WhiteBox_Logistic_Regression.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Logistic regression"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Definition and intuition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The <a href=Linear_models.ipynb>linear regression</a> does not work well with classification cases, a solution for classification tasks is using logistic regression (a generalized linear model). using a link function (the logistic function) it allows the target to exist only between 0 and 1 making it possible to achieve binary classification and also giving a probability of a sample belonging to one class or another.    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Logistic Regression and interpretability "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As mentioned above, the logistic regression model outputs a probability of belonging to a certain  class (the class that has the label 1). Let's call this probability $Y$:  \n",
    "We have :\n",
    "$$ log(Y) = \\beta_0 + \\beta_1 X_1 + ... + \\beta_n X_n  $$\n",
    "In Linear models, in order to achieve explainability we plotted the coefficients $(\\beta_0, \\beta_1, ..., \\beta_n)$ but here we will plot $(exp(\\beta_0), exp(\\beta_1), ..., exp(\\beta_n))$ as they represent :\n",
    "\n",
    "* **Numerical feature**: For an increase of one unit in the corresponding feature, the odds are multiplied by $exp(\\beta)$ \n",
    "* **Binary categorical feature**: For a change of the feature from the reference level (0) to the level (1) the odds are multiplied exp(βj)\n",
    "* **Categorical feature with many levels**: to deal with multiple categories we can one-hot code them, returning to the case of binary features.\n",
    "> Odds are a way to describe the chances of an event occurring, it is the probability of success divided by the probability of failure"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn as sk\n",
    "import pandas as pd\n",
    "\n",
    "# Importing lending club dataset\n",
    "from mlinutils.datasets import Loans\n",
    "\n",
    "# Libraries specific to Logistic regression models \n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "\n",
    "import mlinutils.linear_model_plots as lmp\n",
    "import mlinutils.sanity as san"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "#  make plot outputs appear and be stored within the notebook\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## The datasets we're using "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data description\n",
    "The full description of the Datasets that are used in this entire collection of notebooks is available separately \n",
    "* Lending club: information on users and wether or not they default on their payments from a peer to peer lending online service <a href=\"lending_club_dataset.ipynb\">see lending club dataset exploration</a> "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data importation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "loans = Loans()\n",
    "ln_X_train, ln_X_test, ln_y_train, ln_y_test = loans.load()\n",
    "loan = loans.data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Data preview\n",
    "Take a peak into the data we'll be using !"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "This dataset contains 98339 instances with 19 attributes\n"
     ]
    },
    {
     "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>loan_amnt</th>\n",
       "      <th>int_rate</th>\n",
       "      <th>installment</th>\n",
       "      <th>grade</th>\n",
       "      <th>emp_length</th>\n",
       "      <th>home_ownership</th>\n",
       "      <th>annual_inc</th>\n",
       "      <th>verification_status</th>\n",
       "      <th>purpose</th>\n",
       "      <th>dti</th>\n",
       "      <th>delinq_2yrs</th>\n",
       "      <th>inq_last_6mths</th>\n",
       "      <th>open_acc</th>\n",
       "      <th>revol_bal</th>\n",
       "      <th>revol_util</th>\n",
       "      <th>open_acc_6m</th>\n",
       "      <th>inq_fi</th>\n",
       "      <th>mths_since_crl_was_opened</th>\n",
       "      <th>bad_loan_status</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>12994</th>\n",
       "      <td>17700.0</td>\n",
       "      <td>17.57</td>\n",
       "      <td>445.34</td>\n",
       "      <td>4.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>MORTGAGE</td>\n",
       "      <td>46000.0</td>\n",
       "      <td>Not Verified</td>\n",
       "      <td>debt_consolidation</td>\n",
       "      <td>28.39</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.0</td>\n",
       "      <td>19898.0</td>\n",
       "      <td>75.9</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>82.991437</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84331</th>\n",
       "      <td>14500.0</td>\n",
       "      <td>8.90</td>\n",
       "      <td>460.43</td>\n",
       "      <td>7.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>MORTGAGE</td>\n",
       "      <td>71000.0</td>\n",
       "      <td>Verified</td>\n",
       "      <td>debt_consolidation</td>\n",
       "      <td>1.12</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>2113.0</td>\n",
       "      <td>70.4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>192.923879</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5039</th>\n",
       "      <td>19000.0</td>\n",
       "      <td>15.80</td>\n",
       "      <td>460.03</td>\n",
       "      <td>5.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>MORTGAGE</td>\n",
       "      <td>49000.0</td>\n",
       "      <td>Verified</td>\n",
       "      <td>debt_consolidation</td>\n",
       "      <td>22.65</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>8374.0</td>\n",
       "      <td>43.8</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>141.998809</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40826</th>\n",
       "      <td>21000.0</td>\n",
       "      <td>9.99</td>\n",
       "      <td>677.52</td>\n",
       "      <td>6.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>MORTGAGE</td>\n",
       "      <td>175000.0</td>\n",
       "      <td>Not Verified</td>\n",
       "      <td>credit_card</td>\n",
       "      <td>9.14</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>47073.0</td>\n",
       "      <td>56.1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>145.021458</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26510</th>\n",
       "      <td>28000.0</td>\n",
       "      <td>7.90</td>\n",
       "      <td>876.13</td>\n",
