjYgDwG2q57lWStUREcepNqDfAy+BT01C7C0NBn+UOdfLOvqBJxHxDbgGHKSq6TwMvAGe07xHylbgbEQcomrgfYnq/UrSlBSZzb7YkCRJkiRJ/7uIeA2szsxWDaf/aRHRl5lfSgbzIDCQmYOTvS5JmkoskSFJkiRJktTckZLZ/QIYAa5O8nokacoxg1mSJEmSJEmS1BUzmCVJkiRJkiRJXXGDWZIkSZIkSZLUFTeYJUmSJEmSJEldcYNZkiRJkiRJktQVN5glSZIkSZIkSV35CXwAHCwWZRRiAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1440x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Search in narrow window\n",
    "epsilon,bestF1,recall,precision,accuracy,x,y,dev_pval = find_bestF1(12,min_prob=np.min(x[y>bestF1*.995]),max_prob=np.max(x[y>bestF1*.995]),log=True)\n",
    "    \n",
    "plot_score(dev_pval,x,y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Scores after narrow search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epsilson: 3.695418926574653e-85 \n",
      "F1-score: 0.7682403433476395 \n",
      "Recall: 0.7276422764227642 \n",
      "Precision: 0.8136363636363636 \n",
      "Accuracy: 0.9981088795111104\n"
     ]
    }
   ],
   "source": [
    "# Scores\n",
    "print('Epsilson: {} \\nF1-score: {} \\nRecall: {} \\nPrecision: {} \\nAccuracy: {}'.format(epsilon,bestF1,recall,precision,accuracy))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Not much improvement in that narrow search range"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[56822,    41],\n",
       "       [   67,   179]], dtype=int64)"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Confusion matrix - Dev dataset\n",
    "pred=[1 if (x<epsilon) else 0 for x in dev_pval]\n",
    "confusion_matrix(df_dev['Class'],pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Search for optimum number of features that would give best F1-score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training the model with 1 feature/s...\n",
      "Training the model with 2 feature/s...\n",
      "Training the model with 3 feature/s...\n",
      "Training the model with 4 feature/s...\n",
      "Training the model with 5 feature/s...\n",
      "Training the model with 6 feature/s...\n",
      "Training the model with 7 feature/s...\n",
      "Training the model with 8 feature/s...\n",
      "Training the model with 9 feature/s...\n",
      "Training the model with 10 feature/s...\n",
      "Training the model with 11 feature/s...\n",
      "Training the model with 12 feature/s...\n",
      "Training the model with 13 feature/s...\n",
      "Training the model with 14 feature/s...\n",
      "Training the model with 15 feature/s...\n",
      "Training the model with 16 feature/s...\n",
      "Training the model with 17 feature/s...\n",
      "Training the model with 18 feature/s...\n",
      "Training the model with 19 feature/s...\n",
      "Training the model with 20 feature/s...\n",
      "Training the model with 21 feature/s...\n",
      "Training the model with 22 feature/s...\n",
      "Training the model with 23 feature/s...\n",
      "Training the model with 24 feature/s...\n",
      "Training the model with 25 feature/s...\n",
      "Training the model with 26 feature/s...\n",
      "Training the model with 27 feature/s...\n",
      "Training the model with 28 feature/s...\n"
     ]
    }
   ],
   "source": [
    "# Search for number of features with best F1-score\n",
    "eps,f1,rec,prec,acc = defaultdict(int),defaultdict(int),defaultdict(int),defaultdict(int),defaultdict(int)\n",
    "for i in range(28):\n",
    "    print('Training the model with {} feature/s...'.format(i+1))\n",
    "    eps[i+1],f1[i+1],rec[i+1],prec[i+1],acc[i+1],_,_,_ = find_bestF1(i+1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the results\n",
    "plt.figure(figsize=(20,8))\n",
    "x1 = [i+1 for i in range(28)]\n",
    "y1 = [f1[i] for i in x1]\n",
    "y2 = [rec[i] for i in x1]\n",
    "y3 = [prec[i] for i in x1]\n",
    "y4 = [acc[i] for i in x1]\n",
    "\n",
    "plt.plot(x1,y1,label='Best F1',color='lightblue')\n",
    "plt.plot(x1,y2,label='recall',color='lightgreen')\n",
    "plt.plot(x1,y3,label='precision',color='red')\n",
    "plt.plot(x1,y4,label='Accuracy',color='black')\n",
    "plt.axvline(x=x1[np.argmax(y1)],ls='--',color='red')\n",
    "best_feat = x1[np.argmax(y1)]\n",
    "best_eps = eps[np.argmax(y1)]\n",
    "plt.text(x1[np.argmax(y1)]*.97,0.9,'Features='+str(best_feat),rotation=90)\n",
    "plt.text(x1[np.argmax(y1)]*1.03,0.83,'Epsilon='+str(best_eps),rotation=90)\n",
    "plt.xlabel('No of features')          \n",
    "plt.ylabel('Score')\n",
    "plt.title('Score vs different no of features')\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show();"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final figures on dev data\n",
    "using epsilon against optimum number of features for best F1-score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimum no of Features: 9 \n",
      "Epsilon: 2.084402716250594e-54, \n",
      "F1-score: 0.8044444444444444, \n",
      "Recall: 0.7357723577235772, \n",
      "Precision: 0.8872549019607843, \n",
      "Accuracy:0.9984590870090528\n"
     ]
    }
   ],
   "source": [
    "#Print Best scores\n",
    "print('Optimum no of Features: {} \\nEpsilon: {}, \\nF1-score: {}, \\nRecall: {}, \\nPrecision: {}, \\nAccuracy:{}'\n",
    "      .format(best_feat,best_eps,y1[np.argmax(y1)],y2[np.argmax(y1)],y3[np.argmax(y1)],y4[np.argmax(y1)]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[56834,    29],\n",
       "       [   61,   185]], dtype=int64)"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Prediction on dev data\n",
    "impcol=Feat_imp[:best_feat]['columns']\n",
    "train_mean,train_var,train_cov = df_train[impcol].mean(),df_train[impcol].var(),df_train[impcol].cov()\n",
    "\n",
    "dev_pred = [1 if (x<best_eps) else 0 for x in \n",
    "             multivariate_normal(mean=train_mean, cov=train_cov).pdf(df_dev[impcol])\n",
    "             ]\n",
    "confusion_matrix(df_dev['Class'],dev_pred)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prediction on test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[56836,    27],\n",
       "       [   52,   194]], dtype=int64)"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Prediction on test data\n",
    "test_pred = [1 if (x<best_eps) else 0 for x in \n",
    "             multivariate_normal(mean=train_mean, cov=train_cov).pdf(df_test[impcol])\n",
    "             ]\n",
    "confusion_matrix(df_test['Class'],test_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test data --- \n",
      "F1-score: 0.8308351177730193 \n",
      "Recall: 0.7886178861788617 \n",
      "Precision: 0.8778280542986425 \n",
      "Accuracy: 0.998616680383127\n"
     ]
    }
   ],
   "source": [
    "print('Test data --- \\nF1-score: {} \\nRecall: {} \\nPrecision: {} \\nAccuracy: {}'\n",
    "      .format(f1_score(df_test['Class'],test_pred),\n",
    "              recall_score(df_test['Class'],test_pred),\n",
    "              precision_score(df_test['Class'],test_pred),\n",
    "              accuracy_score(df_test['Class'],test_pred)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}