       "      <td>7.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>OWN</td>\n",
       "      <td>125000.0</td>\n",
       "      <td>Source Verified</td>\n",
       "      <td>debt_consolidation</td>\n",
       "      <td>12.33</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>12860.0</td>\n",
       "      <td>31.9</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>276.999528</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       loan_amnt  int_rate  installment  grade  emp_length home_ownership  \\\n",
       "12994    17700.0     17.57       445.34    4.0         1.0       MORTGAGE   \n",
       "84331    14500.0      8.90       460.43    7.0        10.0       MORTGAGE   \n",
       "5039     19000.0     15.80       460.03    5.0        10.0       MORTGAGE   \n",
       "40826    21000.0      9.99       677.52    6.0        10.0       MORTGAGE   \n",
       "26510    28000.0      7.90       876.13    7.0        10.0            OWN   \n",
       "\n",
       "       annual_inc verification_status             purpose    dti  delinq_2yrs  \\\n",
       "12994     46000.0        Not Verified  debt_consolidation  28.39          0.0   \n",
       "84331     71000.0            Verified  debt_consolidation   1.12          0.0   \n",
       "5039      49000.0            Verified  debt_consolidation  22.65          2.0   \n",
       "40826    175000.0        Not Verified         credit_card   9.14          1.0   \n",
       "26510    125000.0     Source Verified  debt_consolidation  12.33          1.0   \n",
       "\n",
       "       inq_last_6mths  open_acc  revol_bal  revol_util  open_acc_6m  inq_fi  \\\n",
       "12994             0.0      19.0    19898.0        75.9          1.0     1.0   \n",
       "84331             0.0       4.0     2113.0        70.4          1.0     1.0   \n",
       "5039              1.0      15.0     8374.0        43.8          1.0     1.0   \n",
       "40826             0.0      16.0    47073.0        56.1          1.0     1.0   \n",
       "26510             0.0      11.0    12860.0        31.9          1.0     1.0   \n",
       "\n",
       "       mths_since_crl_was_opened  bad_loan_status  \n",
       "12994                  82.991437                1  \n",
       "84331                 192.923879                0  \n",
       "5039                  141.998809                1  \n",
       "40826                 145.021458                1  \n",
       "26510                 276.999528                0  "
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loans.preview()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Application"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plan\n",
    "For logistic regression models we will only see its interpretability for the task: classification with lending club dataset (for regression with bike sharing <a href=Linear_Models.ipynb>see the linear models notebook</a>). We can interpret the our results (globally and locally). We will add LASSO and Ridge penalization and compare their effect on interpretability.  \n",
    "* <a href=#lr>Classification with lending club dataset</a>  \n",
    "    * <a href=#lrnone> Logistic regression without penalization </a>\n",
    "        * <a href=#lrnonem> Model creation and sanity checks </a>\n",
    "        * <a href=#lrnoneg> Global explanation </a>\n",
    "        * <a href=#lrnonel> Local explanation </a>\n",
    "    * <a href=#lrc>Conclusions </a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Classification with lending club dataset <a name=lr>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Logistic regression without penalization <a name=lrnone>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Model creation and sanity checks <a name=lrnonem>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Model creation\n",
    "scaler = MinMaxScaler()\n",
    "scaler.fit(ln_X_train)\n",
    "\n",
    "scaled_ln_X_train = scaler.transform(ln_X_train)\n",
    "scaled_ln_X_test = scaler.transform(ln_X_test)\n",
    "\n",
    "lr = LogisticRegression(C=0.05).fit(scaled_ln_X_train, ln_y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Sanity checks\n",
    "san.classification_sanity_check(lr, \n",
    "                                scaled_ln_X_train, scaled_ln_X_test,\n",
    "                                ln_y_train, ln_y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Global explanation <a name=lrnoneg>\n",
    "The following figure explains what happens in the model globally, the global trends and the importance of each and every feature.\n",
    "Logistic regression coefficients are reported as log odds. \n",
    "    \n",
    "It's a representation of the the exponential of the coefficients \n",
    "* If $\\beta \\geq 0$: the value <span style=\"color:#3572C6\">augments</span> the probability of belonging to the \"positive\" class and it is represented as \"the target outcome (e.g., a correct response) is $exp(\\beta)$ times <b>more likely</b> than the non-target outcome (e.g., an incorrect response).\"\n",
    "* If $\\beta < 0$ the value <span style=\"color:##C44E52\">diminishes</span> the probability of belonging to the default class and it is represented as \"the target outcome (e.g., a correct response) is $exp(-\\beta)$ times <b>less likely</b> than the non-target outcome (e.g., an incorrect response).\"\n",
    "    \n",
    "A more detailed guidance on how to interpret logistic regression coefficients can be find at [Interpreting logistic regression models (pdf)](http://www-hsc.usc.edu/~eckel/biostat2/notes/notes14.pdf)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lmp.global_explanation_plot(ln_X_train.columns, lr, selected_ft=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Local explanation <a name=lrnonel>\n",
    "After looking at global explanations, we need to understand why for a particular sample the prediction was made.\n",
    "For a single sample we explain the numerical contribution of each feature to the probability. (in a mutiplication sense\n",
    "we evaluate and display the contribution of each feature to the predicted total."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>loan_amnt</th>\n",
       "      <th>int_rate</th>\n",
       "      <th>installment</th>\n",
       "      <th>grade</th>\n",
       "      <th>emp_length</th>\n",
       "      <th>home_ownership</th>\n",
       "      <th>annual_inc</th>\n",
       "      <th>verification_status</th>\n",
       "      <th>purpose</th>\n",
       "      <th>dti</th>\n",
       "      <th>delinq_2yrs</th>\n",
       "      <th>inq_last_6mths</th>\n",
       "      <th>open_acc</th>\n",
       "      <th>revol_bal</th>\n",
       "      <th>revol_util</th>\n",
       "      <th>open_acc_6m</th>\n",
       "      <th>inq_fi</th>\n",
       "      <th>mths_since_crl_was_opened</th>\n",
       "      <th>bad_loan_status</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>80149</th>\n",
       "      <td>8000.0</td>\n",
       "      <td>14.64</td>\n",
       "      <td>275.92</td>\n",
       "      <td>5.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>MORTGAGE</td>\n",
       "      <td>96000.0</td>\n",
       "      <td>Verified</td>\n",
       "      <td>home_improvement</td>\n",
       "      <td>16.89</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>11.0</td>\n",
       "      <td>10899.0</td>\n",
       "      <td>31.6</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>244.966016</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       loan_amnt  int_rate  installment  grade  emp_length home_ownership  \\\n",
       "80149     8000.0     14.64       275.92    5.0         6.0       MORTGAGE   \n",
       "\n",
       "       annual_inc verification_status           purpose    dti  delinq_2yrs  \\\n",
       "80149     96000.0            Verified  home_improvement  16.89          0.0   \n",
       "\n",
       "       inq_last_6mths  open_acc  revol_bal  revol_util  open_acc_6m  inq_fi  \\\n",
       "80149             4.0      11.0    10899.0        31.6          1.0     1.0   \n",
       "\n",
       "       mths_since_crl_was_opened  bad_loan_status  \n",
       "80149                 244.966016                0  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loan_sample = scaled_ln_X_test[42]\n",
    "\n",
    "## recovers the original format just for reading convenience\n",
    "loans.data.iloc[[ln_X_test.iloc[42].name],:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Probability of \"default\" is: 0.19\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2cAAAJVCAYAAABajo13AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAIABJREFUeJzs3Xvcl/Phx/HXhxxyWAxtEmozH+l0d7BKhBlmc5rym2QkOR+Grc3mMMw2c5yc8sNEmvwcKxITychCbofwYSkrxpyKhE7X74/ruu99u7u7uzvo/lav5+PRo/t7XZ/TdX3v+L7vz+f63CHLMiRJkiRJDWuthh6AJEmSJMlwJkmSJEllwXAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJWiXEGLeJMc6KMa5dvB4bY+y/Att/MMZ45Ipq76sUY8xijNutoLZ+HGOcVtzbDjHGSTHG3VdAu31jjH9fivJTY4zfX84+B8UYz6lHuRVyjZK0ojVq6AFI0uokxjgV+AYwv+Tw9imld5ajzd2B21JKzZdrcGWsuG/9U0qPLK5MSulfwEYrqL/zgO1SSoeXtL/vimi7lr4GA9NTSmcvY/2x5O//jStyXCUuBU5OKQ0vXrf+ivr5yqWUjq9nuVX2GiWt3gxnkrTi7V9XyFjZYoyNUkrzGnocy2N1uIYyti0wqaEHIUkynEnSShNj7ApcDuwIvAX8LKU0tjh3FPBLoDnwPvCnlNL1McYNgQeB9WKMs4qmtgf+QMlsTM3ZtWIm6jqgT/4ybgg0Ba4CegCzgCtSSgOL8t8Fri3a/hwYmlI6YzHXcSBwPvCtYqwnpZRGxxibAYOAXYCPimu4oahzXnHdXwA/Bv4FHJlSejbGOATYBhgZY5wPXAD8HzAF6A/8FpgaYzyiOLZOSVD7doxxAhCBscBRKaWPapttrJqdI/9/32+AEGM8CJicUmpfOkMVY1yrKHMM0BgYDZySUpoZY2xRjKMv8Dtgg+Je/r6We3Vs8R5kMcbTgMdSSvvHGFsV708F8Dbw65TSiFrq/x7YFegaY/wzMDildHJx+vsxxgeBzYG/ks9+ZUW9fsAA4JvABODYlNJbNdpeD/gQWBt4Icb4bkrp26WzmHW9b0UbZxb3qCkwDTgrpXRvzeuoTYzxp8CF5LOhl9c4txb5v4djgE2AMcDxKaWPivO7ABcXY/sUOCelNLh0ljLGuDkwmPz7cQF5AN0tpbSgxjWuB/wJ+J+i+/8DfpVS+rLq+wi4AvgV+Yz4b1JKN9fnGiVpafnMmSStBDHGrYAHyD+Mfh34BXB3jHGLosh/gP2ArwFHAVfEGDumlD4D9gXeSSltVPyp7xLJ3sCPyD/cLgBGAi8AWwF7AqfFGPcpyl4JXJlS+hrwbfIPqLVdx3eBW8k/+G9CHvSmFqdvB6YDzYBewB9ijHuWVD8AGFbUGwFcDZBS+in5h/79i+u7uKTObkArYB9qdwTQr+hzHjCwrhtS9DeaPNzeUfTXvpZifYs/e5CH0I2qxltiF/JQuCdwbhG4avb1v8BQ4OKir/1jjOuQvxcPk4eaU4ChMcZYS/2zgCfIg9dGJcEM8u+XnYD25MFiH4AicP4GOBjYoqh/ey1tf5lSqlom2j6l9O1a7gMs5n0rTCYPj03IA/ttMcYtF9NOtRjjjuTh9Kfk791m5D+YqHIqcBD5+98M+Bi4pqi7DfkPLK4qrq8CqKylm5+Tfz9uQb7U+DdAVku5s4CuRTvtge8CpUtQv1lc31bA0cA1McZNl3SNkrQsnDmTpBXvvhhj1czO2JTSQcDhwKiU0qji+N9ijM8CPwRuSSk9UFL/8Rjjw+QfeicuxzgGppSmAcQYuwBbpJQuKM69GWO8ATgUeAiYC2wXY9w8pfQB8PRi2jwa+EtK6W/F67eL9rcmDyv7pZS+ACpjjDeSf/geU5T9e9X1F7Nlp9XjGs4rAiq1ZBeAISmll4vz5xT9rohNPfoAl6eU3iza/jXwcjHDWeX8lNLn5LNOL5B/sH+1Hm13JQ97F6WUFgCPxhjvJw/T5y3FGC9KKc0AZsQYHyMPF6OB44A/ppReLcb+B+A3McZta86e1dNi37eU0p0l5e4o7tN3geHUrRdwf0ppXNHuOUBp8DyOPJBOL86fB/yrmG3rAzySUqoKnB8Wf2qaC2wJbJtS+id5SK1NH/JZ0f8UfZ0PXA9UbSwyF7igmK0dVcxgRxb/b0SSlpnhTJJWvINqeeZsW+CQGOP+JcfWAR4DiDHuS758b3vyVQ0bAC8t5zim1ei/WYxxRsmxtfnvB9ajyZcTvhZjnEIePO6vpc2tgVG1HG8GfJRS+rTk2FtA55LX75Z8PRtYvx7Pkk2r41zN82+R39PNl1CnPpoV7ZW23Yh8BqZKzeup72YlzYBpRTArbX+rpRzj4vrfFrgyxnhZyflQtL8s4Wyx71ux1PQMoEVxfiPqd/+bUfLepZQ+izGWBqxtgXtjjKX3aD75/d+afMZuSS4hD7sPF8H+f1NKFy1mLDXf62Ylrz+s8T26NO+1JC0Vw5kkrRzTyGd5jql5onjm5W7yJXrDU0pzY4z3kX+ghtqXYn1GHuCqfLOWMqX1pgFTUkrfqW1wKaU3gN7Fsz4HA3fFGDermrWq0U5ty9/eAb4eY9y4JKBtQzGzVg+1XWNdx6tsXfL1NuSzHB9Q4/7EfPv9LUrKLqndd8gDQmnb84D3WHj5XX3U7OsdYOsY41olAW0b4PV61l+SacDvU0pDl7LeUokxbgvcQL6sc3xKaX6MsZL/ft/W5d/ky1Wr2tqAfGljlWlAv5TSk7X0O418dq5Oxffhz4GfxxhbA4/FGJ9JKY2pUbTqva7aFGWb4pgkrXSGM0laOW4Dnime8XqEfIanK/BPYCawHvnmGvOKWbS9gZeLuu8Bm8UYm6SUZhbHKsk/dF4IrMuSlwhOAD6JMf6K/LmsOeQfjhunlJ6JMR4OPJRSer9kdm1+Le3cRD4TcT/5rN+WwMYppddijE8Bf4wx/oJ8BvBo8uWc9fEe+bNdS+vwGOOt5M+9XQDcVYSE18lneH5E/mzXb8jvcWl/e9UISKVuB35VbLjxPv99Rm3eYpZX1qXmtf2DPDz+spjd6g7sT/78WH3qL8kg4HcxxsqU0qQYYxNg7xpLEFeEDcmD4/tQvalNm3rWvQv4R7GxxwTy9670OfhBwO9jjEemlN4qns3cudjufyj5Ms3/Ae4hfx5s65TSQs+dxRj3A14jn2X7hPz7ubbv6duBs2OMzxTXcy75v1dJWuncEESSVoLi2a8DyUPC++QzAwOAtYqf8J9KvgnHx8Bh5BsvVNV9jfwD5JsxxhnFrohDyDf3mEoePu5YQv/zyQNABflOgx8AN5J/sAX4ATCpeJ7mSuDQ4tmxmu1MoNiwhDxUPs5/Z5h6ky9vewe4F/htybNpS/JH8g/IM4pwV19DyHfkexdYn/w+UoTYE4trfJs8DE0vqVcVVD6MMdb2XN9firbHkd+vL8g37lgWNwE7Ftd2X0ppDvkmG/uSvw/XAkcU73NtrgR6xRg/jjHWZ8OTe8l3HxwWY/yEPOSv8N/hllJ6BbgMGE8eINsCi8x0LabuJOAk8l0m/03+fV/6/lxJ/m/g4Rjjp+TPd3Up6v6L/FnNn5PvClpJ/rxfTd8h/0HIrGKM11btjlrDhcCzwIvkS4knFsckaaULWba0qyUkSZIkSSuaM2eSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAbcSl9rhOL3SO1EvitYbVspS5IklYu1yX9VyTMppS8bejBaeQxnWlPsBDzR0IOQJElaCrsCf2/oQWjlMZxpTfFvgKFDh/LNb36zocciSZK0WO+++y59+vSB4vOL1hyGM60p5gN885vfpHnz5g09FkmSpPrwUYw1jBuCSJIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJK9iCOXMaegiS1jD+d0eSVg+NGnoA0upmrXXX5ckDezb0MCStQboPv7uhhyBJWgGcOZMkSZKkMmA4kyRJkqQyYDiTJEmSpDJgOJMkSZKkMmA4kyRJkqQyYDiTJEmSpDJgOJMkSZKkMmA4kyRJkqQyYDiTJEmSpDJgOJMkSZKkMmA4kyRpNTVt2jT22GMPWrVqRevWrbnyyitrLTd06FDatWtHu3bt2HnnnXnhhReqz11xxRW0bt2aNm3a0Lt3b7744ot697/77rvz7LPPLvd11DR16lTatGkDwLPPPsupp54KwHnnncell15a73aWtvzgwYN55513lm6wJUrHXZcBAwbQunVrBgwYsEz9bLTRRtX9/fWvf12mNiQ1DMOZJEmrqUaNGnHZZZfx6quv8vTTT3PNNdfwyiuvLFKuZcuWPP7447z44oucc845HHvssQC8/fbbDBw4kGeffZaXX36Z+fPnM2zYsJV9GXXq3LkzAwcOXCl9LW84q6/rr7+eiRMncskllyxXO4YzadVjOJMkaTW15ZZb0rFjRwA23nhjWrVqxdtvv71IuZ133plNN90UgK5duzJ9+vTqc/PmzePzzz9n3rx5zJ49m2bNmjFz5kxijKSUAOjduzc33HBDnWN5+OGH6datGx07duSQQw5h1qxZAJx55pnsuOOOtGvXjl/84hcA3HnnnbRp04b27dvTo0ePOtsdO3Ys++233yLHb7jhBvbdd18+//xzJk+ezA9+8AM6derErrvuymuvvbZQ2cmTJ1ffJ4A33niDTp06LVTmrrvu4tlnn6VPnz5UVFTw+eefM2bMGDp06EDbtm3p168fX3755SLjeO6552jfvj3dunXjmmuuqT4+f/58BgwYwE477US7du24/vrrATjggAP47LPP6NKlC3fccQcjR46kS5cudOjQge9///u89957wKKzfm3atGHq1KkL9X3mmWfyxBNPUFFRwRVXXFHnfZRUHgxnkiStAaZOncrzzz9Ply5d6ix30003se+++wKw1VZb8Ytf/IJtttmGLbfckiZNmrD33nvTpEkTrr76avr27cuwYcP4+OOPOeaYYxbb5gcffMCFF17II488wsSJE+ncuTOXX345H330Effeey+TJk3ixRdf5Oyzzwbgggsu4KGHHuKFF15gxIgRS32tV199NSNHjuS+++6jcePGHHvssVx11VU899xzXHrppZx44okLlf/2t79NkyZNqKysBODmm2+mb9++C5Xp1asXnTt3ZujQoVRWVhJCoG/fvtxxxx289NJLzJs3j+uuu26RsRx11FEMHDiQ8ePHL3KfmzRpwjPPPMMzzzzDDTfcwJQpUxgxYgSNGzemsrKSn/zkJ+yyyy48/fTTPP/88xx66KFcfPHF9b4PF110EbvuuiuVlZWcfvrp9a4nqeEYzlR2YoyDY4wnN/Q4JGl1MWvWLHr27Mmf//xnvva1ry223GOPPcZNN93En/70JwA+/vhjhg8fzpQpU3jnnXf47LPPuO222wDYa6+9aNu2LSeddBI33nhjnf0//fTTvPLKK3Tv3p2KigpuueUW3nrrLb72ta+x/vrr079/f+655x422GADALp3707fvn254YYbmD9//lJd65AhQ3jwwQe5++67WW+99Zg1axZPPfUUhxxyCBUVFRx33HH8+9//XqRe//79ufnmm5k/fz533HEHhx12WJ39pJRo2bIl22+/PQBHHnkk48aNW6jMzJkzmTFjBrvtthsAP/3pT6vPPfzww9x6661UVFTQpUsXPvzwQ954441F+pk+fTr77LMPbdu25ZJLLmHSpElLdT8krVoMZ/rKxRgbNfQYJGlNNXfuXHr27EmfPn04+OCDF1vuxRdfpH///gwfPpzNNtsMgEceeYSWLVuyxRZbsM4663DwwQfz1FNPAbBgwQJeffVVGjduzEcffVTnGLIsY6+99qKyspLKykpeeeUVbrrpJho1asSECRPo2bMn9913Hz/4wQ8AGDRoEBdeeCHTpk2joqKCDz/8sN7XW7W8r2pp5oIFC9hkk02q+66srOTVV19dpF7Pnj158MEHuf/+++nUqVP1PajrmpYkyzJCCIs9d9VVV1WPacqUKey9996LlDvllFM4+eSTeemll7j++uurN2Rp1KgRCxYsqC63NBu1SCpffmjWcokx9gR+D3wO3Fl8vTHwKfBL4EfAEzHG/wOuBTYE1gf+N6X056KNrYBbgc2BKZR8X8YYvwZcDrQr6j0GnJFSWrofpUrSGijLMo4++mhatWrFGWecsdhy//rXvzj44IMZMmRI9UwQwDbbbMPTTz/N7Nmzady4MWPGjKFz585Avotjq1at+MMf/kC/fv0YP34866yzTq3td+3alZNOOol//vOfbLfddsyePZvp06fTrFkzZs+ezQ9/+EO6du3KdtttB+TPgHXp0oUuXbowcuRIpk2btsSwVKVDhw6ccMIJHHDAATz00EM0a9aMli1bcuedd3LIIYeQZRkvvvgi7du3X6je+uuvzz777MMJJ5zATTfdVGvbG2+8MZ9++ikAO+ywA1OnTq2+piFDhlTPkFXZZJNNaNKkCX//+9/ZZZddGDp0aPW5ffbZh+uuu47vfe97rLPOOrz++utstdVWbLjhhgu1MXPmTLbaaisAbrnllurjLVq04P777wdg4sSJTJkypc7xSlo1OHOmZRZjbAr8L7B/SqkDeUArtVZKafeU0jnAVOD7KaWOwHeBY2OMrYpyA4FxKaX2wBlA6f/dLgceTyl9F6gAmgL9vqprkqTVyZNPPsmQIUN49NFHqaiooKKiglGjRgH57NSgQYOA/BmvDz/8kBNPPJGKiorqANalSxd69epFx44dadu2LQsWLODYY4/l9ddf58Ybb+Syyy5j1113pUePHlx44YWLHccWW2zB4MGD6d27N+3ataNr16689tprfPrpp+y33360a9eO3XbbrXrTigEDBtC2bVvatGlDjx49FglSS7LLLrtw6aWX8qMf/YgPPviAoUOHctNNN9G+fXtat27N8OHDa63Xp08fQgi1zmAB9O3bl+OPP56KigqyLOPmm2/mkEMOoW3btqy11locf/zxi9S5+eabOemkk+jWrRuNGzeuPt6/f3923HFHOnbsSJs2bTjuuOOYN2/eIvXPO+88DjnkEHbddVc233zz6uM9e/bko48+oqKiguuuu26hUF2lXbt2NGrUiPbt27shiLSKCPWZlpdqE2M8ADglpbRX8boJMIP/zpxtmVJ6tzj3DeAyoD2wANgWOD6lNCzG+BHQNqX0dlH2PuCRlNLVMcb/AO8WdQA2AIanlJbql7/EGFsAU8aMGUPz5s2X57Lr5ckDe37lfUhSle7D727oIawWLr30UmbOnMnvfve7hh6K1nDTp09nzz33BGiZUprawMPRSuSyRi2PANSV7meVfP0H8pDVN6U0L8b4MPkyxfr0cVBK6c1lH6YkSXX78Y9/zOTJk3n00UcbeiiS1mAua9TyeBroFGPcrnjdt46ymwDTimDWBti15NyjwFEAMcaWwJ4l50YAZ8YY1y7Ob16UkSRphbn33nt58cUXF1o6KEkrm+FMyyyl9B5wPPBAjPFJoDEwF5hdS/ELgWNijM8AZwOl+w3/DNgjxvgC8CfgbyXnTgPmAy/EGF8CRgNbrehrkSRJkhqayxq1vEanlO4EiDEeBUxIKS0gX45YLaX0PNCmtgaKZ832XMy5T4ETVuiIJUmSpDJkONPyOjXGeAj599JHwDENPB5JkiRplWQ403JJKf2e/HebSZIkSVoOPnMmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWWgUUMPQFrdLJgzh+7D727oYUhagyyYM4e11l23oYchSVpOzpxJK5gfkCStbP53R5JWD4YzSZIkSSoDhjNJkiRJKgOGM0mSJEkqA4YzSZIkSSoDhjNJkiRJKgOGM0mSJEkqA4YzSZIkSSoDhjNJkiRJKgOGM0mSVnFz5s5v6CFIaxT/zemr0qihByBJkpbPuuuszf4/H97Qw5DWGCMvO7Chh6DVlDNnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSdIK0q9fP5o2bUqbNm0WW+aSSy6hoqKCiooK2rRpw9prr81HH30EwIwZMzjuuOOYMmUKr7/++iMhhG4ra+zLIoQwOITQq/j6xhDCjsXXs5aynXqXDyG0CCEctnQjXaSN6nHXUWaHEEJlCOH5EMK3l6GP80IIvyi+7htCaLakOoYzSZIkaQXp27cvo0ePrrPMgAEDqKyspLKykj/+8Y/stttufP3rXwfgZz/7GbvvvjstW7Zk22233Rd4dSUMeyEht9Q5Icuy/lmWvfJVjKmGFsByhbN6OggYnmVZhyzLJi9nW30Bw5kkSZK0svTo0aM6aNXH7bffTu/evQH45JNPGDduHIceeigA66233twsy2aEEBqFEJ4JIewOEEL4Ywjh9zXbCiGMDSFcEUIYF0J4NYSwUwjhnhDCGyGEC0vKnRFCeLn4c1pxrEVR51pgIrB1CGHvEML4EMLEEMKdIYSN6rqWov/ONY5tXrTxo+L1gOJaXgwhnF9LG0NCCAeWvB4aQjigRrGLgF2LWa3TQwjrhxBuDiG8VMxy7VFLuyGEcHUI4ZUQwgNA05JznUIIj4cQngshPBRC2DKE8EPgNKB/COGxotx9RZlJIYRjS+rPKvm6VwhhcI2+ewGdgaHFmBsv7h4aziRJkqQGMHv2bEaPHk3Pnj0BePPNN9liiy0444wzeOutt/jnP/95UQhhwyzL5pHPvFwXQtgL+AGwSLApzMmyrAcwCBgOnAS0AfqGEDYLIXQCjgK6AF2BY0IIHYq6Ebg1y7IOwGfA2cD3syzrCDwLnLE01xdC+AbwAHBulmUPhBD2Br4DfBeoADqFEHrUqHZjMT5CCE2AnYFRNcqcCTyRZVlFlmVXFNdIlmVtgd7ALSGE9WvU+XFxfW2BY4p2CSGsA1wF9MqyrBPwF+D3WZaNKu7hFVmWVYW9fkWZzsCpIYTN6nMfsiy7i/z+9SnG/PniyhrOtIgYY2WMcbGJvqTcaTHGpksqt5R9HxRj/O6KbFOSJKkcjRw5ku7du1fPtM2bN4+JEydyxBFHsO222xJCmE0eRMiybBIwBBhJHhLmLKbZEcXfLwGTsiz7d5ZlXwJvAlsDuwD3Zln2WZZls4B7gF2LOm9lWfZ08XVXYEfgyRBCJXAksO1SXN46wBjgl1mW/a04tnfx53ny2bkdyMNatSzLHge2CyE0JQ9adxfhtC67kN8bsix7DXgL2L5GmR7A7VmWzc+y7B3g0eJ4JA+vfyuu82yg+WL6OTWE8ALwNPm9/M5iyi2zRiu6Qa36UkoV9Sx6GvAI8J/6th1jbJRSqusf2EHkP1mYUN82JUmSVkXDhg2rXtII0Lx5c5o3b06HDvlE1nrrrffg7Nmz+5ZUaQvMAL5RR7NfFn8vKPm66nUjINRR97OSrwPwtyzLei+u8BLMA54D9gEeL2nzj1mWXb+EukOAPsChQL969FXXNZXKFlN3UpZldW68Uiwp/T7QLcuy2SGEsUDV7FxpuzVn7JaKM2daRIwxizFuVHw9NcZ4QYxxfPH1ycXxs8gfaryrmGnbcQntDYgxjgV+G2NsG2N8IsY4Mcb4SozxtKLcPsABwJlFm0cUx4+MMf4jxvhcjPHRGGP8au+AJEnSV2vmzJk8/vjjHHhg9eNVfPOb32Trrbdm8uR874k5c+Z0B14BCCEcDGxGPgM0MISwyTJ2PQ44KISwQQhhQ/Llfk/UUu5poHsIYbui/w1CCDVno+qSkQerHUIIZxbHHgL6VT27FkLYqpghq2kw+SRA1YxhTZ8CG9e4pj5Fm9sD2wCpRp1xwKEhhLVDCFsCVUsVE7BF1a6YIYR1Qgita+mzCfBxEcx2IJ9ZrPJeCKFVsYnKj2upW9uYa2U4U31skFLqBuwOXBRj3Cil9HvgHaBXSqkipbSknXnWSintnlI6B5gKfD+l1JF8zfGxMcZWKaWHyKfiLyravDXGuCvwP0CPlFIn4BLytcCSJEllp3fv3nTr1o2UEs2bN+emm24CYNCgQQwaNKi63L333svee+/NhhtuuFD9q666ilNOOYWpU6eyYMGCHYE/hBA2J98E4+gsy14HrgauXJbxZVk2kTz8TAD+AdyYZdnztZR7n/w5t9tDCC+Sh7UdlrKv+eSzX3uEEE7Msuxh4K/A+BDCS8Bd1BJYsix7j3yXypsX0/SLwLwQwgshhNOBa4G1izbvAPoWSzlL3Qu8Qb7c8zqK2bxieWgv4E/FksVKiufRahgNNCruxe/I70eVM4H7yZdK/nsxYx4MDFrShiAhy2qb3dOaLMaYARunlGbFGKeSB7Bni3PTgL1SSq8V5/ZLKb1cj/a2TCm9W7z+BnAZ0J58in1b4PiU0rAY42Dg2ZTS1UXZi8m3Sv2gaC4Am6aUtlnKa2oBTBkzZgzNmy9uGbEkrbr2//nwhh6CtMYYedmBSy60HKZPn86ee+4J0DKlNPUr7awMhRA2IA9RHbMsm9nQ41mZfOZM9fFFydfzWbbvm9JfLPgH4F2gb0ppXozxYRa/PjcAf0kpnbsMfUqSJGkVEkL4PvkqqcvXtGAGLmvU8vmEfP3t0toEmFYEszb8d4eg2tocCRwRY2wOEGNcO8bYaVkHLEmSpPKVZdkjWZZtk2XZnxt6LA3BcKblMRC4eUkbgtTiQuCYGOMz5NuVjis5NwStDcWbAAAgAElEQVQ4rGpDkJTSOOAsYESM8QXgZeCrXUsgSZIkNQCfOdMawWfOJK3ufOZMWnl85kxfFWfOJEmSJKkMuCGIVogY47nAwbWc2julVO9fUi1JkiStqQxnWiFSShcAFzT0OCRJkqRVlcsaJUmSJKkMGM4kSZIkqQwYziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkiSpDDRq6AFIkqTlM2fufEZedmBDD0NaY8yZO59111m7oYeh1ZAzZ5IkreL8kCitXP6b01fFcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSdIqbs7c+Q09BGmV4r8ZlatGDT0ASZK0fNZdZ232//nwhh6GtMoYedmBDT0EqVbOnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSSX69etH06ZNadOmTZ3lxo4dS0VFBa1bt2a33XZb6Nz8+fPp0KED++2331c51BUihDA2hNC5+HpUCGGTEEKLEMLLS9HG0pavCCH8cFnGW9JG9bjrKLNrCGFSCKEyhNB4GfoYHELoVXx9Wghhg2Udb30YziRJkqQSffv2ZfTo0XWWmTFjBieeeCIjRoxg0qRJ3HnnnQudv/LKK2nVqtVXOcw6hRDWXpZ6WZb9MMuyGSt6PLWoAJYrnNVTH+DSLMsqsiz7fDnbOg0wnEmSJEkrS48ePfj6179eZ5m//vWvHHzwwWyzzTYANG3atPrc9OnTeeCBB+jfv3/1sZkzZxJjJKUEQO/evbnhhhsWabdFixZcdNFF/Otf/+KNN94YEULoGEJ4KIQwOYRwPEDIXRJCeDmE8FII4SfF8d1DCI+FEP4KvFQcOzyEMKGYObp+SaEthDA1hLB5jWPfCiE8H0LYKYSwdtH3MyGEF0MIx9XSxhMhhIqS10+GENqVvF4XuAD4STGun4QQvh5CuK9o8+nS8iX1GocQhhVl7gAal5zbO4QwPoQwMYRwZwhhoxBCf+B/gHNDCEOLY2OKMi+FEA4s6i406xdC+EUI4bwafZ8KNAMeCyE8Vtc9XB6GM0mSJGkpvf7663z88cfsvvvudOrUiVtvvbX63GmnncbFF1/MWmv996N2kyZNuPrqq+nbty/Dhg3j448/5phjjqm17WbNmrHNNtuw1lprPQMMBnoBXckDDcDB5DNP7YHvA5eEELYszn0XOCvLsh1DCK2AnwDdsyyrAOaTzyTVWwghAncDR2VZ9gxwNDAzy7KdgJ2AY0IILWtUuxHoW9TfHlgvy7IXq05mWTYHOBe4o5jRugM4H3g+y7J2wG+AW1nUCcDsoszvgU5FH5sDZwPfz7KsI/AscEaWZTcCI4ABWZb1Ab4AflyU2QO4LIQQ6nMfsiwbCLwD7JFl2R71qbMsGn1VDUvLI8aYARsDvwD+kFKaUxy/AJiUUrqjIccnSZLWbPPmzeO5555jzJgxfP7553Tr1o2uXbvy+uuv07RpUzp16sTYsWMXqrPXXntx5513ctJJJ/HCCy8stu299tqLW265hbXXXvu1+fPnL8iy7FPg0xDCFyGETYBdgNuzLJsPvBdCeJw8KH0CTMiybErR1J7kAeaZIoM0Bv6zFJe5BTAc6Jll2aTi2N5Au6rnsIAmwHeA10vq3QmcE0IYAPQjD5hLsgvQEyDLskdDCJuFEJpkWTazpEwPYGBR5sUQQlXg6wrsCDxZXOe6wPha+gjAH0IIPYAFwFbAN+oxtpXGcKZy91vgUmAOQErp3IYdjiRJEjRv3pzNN9+cDTfckA033JAePXrwwgsvMHHiREaMGMGoUaP44osv+OSTTzj88MO57bbbWLBgAa+++iqNGzfmo48+onnz5rW2vd5661V9mQFflpxaQP75va7Zns9Kvg7ALVmW/XoZL3MmMA3oDlSFswCckmXZQ6UFQwgtqgedZbNDCH8DDiRfVljnph0l7daU1fNYAP6WZVnvJfTRhzxwdsqybG4IYSqwPjCPhVcUrr/k4X41XNaoshBjPDjG+FqM8akY49nF4SuLv5+KMVbGGDeJMQ6OMZ7cUOOUJEkCOPDAA3niiSeYN28es2fP5h//+AetWrXij3/8I9OnT2fq1KkMGzaM733ve9x2220AXHHFFbRq1Yrbb7+dfv36MXfu3GXtfhz581prhxC2IJ9RmlBLuTFArxBCU4Diua5tl6KfOcBBwBEhhMOKYw8BJ4QQ1ina3D6EsGEtdW8kn+V6Jsuyj2o5/yn5KqnSa+pTtLk78EGWZZ/UqFNapg1Q9Vza00D3EMJ2xbkNiuWUNTUB/lMEsz2AqnvxHtC0mK1bD1jcFps1x7zCGc7U4GKMTYEbgANTSjtTzJIBPyv+3jmlVJFSWhk7B0mSpDVc79696datGyklmjdvzk033QTAoEGDGDRoEACtWrXiBz/4Ae3ateO73/0u/fv3r3Pr/ddff50bb7yRyy67jF133ZUePXpw4YUXLusQ7wVeBF4AHgV+mWXZuzULZVn2CvmzWA8XSwD/BmxZs1xdsiz7jDysnF5soHEj8AowsdhE43pqWY2XZdlz5Mssb15M048BO1ZtCAKcB3QuxnkRcGQtda4DNirK/JIikGZZ9j75M263F+eeBnaopf7Qoo9nyUPea0X9ueTP8/0DuL/qeC3+F3jwq9wQJGRZbTOD0soTYzwAOCWltFfxehPgY/KfTHwKbJxSmlWcGww8m1K6ein7aAFMGTNmzGKXEEjSqmz/nw9v6CFIq4yRlx3Y0EOo0/Tp09lzzz0BWqaUpjbwcJZJCKEZMBbYIcuyBQ08nFWGM2cqB/XaJUeSJEnlL4RwBPks1FkGs6VjOFM5GA90iDF+p3jdv+Tcp+TrgyVJkrQKyLLs1izLts6y7M4ll1Ypw5kaXErpP8CxwMgY41PkO+ZUuQx4tGpDkAYZoCRJkrQSuJW+ykJK6R7gnpJDfy7+Pr/4U6XvyhqTJEmStDI5cyZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWXAcCZJkiRJZcBwJkmSJEllwHAmSZIkSWWgUUMPQJIkLZ85c+cz8rIDG3oY0ipjztz5rLvO2g09DGkRzpxJkrSK80OmtHT8N6NyZTiTJEmSpDJgOJMkSZKkMmA4kyRJkqQyYDiTJEmSpDJgOJMkSZKkMmA4kyRJkqQyYDiTJEmSpDJgOJMkSZKkMmA4k7RaWTBnTkMPQVrp/L6XpNVDo4YegCStSGutuy5PHtizoYchrVTdh9/d0EOQJK0AzpxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmSJElSGTCcSZIkSVIZMJxJkiRJUhkwnEmStBrr168fTZs2pU2bNostM3bsWJo0aUJFRQUVFRVccMEFS1V/cfr27ctdd921TONeko022giAd955h169egEwePBgTj755Hq3sbTl77vvPl555ZWlG2gNVeOuy8CBA2nVqhV9+vRZpj5atGjBBx98wIwZM7j22muXqQ1JDcNwJknSaqxv376MHj16ieV23XVXKisrqays5Nxzz13q+g2lWbNmX1kArGlFhLP6uPbaaxk1ahRDhw5drnYMZ9Kqx3AmSdJqrEePHnz9619fofXnzZvHTjvtxNixYwH49a9/zVlnnVVnO8899xy77bYbnTp1Yp999uHf//43kM8S7bjjjrRr145DDz0UgMcff7x6Fq9Dhw58+umni2136tSptc7qPfDAA3Tr1o0PPviA999/n549e7LTTjux00478eSTTy5U9tNPP6Vly5bMnTsXgE8++YQWLVpUvwZ46qmnGDFiBAMGDKCiooLJkydTWVlJ165dadeuHT/+8Y/5+OOPFxnHlClT6NatGzvttBPnnHPOQucuueQSdtppJ9q1a8dvf/tbAI4//njefPNNDjjgAK644gomTJjAzjvvTIcOHdh5551JKQGLzvrtt99+1e9HlTPPPJPJkydTUVHBgAEDFnsPJZWPRg09AEmS1PDGjx9P+/btadasGZdeeimtW7debNlGjRoxePBgevXqxcCBAxk9ejT/+Mc/Flt+7ty5nHLKKQwfPpwtttiCO+64g7POOou//OUvXHTRRUyZMoX11luPGTNmAHDppZdyzTXX0L17d2bNmsX666+/VNdy7733cvnllzNq1Cg23XRTDjvsME4//XR22WUX/vWvf7HPPvvw6quvVpffeOON2X333XnggQc46KCDGDZsGD179mSdddapLrPzzjtzwAEHsN9++1Uvo2zXrh1XXXUVu+22G+eeey7nn38+f/7znxcay89+9jNOOOEEjjjiCK655prq4w8//DBvvPEGEyZMIMsyDjjgAMaNG8egQYMYPXo0jz32GJtvvjmffPIJ48aNo1GjRjzyyCP85je/4e67767Xfbjooot4+eWXqaysXKr7J6nhGM4kSVrDdezYkbfeeouNNtqIUaNGcdBBB/HGG2/UWad169b89Kc/Zf/992f8+PGsu+66iy2bUuLll19mr732AmD+/PlsueWWQB5w+vTpw0EHHcRBBx0EQPfu3TnjjDPo06cPBx98MM2bN6/3tTz22GM8++yzPPzww3zta18D4JFHHlloOeInn3yyyGxc//79ufjiiznooIO4+eabueGGG+rsZ+bMmcyYMYPddtsNgCOPPJJDDjlkkXJPPvlkdZj66U9/yq9+9SsgD2cPP/wwHTp0AGDWrFm88cYb9OjRY5F+jjzySN544w1CCAvN5kla/SwxnMUYM2DjlNKslTCeVV6MsQXwbEpp81rONQOGppT2WEF9DQaOBNqklCYVx1oCk4F7Ukq9imNbA5cDnYAFxfmfp5ReLs6fB5wIvAOsD0wEjkkpfRZj/AewHrAusD3wctH98ymlo2KM3wD+BPQAPiFfKjsOOCulNLNkrBOAdVNKFTWuoc76xTV+H/igpNqFKaWV84CBJK0BqkIMwA9/+ENOPPFEPvjgAzbffJH/lS3kpZdeYpNNNuG9996rs1yWZbRu3Zrx48cvcu6BBx5g3LhxjBgxgt/97ndMmjSJM888kx/96EeMGjWKrl278sgjj7DDDjvU61q+9a1v8eabb/L666/TuXNnABYsWMD48eNp3LjxYut1796dqVOn8vjjjzN//vxl2gBlcUIIixzLsoxf//rXHHfccXXWPeecc9hjjz249957mTp1KrvvvjuQz14uWLCgutwXX3yxwsYrqeH4zNkyiDGuFWNc9L+0S5BSemdFBbMSz5MHtCp9ycMVADHGdYCHgfEppW+llLYDbgAeiTFuWlLv1iI4tQO2Bk4uxtylOP5DYEZKqaL4c1SMcQPyIPUG8J2i3HeB94CmJWNoDXwD2DDG2LHkeL3qAxeV9FthMJOkFevdd98lyzIAJkyYwIIFC9hss83qrHPPPffw4YcfMm7cOE499dTqJYm1iTHy/vvvV4ezuXPnMmnSJBYsWMC0adPYY489uPjii5kxYwazZs1i8uTJtG3bll/96ld07tyZ1157rd7Xsu2223LPPfdwxBFHMGnSJAD23ntvrr766uoyi1vmd8QRR9C7d2+OOuqoWs9vvPHG1TNuTZo0YdNNN+WJJ54AYMiQIdWzaKW6d+/OsGHDABba4GOfffbhL3/5C7Nm5T/7fvvtt/nPf/6zSP2ZM2ey1VZbAflzZlVatGhBZWVl9T2cMGFCneOVtGqo77LGU2OMPwY2AwaklO4GiDH+APgjsDbwPnBcSumfMcbdgSuBCUBXYC7wU+C3QBtgGnBwMTOzLvB7YDfy2ZmXgBPqmqmLMR4BDAAy8lmg41JK/4kxjgdOTSk9E2O8FtgtpdQ6xtgIeBfYFjgEOAz4uBjLDKBnSundou1fAr2Ke/M2+QzSu8Xs0nbARsC3gd1jjL8Dvgd8CcxKKXUvGePvyQPNBsDRKaW/15xVK2Ylzwf2Lu7tb6ru7VL4P+DoGOOvyWfFfkIevroV53sDM1NKl1dVSCndFWM8hDyA/a60sZTSnBjjU8A29ej7MOCjlNLvS+p/UbNN4GjgVvL71I//hsf61pckLaPevXszduxYPvjgA5o3b87555/P0UcfzaBBg4B8A4q77rqL6667jkaNGtG4cWOGDRtWPdtTW/0DDzyQM888kzFjxrD11ltz8skn87Of/Yxbbrml1jGsu+663HXXXZx66qnMnDmTefPmcdppp7H99ttz+OGHM3PmTLIs4/TTT2eTTTbhnHPO4bHHHmPttddmxx13ZN99912qa44xMnToUA455BBGjhzJwIEDOemkk2jXrh3z5s2jR48e1ddfqk+fPpx99tn07t271nYPPfRQjjnmGAYOHMhdd93FLbfcwvHHH8/s2bP51re+xc0337xInSuvvJLDDjuMK6+8kp49e1Yf33vvvXn11Vfp1i3/3/VGG23EbbfdRtOmTReq/8tf/pIjjzySyy+/nO9973vVx7t3707Lli1p27Ytbdq0oWPHjtS02Wab0b17d9q0acO+++7LJZdcUr8bKKnBhKqflC1OESBOSSldHWPsDvxfSmmrGGNTYBJ5AHolxng0cGxKqUsRzv4G7JRSqowxXgMcCHRNKU2PMY4iX3Z3Y4zxbICU0oVFf38C5qWUat32KcbYhnwmqFNK6d9FQNo+pfSTGOOF5CHpohjji8AcYH+gBXBxSmnXGGNf4AqgXUppWozxBuA/KaWzYoyHky+vOz6ltCDGeAKwS0qpTxHO+gEdU0ofxBg7AHcAOxRlN00pfVwEsCnA/iml+2OMfYATU0rdFxPOfptSuiDGGIGngFZF0LyLPAzWpltK6fNiyd+z5LNNtwNfACcB9wP7pZR6xRgvBRqllE6rcR9PB7oXZc4DNkop/SLGuHHx3v0mpfRoSfmFxl4cuxaYU7PtGv2sA0wHdi7ej+eBrVJKX9az/mAWXdbYN6W0VE83V70vY8aMWapnF7RqevLAnksuJK1Gug9f2p/rqTZ33XUXw4cPZ8iQIQ09FK3hpk+fzp577gnQMqU0tYGHo5WovjNnw4q/nwaaxRjXB7oAL6SUqp6wvRm4tvhwD5BKPkBPBLZNKU0vXj/Hf4PHAcDXYoy9itfrAS/UMZY9gFEppX8Xr68vKf8o8JsY41DgQ+BxYE+gJTCmpI0nU0rTSq5pr5KxdAYm5lmJRsDMknqjUkpVIeFN8hnDm2KMj5IHoiqzUkpVr58GLqvjem6C/GbFGCeSzzSOqHperJ4GA8eSz0wNBkofEqjv8ssjYox7Ad8BRgKPLUX/QPWM5hnAJsCvUkp3kIfjlFKaXJR5Hvgx//2eWlJ9yJc1Xl2zvCRJK8opp5zCgw8+yKhRoxp6KJLWYPUNZ18ApJTml4SWQL6ssM46hfm1vK56KjeQzyw9Sv3U1m/V6yeBjsCPyMPY4+SzXS2Bc0vK1xxL1X0I5JtN/GUxfVcvtSw2q2gN7E4eAP9U8jzVl4tpf0mqr60+M2clrx8DrgHWIV9CeHjJuRfIN/uoqSvwYsnrW4uZs28CfweOB65bwnifB6oX5qeUbgVuLcZe9f72A3aMMU4tXm9YHBtWz/qSJH3lrrrqqoYegiQt14Yg44GKGGPV9klHku/gt7RPno4AzogxNgaIMW4cY2xVR/kxwA+LEAFwDPAIQErpS/JZujOLY08D3ck3uXi6nmM5sWqjjBjjejHG9rUVjDFuATROKY0u+psJfKsefdR0VNHed4AK4B/FtfSqsQlG6Z/SYEZKKQNOB05PKc2r0f4wYNMY4xklY+9FHioXmY0qnr37GXBOMUNal78CW8QYz4wxrl20HSiCVYxxS/Jloi1TSi1SSi3INxvpHGPcZkn1JUmSpDXJMoezlNL75Jt8/LV4vutwFp6xqa+LyGd3nina+Tuw2HBWbBn/a+BvRfn25GGiyhhgU/Lno+YC/wSeSSnNqcc1DQGGAo8XbT9HHu5qszX5jocvkM9APUj9AmBNX8YYnyRfFnlcSmnRrZrqIaU0OqU0spbjc8g3HOkeY5wSY5wMHAfslVL6aDFtPQC8VpSrq8/PyMNXK+CfxZLFJ8nv+WjgCODB0sBebPhxH/lzY0uqX+XMGGNlyZ/j63FLJEmSpFXKEjcE0Vcn+jvkVho3BFmzuCGI1jRuCCKtXtwQZM3l7zmTJEmSpDJQ340qVroY4yDyTStKzUspdW6I8XwVUkpL/YusJUmSJK2eyjacpZR8rkiSJEnSGsNljZIkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYaNfQAJGlFWjBnDt2H393Qw5BWqgVz5rDWuus29DAkScvJmTNJqxU/oGpN5Pe9JK0eDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnkiRJklQGDGeSJEmSVAYMZ5IkSZJUBgxnklRiwZw5DT0Eaan5fStJq4dGDT0ASSona627Lk8e2LOhhyEtle7D727oIUiSVgBnziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkiSpDBjOJEmSJKkMGM4kSZIkqQwYziRJkv6fvXuPj6q69///WhAuCYZQFVQMFSqHJZCEcAkXMQGKIYoeKCJi5BRDjApi+dnW9FhrLWC8VFHOD7GAgCCKEAEFsWIVlIMgHII4BAVWEInlVgWBKBe5hP39YybThISQcMsA7+fj4aMze6+91mfvGW3eWWvviIiEAIUzERGRC1h6ejoNGjQgJibmhG3mzZtHXFwc8fHxtGvXjqVLlwb3/fd//zcxMTHExMSQnZ1dqbG7du3KqlWrTrn2E8nPzw+ez6pVqxg2bBgAw4cPZ9SoURXup7Ltp06dyvbt2ytXbDHF6y5PZmYmLVu2JDMz85TGueSSS4LjvfHGG6fUh4hUjbCqLkBERETOnrS0NB588EEGDhx4wjbdu3enV69eGGPIzc3ljjvuYMOGDfz9739n9erV+Hw+Dh06RJcuXbj55pupW7fuOTyD8rVr14527dqdk7GmTp1KTEwMDRs2PKvjTJgwgZ07d1KrVq3T6qconN11111nqDIROds0cyYiInIBS0pK4tJLLy23zSWXXIIxBoD9+/cHX69bt44uXboQFhZGnTp1aNWqFe+//z4FBQVYa3HOAZCamsrEiRPLHeODDz6gU6dOtGnThn79+rFv3z4AHnnkEVq0aEFcXBwPP/wwALNmzSImJoZWrVqRlJRUbr+LFy/m1ltvLbV94sSJ3HzzzRw8eJBNmzZx00030bZtWxITE9mwYUOJtps2baJNmzbB9xs3bqRt27Yl2syePZtVq1YxYMAA4uPjOXjwIIsWLaJ169bExsaSnp7OoUOHStXx2Wef0apVKzp16sRLL70U3F5YWEhmZiYJCQnExcUxYcIEAHr16sX+/fvp0KED2dnZzJ8/nw4dOtC6dWtuvPFGvv32W6D0rF9MTAz5+fklxn7kkUf45JNPiI+PZ/To0eVeRxEJDQpnIiIiwttvv811113HLbfcwiuvvAJAq1atWLBgAQcOHGDXrl18/PHHbNmyhaioKMaOHUtaWhozZ85kz5493HvvvSfse9euXWRlZbFw4UJWr15Nu3bteOGFF9i9ezdvv/02X375Jbm5uTz22GMAjBw5kn/84x+sWbOGd955p9LnMnbsWObPn8/cuXMJDw/nvvvu48UXX+Szzz5j1KhRPPDAAyXaX3vttURFReHz+QCYMmUKaWlpJdrcfvvttGvXjunTp+Pz+TDGkJaWRnZ2NmvXruXo0aOMGzeuVC2DBg1izJgxLF++vMT2yZMnExUVRU5ODjk5OUycOJHNmzfzzjvvEB4ejs/no3///txwww2sWLGCzz//nDvvvJNnn322wtfhmWeeITExEZ/Px29/+9sKHyciVUfLGkVERIQ+ffrQp08flixZwp///GcWLlxIjx49yMnJ4frrr6d+/fp06tSJsDD/jw7JycnMmjWLoUOHsmbNmnL7XrFiBevWraNz584AHD58mE6dOlG3bl1q165NRkYGt9xyS3AGrHPnzqSlpXHHHXdw2223Veo8XnvtNaKjo5k7dy41atRg3759fPrpp/Tr1y/YpqwZroyMDKZMmcILL7xAdnY2K1euLHcc5xxNmjShWbNmANx999289NJLPPTQQ8E2BQUF7N27ly5dugDw61//mgULFgD+mcTc3Fxmz54dbLtx40aaNGlSYpytW7fSv39/duzYweHDh0vtF5ELS0jPnFlrfdba8MDr6621X1hrP7fWdrPWvmetvfY0+n7IWtug2PvB1tpz9msla228tfaOCrZtbK297xzU9L619v7jthlr7WZrbfnrSsru74x/ftbafGvtye+mFhGRU5KUlMSmTZvYtWsXAH/605/w+Xx8+OGHeJ7Hf/zHfwBw7Ngx1q9fT3h4OLt37y63T8/zSE5Oxufz4fP5WLduHZMnTyYsLIyVK1fSt29f5s6dy0033QTA+PHjycrKYsuWLcTHx/P9999XuP6i5X1bt24N1lmvXr3g2D6fj/Xr15c6rm/fvixYsIB3332Xtm3bctlll530nLnGeIgAACAASURBVE7G87zgEtGy9r344ovBmjZv3kyPHj1KtfvNb37Dgw8+yNq1a5kwYQI//fQTAGFhYRw7dizYrmi7iJzfQjKcWWvDAJxz8c65g4HNvwZedc61ds597Jzr6ZzbdBrDPAQEw5lzbrxz7lwuyI4HKhTOgMbAWQ9nwCvAoOO2dQWOOueWVLSTc/T5iYjIGfLVV18Fw8bq1as5fPgwl112GYWFhcFglJubS25ubjBAjB49mubNmzNjxgzS09M5cuTICfvv2LEjy5Yt46uvvgLgwIED5OXlsW/fPgoKCujZsyf/8z//E1xWuGnTJjp06MDIkSO5/PLL2bJlS4XPpXXr1kyYMIFevXqxfft26tatS5MmTZg1axbgD0VlzfTVrl2blJQUhgwZwqBBx/9foV9kZCQ//vgjANdddx35+fnBc3rttdeCM2RF6tWrR1RUVPDpl9OnTw/uS0lJYdy4ccHrlpeXx/79+0uNWVBQwNVXXw3Aq6++GtzeuHFjVq9eDfg/s82bN5dbr4icH87KskZr7Z+BS51zvw28vwzIA34BPAZ0AWoCa4Ehzrl91tqpwI/AfwD1gbbWWg+IBIYA/YED1toBQCdgPXCrc+4La+3VwJjAsQAznHNPW2vvAv6/wFgADzvnFllr/wQ0BGZba38C7sIflC5xzj1sra0O/BW4KXDc+8B/O+cKA3X+BDQDGgHLgbudc2X+Ci0wO/cGcEVg00IgCxgJ1LXW+oAlzrlh1trpgAVqAV8B6c65PcBLQJNA26+cc7cXXRvn3L7AOEXX6hjwKtASOAI451xFQ+Bc4G/W2hbOuXWBbYOAKYExagJPcuY/v6uAF4GfA+H4P7+nAmMmAn8DDgIrgLJ/BSkiImVKTU1l8eLF7Nq1i+joaEaMGME999zD+PHjARg8eDBz5sxh2rRp1KhRg/DwcLKzszHGcOTIERITEwGoW7cur7/+OmFhYeTl5TFp0iRWrlxJZGQkSUlJZGVlMWLEiDJrqF+/PlOnTiU1NTW4pDArK4vIyEh69+7NTz/9hOd5wYdWZGZmsnHjRjzPo3v37rRq1apS53zDDTcwatQobrnlFj788EOmT5/OkCFDyMrK4siRI9x5551l9jlgwADeeuutMmewwP/ky8GDBxMeHs7y5cuZMmUK/fr14+jRoyQkJDB48OBSx0yZMoX09HQiIiJISUkJbs/IyCA/P582bdrgeR7169dn7ty5pY4fPnw4/fr14+qrr6Zjx47BENa3b1+mTZtGfHw8CQkJweWVxcXFxREWFkarVq1IS0vTfWci5wFTkWn5yrLW/hz4P6CRc+6otfY3QGvgawDnXFag3V/xz8r8KfDDfQzQxTm3P7A/GEAC+1c558YG9uXz7x/uPwbec849F9h3uXNuVyAU7nbOedZaCyxyzkUff3zg/XD+Hc6GAH2BnoFTWgDMds6NC9TRFLgRfxD6HHjIOffhCa7Fb4EY59w9gfc/c87tsdamBca/vVjby51zuwKvs4Aw59wj1tquwCjnXLtibU8UzpKBoc65G48brx6w+AQf2ZfOuQGB9mOAn5xzf7DWRgLbgObOuW3W2sfO0uf3IfCEc25JIAAuwh9el+D/zgxwzi0OLAPNBmKLPreKstY2BjYvWrSI6OjoyhwqF6FlvftWdQkildJ53pyqLuGCMGrUKAoKCnjiiSequhS5yG3dupXu3bsDNHHO5VdxOXIOnZWZM+fcP6216/CHm3eANPzLCJ/HP1tUFEhqAcXXFswu+sG+oqy1lwDX4w8lRePvCry8FpgRmFk7Alxprb3SOfevk3R7IzDVOXc4MMYUoA9Q9Bimuc65nwL7VgfGKTOc4Z/t+Z219jngf4F/lDPuwMDMUk2gDv7ZxspaA1xnrX0Jfxj7O4Bzbi/+pZQnMxl431r7KP7ZrqXOuW2Bfb04859fHfxLJ+v78zPgD5nNgW+BA865xYFzeNNa+3Jl+hcREamIPn36sGnTJj766KOqLkVELmJn82mNU4G7rbVfA1HOuU+stQZ4wDl3ov/y7TvDNcwAfu+cm2utrQYcAGpX4DgDHD+lWPx98btuCynnOjrnlltr4/GHx18DjwA3HN8usHxvCHC9c25nYElmefeZFRK4Z9BaGzwn59zX1trmQHfgZuApa20s/vNefIK+gjNnzrk11tod+Jd0DgKK34d3Nj6/avivbYJzrsQNC9bayq1jEREROUVvv/12VZcgInJWHwgyB0gCHsYf1MA/i/a7Yk/wiwwEiVMWWNb3KRBcSG2tvTzwsh5QdIfsPfhneor8AESdoNsPgTRrbQ1rbQ3gbvz3ilWatbYJ8INzbibwO/z3YlUrY/x6QAHwvbW2FpB+klo3AQmB13cVGy8aKHTOzcV/Terjv/9vb+ABHWX9M+C4vl8BhuO/r674H5g5G5/fj8An+ENr0Tk0stZeCWwAwoueFBmYsTvRZyYiIiIicl47a+HMOXcAmId/tmhaYPMz+JfB5Vhrc4Gl+Jevna7/AjoHHtW+Bn8QA/9SyrnW2qX4n3hY/Fm8Y4Apgce9tziuv5eBXPz3k30eeD3xFGvrCnweeJjHAmCwc+4Y/vuq6lhr1wTu81qAP3BtCLxeXayPXMAFzm92YNtvgQnW2k/wB7AiscDywHVYCTztnNteyZqn43+gyPSipZ0BZ+vzGwC0sNautdauxX9fWT3n3CEgFXjJWrsSaAf88wyMJyIiIiIScs7KA0FEQo0eCCKVoQeCyPlGDwQRubDogSAXr5D8O2ciIiIiIiIXm7P5QJCLirV2PNDxuM1Hiz/+XkRERERE5EQUzs4Q51zpvzwpIiIiIiJSQVrWKCIiIiIiEgIUzkREREREREKAwpmIiIiIiEgIUDgTEREREREJAQpnIiIiIiIiIUDhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAgIq+oCRERCybHDh+k8b05VlyFSKccOH6ZazZpVXYaIiJwmzZyJiBSjH3DlfKTvrYjIhUHhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMRETkPHf4SGFVlyAXKH23RM6tsKouQERERE5PzRrV+c/fz6vqMuQCNP/53lVdgshFRTNnIiIiIiIiIUDhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiEilpaen06BBA2JiYk7YZvr06cTFxREXF8f111/PmjVrgvvef/99rLU0bdqUZ5555lyUfFq6du3KqlWrAOjZsyd79+4lPz+/3PM/XmXb79ixo4Uxpmeliy3GGLPYGNPuJG0SjTFfGmN8xpjwUxhjqjHm9sDrh4wxEada78VO4UxEREREKi0tLY3333+/3DZNmjThf//3f8nNzeXPf/4z9913HwCFhYUMHTqUBQsWsG7dOmbMmMG6devORdklFBYWntJx7733HvXq1TvD1ZR26NChFsBphbMKGgCM8jwv3vO8g6fZ10OAwtkpUjgTERERkUpLSkri0ksvLbfN9ddfz89+9jMAOnbsyNatWwFYuXIlTZs25Re/+AU1a9bkzjvvZN68eRQUFGCtxTkHQGpqKhMnTizVb+PGjXn00Ufp1KkT7dq1Y/Xq1aSkpHDttdcyfvx4ADzPIzMzk5iYGGJjY8nOzgZg8eLFdOvWjbvuuovY2FgAXn/9ddq3b098fDz333//SUNb48aN2bVrV4ltX3/9Na1btyYnJ4fCwkIyMzNJSEggLi6OCRMmlOojMTERn88XfN+5c2dyc3OD7z3P4/Dhw78F+gdmtPobYy41xsw1xuQaY1YYY+KO79cYE26MmRlokw2EF9vXwxiz3Biz2hgzyxhziTEmA7gDeNwYMz2wbVGgzVpjTO/AsY2NMV8U6+thY8zw48YeBjQEPjbGfFzuRZQyKZyJiIiIyFk3efJkbr75ZgC2bdtGo0aNgvuio6PZtm0bUVFRjB07lrS0NGbOnMmePXu49957y+yvUaNGLF++nMTERNLS0pg9ezYrVqzg8ccfB+Ctt97C5/OxZs0aFi5cSGZmJjt27AD84fDJJ59k3bp1rF+/nuzsbJYtW4bP56N69epMnz69UufmnKNv375MmTKFhIQEJk+eTFRUFDk5OeTk5DBx4kQ2b95c4piMjAymTp0KQF5eHocOHSIu7t9ZyxhDzZo1RwPZgRmtbGAE8LnneXHAo8C0MsoZAhwItHkSaBvo73LgMeBGz/PaAKuA33meNwl4B8j0PG8A8BPQJ9CmG/C8McZU5Dp4njcG2A508zyvW0WOkZLCqroAEREREbmwffzxx0yePJmlS5cC/lmh4xX9/J+cnMysWbMYOnRoiXvUjterVy8AYmNj2bdvH5GRkURGRlK7dm327t3L0qVLSU1NpXr16lxxxRV06dKFnJwc6tatS/v27WnSpAkAixYt4rPPPiMhIQGAgwcP0qBBgwqf286dO+nduzdz5syhZcuWAHzwwQfk5uYye/ZsAAoKCti4cSPNmjULHtevXz+eeOIJnnvuOV555RXS0tIqMtwNQF8Az/M+MsZcZoyJ8jyvoFibJGBMoE2uMaZoOq4j0AJYFrjWNYHlZYxhgKeMMUnAMeBq4IqKFCenT+HsAmSt9YBI59y+qq7lTLHWxgPNnHNvVnUtIiIiUnG5ublkZGSwYMECLrvsMsA/U7Zly5Zgm61bt9KwYUMAjh07xvr16wkPD2f37t1ER0eX2W+tWrUAqFatWvB10fujR4+WGQCL1KlTJ/ja8zzuvvtunn766VM6v6ioKBo1asSyZcuC4czzPF588UVSUlJKtM3Pzw++joiIIDk5mXnz5vHmm28GHzZyEmXNYJV1omVtM8CHnuelnmSMAUB9oK3neUeMMflAbeAoJVfd1T55uVJZWtYo54t4/OuhRURE5Dzxz3/+k9tuu43XXnutxKxRQkICGzduZPPmzRw+fJiZM2cGZ8JGjx5N8+bNmTFjBunp6Rw5cuSUxk5KSiI7O5vCwkJ27tzJkiVLaN++fal23bt3Z/bs2Xz33XcA7N69m2+++abC49SsWZO5c+cybdo03njjDQBSUlIYN25csPa8vDz2799f6tiMjAyGDRtGQkJCmffvGWP2AZHFNi3BH54wxnQFdnme98NxhxVvEwMUrZVcAXQ2xjQN7IswxjSjtCjgu0Aw6wZcE9j+LdAgMFtXC7j1BJfkx+NqlkrQzNkFzlqbgH9quw6wHxjmnMux1oYBfwcuw3+j6ErgfufcYWttGnAXsAeIAfYCfZ1z/ypnnFFAF/xT5LuAdOfcN9baxvjXNE8EbgqMNQAYDHQADgK9nXP/OtG4wBFgJFDXWusDljjnhp2RCyQiIiKnJDU1lcWLF7Nr1y6io6MZMWIE99xzT/CBHIMHD2bkyJF8//33PPDAAwCEhYWxatUqwsLCGDt2LCkpKRQWFpKenk7Lli3Jy8tj0qRJrFy5ksjISJKSksjKymLEiBGVrq9Pnz4sX76cVq1aYYzh2Wef5corr2TDhg0l2rVo0YKsrCx69OjBsWPHqFGjBi+99BLXXHPNCXourU6dOrz77rskJydTp04dMjIyyM/Pp02bNnieR/369Zk7d26p49q2bUvdunUZNGhQmf3Wq1dv+bfffnuPMcYHPA0MB6YElioeAO4u47Bxxdr48P+Mh+d5O40xacCMQLgC/z1oeccdPx2Yb4xZFTh+Q+D4I8aYkcD/AZuLtpfhZWCBMWaH7jurPFPelK+cn4qWNQKHga/wB6WF1truwBSgKf7Ac6lz7ntrrQFeBT51zo0PhKTRQJxzbou1diLwnXPuT+WMeblzblfgdQZwo3PuzkA42wzc6pz7u7U2E/gT0NU557PW/g3Y7Zx7rLxxA/tudc7dforXpDGwedGiRSdcHiEicj77z9/Pq+oS5AI0//neVV3CBW379u107dqVDRs2UK3avxe0bd26le7duwM0cc7lV1V9cu5pWeOFzQKHnXMLAZxzi/AHNov/s384MBOVC/wS/9LBIsucc0WLwVcA155krJuttSustV8ADx/X1z7n3N8Dr1cDW51zRc+O/Qx/WDzVcUVERETOO9OmTaNDhw48+eSTJYKZXNz0TbiwGU58k+hd+J/4k+iciwX+RskbO38q9rqQcpbAWmuvwT/jleqciwHSj+vr0HF9ldd3hccVEREROV8NHDiQLVu20K9fv6ouRUKIwtmFbQNQy1rbDSDwvzXwry2uB+xyzv1orY3CH9ZOVV38M3L/stZWw38/2Zn2A/4bVEVERERELkgKZxcw59xh/A/UeMpamws8Bdwe2D4NiLTWfgnMAj45jXHWBvr4EvgI/z1mZ9oioI61do21dsxZ6F9EREREpErpgSByUdADQUTkQqcHgsjZoAeCVA09EOTipZkzERERERGREKCHLUiFBR6R/2AZu9KKPX1RREREREROgcKZVJhzbhIwqarrEBERERG5EGlZo4iIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiIiIhACFMxERERERkRCgcCYiIiIiIhICFM5ERERERERCgMKZiIiIiIhICFA4ExERERERCQEKZyIiIiIiIiEgrKoLEBERkdNz+Egh85/vXdVlyAXo8JFCataoXtVliFw0NHMmIiJyntMPz3K26Lslcm4pnImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYhUgWOHD1d1CXIB0fdJROTCEFbVBYiIXIyq1azJst59q7oMuUB0njenqksQEZEzQDNnIiIiIiIiIUDhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiIiIhACFMxERkQtYeno6DRo0ICYm5oRtNmzYQKdOnahVqxajRo0qsW/06NG0bNmSmJgYUlNT+emnnyo8dteuXVm1atUp134i+fn5wfNZtWoVw4YNA2D48OGl6i9PZdtPnTqV7du3V67YYorXXZ7MzExatmxJZmbmKY1zySWXBMd74403TqkPEakaCmciIiIXsLS0NN5///1y21x66aWMGTOGhx9+uMT2bdu2MWbMGFatWsUXX3xBYWEhM2fOPJvlVlq7du0YM2bMORnrdMNZRU2YMIHVq1fz3HPPnVY/Cmci5x+FMxERkQtYUlISl156abltGjRoQEJCAjVq1Ci17+jRoxw8eJCjR49y4MABGjZsSEFBAdZanHMApKamMnHixHLH+OCDD+jUqRNt2rShX79+7Nu3D4BHHnmEFi1aEBcXFwyHs2bNIiYmhlatWpGUlFRuv4sXL+bWW28ttX3ixIncfPPNHDx4kE2bNnHTTTfRtm1bEhMT2bBhQ4m2mzZtok2bNsH3GzdupG3btiXazJ49m1WrVjFgwADi4+M5ePAgixYtonXr1sTGxpKens6hQ4dK1fHZZ5/RqlUrOnXqxEsvvRTcXlhYSGZmJgkJCcTFxTFhwgQAevXqxf79++nQoQPZ2dnMnz+fDh060Lp1a2688Ua+/fZboPSsX0xMDPn5+SXGfuSRR/jkk0+Ij49n9OjR5V5HEQkNCmciIiJSpquvvpqHH36Yn//851x11VVERUXRo0cPoqKiGDt2LGlpacycOZM9e/Zw7733nrCfXbt2kZWVxcKFC1m9ejXt2rXjhRdeYPfu3bz99tt8+eWX5Obm8thjjwEwcuRI/vGPf7BmzRreeeedStc9duxY5s+fz9y5cwkPD+e+++7jxRdf5LPPPmPUqFE88MADJdpfe+21REVF4fP5AJgyZQppaWkl2tx+++20a9eO6dOn4/P5MMaQlpZGdnY2a9eu5ejRo4wbN65ULYMGDWLMmDEsX768xPbJkycTFRVFTk4OOTk5TJw4kc2bN/POO+8QHh6Oz+ejf//+3HDDDaxYsYLPP/+cO++8k2effbbC1+GZZ54hMTERn8/Hb3/72wofJyJVR+HsAmCt9Vlrw0/x2HrW2j9Uor1nrb0k8DrfWnvyxfNnQGXrFBGR07dnzx7mzZvH5s2b2b59O/v37+f1118HIDk5mdjYWIYOHcqkSZPK7WfFihWsW7eOzp07Ex8fz6uvvso333xD3bp1qV27NhkZGbz11ltEREQA0LlzZ9LS0pg4cSKFhYWVqvm1115jwYIFzJkzh1q1arFv3z4+/fRT+vXrR3x8PPfffz87duwodVxGRgZTpkyhsLCQ7Oxs7rrrrnLHcc7RpEkTmjVrBsDdd9/NkiVLSrQpKChg7969dOnSBYBf//rXwX0ffPAB06ZNIz4+ng4dOvD999+zcePGUuNs3bqVlJQUYmNjee655/jyyy8rdT1E5PyicHYBcM7FO+cOnuLh9YDzIfScL3WKiFwwFi5cSJMmTahfvz41atTgtttu49NPPwXg2LFjrF+/nvDwcHbv3l1uP57nkZycjM/nw+fzsW7dOiZPnkxYWBgrV66kb9++zJ07l5tuugmA8ePHk5WVxZYtW4iPj+f777+vcM1Fy/u2bt0arLNevXrBsX0+H+vXry91XN++fVmwYAHvvvsubdu25bLLLjvpOZ2M53kYY06478UXXwzWtHnzZnr06FGq3W9+8xsefPBB1q5dy4QJE4IPZAkLC+PYsWPBdpV5UIuIhK6wqi5ATp+11gMinXP7rLX5wDQgGbgKGOWcG2utrQaMBX4JHAL2Oec6Ay8B9ay1PuCAc+56a+3vgTvxfz9+AoY453wnqWEx8BnQHmgM/P/ANuA3QEMg0zk3K9C2A/AMUDdw+OPOub9baxsDq4AJQE8gArjHObe0rDpP/YqJiEhF/PznP2fFihUcOHCA8PBwFi1aRLt27QD/UxybN2/OU089RXp6OsuXLy/znjWAjh07MnToUL766iuaNm3KgQMH2Lp1Kw0bNuTAgQP07NmTjh070rRpU8B/D1iHDh3o0KED8+fPZ8uWLScNS0Vat27NkCFD6NWrF//4xz9o2LAhTZo0YdasWfTr1w/P88jNzaVVq1YljqtduzYpKSkMGTKEyZMnl9l3ZGQkP/74IwDXXXcd+fn5wXN67bXXgjNkRerVq0dUVBRLly7lhhtuYPr06cF9KSkpjBs3jl/+8pfUqFGDvLw8rr76aurUqVOij4KCAq6++moAXn311eD2xo0b8+677wKwevVqNm/eXG69InJ+0MzZhSnCOdcJ6Ao8E1iG2Aq4EWjhnGsFFN09PRTYG5h9Kwo805xzCc651sCfgfEVHDca6AJ0AEYCMYE+7wBGg395YqC/u5xzbQN1TAhsB7gMWB4YeyTw13LqFBGRk0hNTaVTp04454iOjg4Gj/HjxzN+vP8/7//617+Ijo7mhRdeICsri+joaH744Qc6dOjA7bffTps2bYiNjeXYsWPcd9995OXlMWnSJJ5//nkSExNJSkoiKyvrhDXUr1+fqVOnkpqaSlxcHB07dmTDhg38+OOP3HrrrcTFxdGlS5fgQysyMzOJjY0lJiaGpKSkUkHqZG644QZGjRrFLbfcwq5du5g+fTqTJ0+mVatWtGzZknnz5pV53IABAzDGlDmDBf4nXw4ePJj4+Hg8z2PKlCn069eP2NhYqlWrxuDBg0sdM2XKFIYOHUqnTp0ID//3HQgZGRm0aNGCNm3aEBMTw/3338/Ro0dLHT98+HD69etHYmIil19+eXB737592b17N/Hx8YwbNy64vLK4uLg4wsLCaNWqlR4IInKeMBWZlpfQVsbM2e3OuVWBfVvwz6LtAFYDS4CPgHedc3uKZqucc5cX6+8m4FHgUuAY0Mw5V/sEY93qnPsiMHM21jk3u9i4fZxzq6y11YGjQDj+mbs3gPxip3AZ0AfYBax1zkUG+rgWWOacu7KsOit5jRoDmxctWkR0dPSpdCFyxi3r3beqS5ALROd5c6q6hAvCqFGjKCgo4IknnqjqUuQit3XrVrp37w7QxDmXX8XlyDmkZY0XpuILzwuBMOdcgbW2Jf7ZtO7AX621bY4/0FpbE5gNJDnnVltrG+Jfnngq4/4E4JwrtNaC//tmgFznXKlnIwcC1KHj+tB3VEREzro+ffqwadMmPvroo6ouRUQuYlrWeJGw1tYHwp1z7wOPAAXAL4AfgAhrbVEIqo0/EG0JvH/g+L5O06fAf1hruxWrLcFaW/Yd0/92fJ0iIiJnzNtvv01ubm6JpYMiIueawtnFoxGw0Fq7BsgFFgArnHO7genAWmvtp865H4DHgRxr7RJg/5kswjm3B+gF/MVau8Zaux4Yjn9GrbzjStR5JmsSEREREQkFuudMLgq650xCke45kzNF95yJXFh0z9nFSzNnIiIiIiIiIUDhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREQoDCmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiIiIhACFMxERERERkRCgcCYiIiIiIhICFM5ERERERERCgMKZiIiIiIhICAir6gJERC5Gxw4fpvO8OVVdhlwgjh0+TLWaNau6DBEROU2aORMRqQL6QVrOJH2fREQuDApnIiIiIiIiIUDhTEREREREJAQonImIiIiIiIQAhTMREREREZEQoHAmIiIiIiISAhTOREREREREz4FBnAAAIABJREFUQoDCmYiIiIiISAhQOBMREREREQkBCmciIiLnucNHCqu6BAlB+l6InH/CqroAEREROT01a1TnP38/r6rLkBAz//neVV2CiFSSZs5ERERERERCgMKZiIiIiIhICFA4ExERERERCQEKZyIiIiIiIiFA4UxERERERCQEKJyJiIiIiIiEAIUzERERERGREKBwJiIiIiIiEgIUzkREREREREKAwpmIiIiIiEgIUDgTERERucikp6fToEEDYmJiTtjG8zyGDRtG06ZNiYuLY/Xq1QB8/PHHxMfHB/+pXbs2c+fOPVeln5KuXbuyatUqAHr27MnevXvJz88v9/yPV9n2Pp+P9957r9K1Frdx48aZxph25bUxxiQaY740xviMMeGVHcMYM9UYc3vg9UPGmIhTrVdOn8KZiIiIyEUmLS2N999/v9w2CxYsYOPGjWzcuJGXX36ZIUOGANCtWzd8Ph8+n4+PPvqIiIgIevTocS7KLqGwsPCUjnvvvfeoV6/eGa6mtDMRzipoADDK87x4z/MOnmZfDwEKZ1VI4UxERETkIpOUlMSll15abpt58+YxcOBAjDF07NiRvXv3smPHjhJtZs+ezc0330xERAQFBQVYa3HOAZCamsrEiRNL9du4cWMeffRROnXqRLt27Vi9ejUpKSlce+21jB8/HvDP2mVmZhITE0NsbCzZ2dkALF68mG7dunHXXXcRGxsLwOuvv0779u2Jj4/n/vvvP2loa9y4Mbt27Sqx7euvv6Z169bk5ORQWFhIZmYmCQkJxMXFMWHChFJ9JCYm4vP5gu87d+5Mbm5u8P3hw4d5/PHHyc7OJj4+nuzsbHbv3s2vfvUr4uLi6NixY4n2RQ4ePMidd95JcnIy27dvB6hVtM8Y08MYs9wYs9oYM8sYc4kxJgO4A3jcGDM9sG1RoM1aY0zvwLGNjTFfFOvrYWPM8OJjG2OGAQ2Bj40xH5d7EeWsUTgTERERkVK2bdtGo0aNgu+jo6PZtm1biTYzZ84kNTUVgKioKMaOHUtaWhozZ85kz5493HvvvWX23ahRI5YvX05iYiJpaWnMnj2bFStW8PjjjwPw1ltv4fP5WLNmDQsXLiQzMzMYDFeuXMmTTz7JunXrWL9+PdnZ2Sxbtgyfz0f16tWZPn16pc7TOUffvn2ZMmUKCQkJTJ48maioKHJycsjJyWHixIls3ry5xDEZGRlMnToVgLy8PA4dOkRcXFxwf82aNRk5ciT9+/fH5/PRv39//vKXv9C6dWtyc3N56qmnGDhwYKlaxo0bR0REBB9++CGXXXYZnufFAhhjLgceA270PK8NsAr4ned5k4B3gEzP8wYAPwF9Am26Ac8bY0xFroPneWOA7UA3z/O6VeoiyhkTVtUFiIiIiEjo8Tyv1LbiP+fv2LGDtWvXkpKSEtyWnJzMrFmzGDp0KGvWrDlh37169QIgNjaWffv2ERkZSWRkJLVr12bv3r0sXbqU1NRUqlevzhVXXEGXLl3Iycmhbt26tG/fniZNmgCwaNEiPvvsMxISEgD/zFODBg0qfI47d+6kd+/ezJkzh5YtWwLwwQcfkJuby+zZswEoKChg48aNNGvWLHhcv379eOKJJ3juued45ZVXSEtLO+lYS5cuZc6cOQD88pe/5Pvvv6egoICoqKhgmyVLljBs2DAAatWqhTFmQ+Bz6Ai0AJYFPoOawPIyhjHAU8aYJOAYcDVwRYUviFQ5hTM546y1XYFRzrlyb2At5/g04FPnXF7gfS8g0TmXebp9i4iISMVER0ezZcuW4PutW7fSsGHD4Ps333yTPn36UKNGjeC2Y8eOsX79esLDw9m9ezfR0dFl9l2rln+1XrVq1YKvi94fPXq0zGBYpE6dOsHXnudx99138/TTT1f+BPHP9jVq1Ihly5YFw5nnebz44oslQif4HwhSJCIiguTkZObNm8ebb74ZfNhIeU4Wdsvbhj90feh5XupJhhkA1Afaep53xBiTD9QGjlJyxVztkxYsVULLGi9y1tpQDOhpQPDXU865d5xzmVVXjoiIyMWnV69eTJs2Dc/zWLFiBVFRUVx11VXB/TNmzAguaSwyevRomjdvzowZM0hPT+fIkSOnNHZSUhLZ2dkUFhayc+dOlixZQvv27Uu16969O7Nnz+a7774DYPfu3XzzzTcVHqdmzZrMnTuXadOm8cYbbwCQkpLCuHHjgrXn5eWxf//+UsdmZGQwbNgwEhISyrx/LzIykh9//LHEORUtuVy8eDGXX345devWLXXeRW0OHTqE53nXBXatADobY5oCGGMijDHNKC0K+C4QzLoB1wS2fws0MMZcZoypBdx6gkvyIxB5gn1yDoTiD+ZylllrPeAPwC3AJ8CfrbV/AG7H/53YBtwL/AD8E7jOObcrcOzzwA/OuRHW2puAp4HqwE7gfufcVxWsYTH+GbB3i7/H/9uedsAYa20W8DAQDdzqnLv99M9eREREUlNTWbx4Mbt27SI6OpoRI0Zwzz33BB/IMXjwYHr27Ml7771H06ZNiYiIYMqUKcHj8/Pz2bJlC126dAluy8vLY9KkSaxcuZLIyEiSkpLIyspixIgRla6vT58+LF++nFatWmGM4dlnn+XKK69kw4YNJdq1aNGCrKwsevTowbFjx6hRowYvvfQS11xzzQl6Lq1OnTq8++67JCcnU6dOHTIyMsjPz6dNmzZ4nkf9+vXL/FMBbdu2pW7dugwaNKjMfrt168YzzzxDfHw8f/zjHxk+fDiDBg0iLi6OiIgIXn311VLHDBkyhEGDBpGcnMzu3bsxxqzxPA/P83YaY9KAGYFwBf570PKO62I6MN8YswrwARsAAmFtJPB/wOai7WV4GVhgjNmh+86qhilv2lguTIFw9ohz7q+B9/8FJAGDnXPHrLVDgBuccwOstZOBNc65MYFZtn8C1wMHgC+BLs65ddbae4D7nHMdKrL08EThzDn3bhn70giEs1Nd1mitbQxsXrRo0QmXWIiInM/+8/fzqroECTHzn+9d1SVc0LZv307Xrl3ZsGED1aqd2cVoW7dupXv37gBNnHP5Z7RzCWla1njxKv7rml7AjcBqa60PGAo0Duybin+ZIcDNwPrAfyQ64A9t6wL7pgDx1lpNhYuIiMgFbdq0aXTo0IEnn3zyjAczubhpWePFa1+x1wbIcs69cnwj59wn1tpIa20s/pA2tdgxpzPtqhtTRURE5Lw0cODAMh+FL3K6FPUF/H8f4wFr7c8ArLW1rLWtiu2fBvwe/9LHOYFty/HPlBXdqHo38Llz7kcqZhOQEBivBRBfbN8P+G9oFRERERG5aGjmTHDOvWatvRz4X2st+EP734CiP1DyKv6bR6c45w4Ejtlprf018EbgXrSdwH9VYti/ArOstTcDucDnxfa9DIyy1j4M6CmNIiIiInJR0ANB5KKgB4KIyIVODwSR4+mBIOcvPRDk4qVljSIiIiIiIiFAyxrlrLHWxvPvB4gUN9Y5N+kclyMiIiIiEtIUzuSscc75KPmgDxEREREROQEtaxQREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiIiIhACFMxERERERkRCgcCYiIiIiIhICFM5ERERERERCgMKZiIiIiIhICFA4ExERERERCQEKZyIiIiIiIiFA4UxERERERCQEhFV1ASIiInJ6Dh8pZP7zvau6DAkxh48UUrNG9aouQ0QqQTNnIiIi5zn9AC5l0fdC5PyjcCYiIiIiIhICFM5ERERERERCgMKZiIiIiIhICFA4ExERERERCQEKZyIiIiIiIiFA4UxERERERCQEKJyJiIiIiIiEAIUzERERERGREKBwJiIicp47fKSwqkuQEKPvhMj5KayqCxAREZHTU7NGdf7z9/OqugwJIfOf713VJYjIKdDMmYiIiIiISAhQOBMREREREQkBCmciIiIiIiIhQOFMREREREQkBCiciYiIiIiIhACFMxERERERkRCgcCYiIiIiIhICFM5ERERERERCgMKZiIiIiIhICFA4ExERERERCQEKZyIiIiIXkfT0dBo0aEBMTMwJ23iex7Bhw2jatClxcXGsXr06uO+f//wnPXr0oHnz5rRo0YL8/PxzUPWp69q1K6tWrQKgZ8+e7N27l/z8/HLP/3iVbe/z+XjvvfcqXWtxW7ZsYfv27bHltTHGJBpjvjTG+Iwx4ZUdwxgz1Rhze+D1Q8aYiFOtV84MhTMRERGRi0haWhrvv/9+uW0WLFjAxo0b2bhxIy+//DJDhgwJ7hs4cCCZmZmsX7+elStX0qBBg7NdcimFhYWndNx7771HvXr1znA1pZ2JcFZBA4BRnufFe5538DT7eghQOKtiCmciIiIiF5GkpCQuvfTSctvMmzePgQMHYoyhY8eO7N27lx07drBu3TqOHj1KcnIyAJdccgkREREUFBRgrcU5B0BqaioTJ04s1W/jxo159NFH6dSpE+3atWP16tWkpKRw7bXXMn78eMA/a5eZmUlMTAyxsbFkZ2cDsHjxYrp168Zdd91FbKx/Qun111+nffv2xMfHc//99580tDVu3Jhdu3aV2Pb111/TunVrcnJyKCwsJDMzk4SEBOLi4pgwYUKpPhITE/H5fMH3nTt3Jjc3N/j+8OHDPP7442RnZxMfH092dja7d+/mV7/6FXFxcXTs2LFE+yIHDx7kzjvvJC4ujiFDhuB5XnCfMaaHMWa5MWa1MWaWMeYSY0wGcAfwuDFmemDbokCbtcaY3oFjGxtjvijW18PGmOHFxzbGDAMaAh8bYz4u9yLKWaVwJiIiIiIlbNu2jUaNGgXfR0dHs23bNvLy8qhXrx633XYbrVu3JjMzk8LCQqKiohg7dixpaWnMnDmTPXv2cO+995bZd6NGjVi+fDmJiYmkpaUxe/ZsVqxYweOPPw7AW2+9hc/nY82aNSxcuJDMzEx27NgBwMqVK3nyySdZt24d69evJzs7m2XLluHz+ahevTrTp0+v1Hk65+jbty9TpkwhISGByZMnExUVRU5ODjk5OUycOJHNmzeXOCYjI4OpU6cCkJeXx6FDh4iLiwvur1mzJiNHjqR///74fD769+/PX/7yF1q3bk1ubi5PPfUUAwcOLFXLuHHjiIiIIDc3l2HDhvHTTz8BYIy5HHgMuNHzvDbAKuB3nudNAt4BMj3PGwD8BPQJtOkGPG+MMRW5Dp7njQG2A908z+tWmWsoZ1ZYVRcgIiIiIqGl+KxNEWMMR48e5ZNPPuHzzz/n5z//Of3792fq1Kncc889JCcnM2vWLIYOHcqaNWtO2HevXr0AiI2NZd++fURGRhIZGUnt2rXZu3cvS5cuJTU1lerVq3PFFVfQpUsXcnJyqFu3Lu3bt6dJkyYALFq0iM8++4yEhATAP/NUmSWWO3fupHfv3syZM4eWLVsC8MEHH5Cbm8vs2bMBKCgoYOPGjTRr1ix4XL9+/XjiiSd47rnneOWVV0hLSzvpWEuXLmXOnDkA/PKXv+T777+noKCAqKioYJslS5YwbNgwAJo3b06tWrWKdnUEWgDLAlmrJrC8jGEM8JQxJgk4BlwNXFHR6yGhQeHsHLPW+oBOzrnTXRd8fL9pwK3OudtP4/hPnXN5J2lXDRgB9Mf/G5otzrlbKjlWPeA+59yzxbYtBkY5596tZOkiIiJyhkVHR7Nly5bg+61bt9KwYUOOHDlC69at+cUvfgHAr371K1asWME999zDsWPHWL9+PeHh4ezevZvo6Ogy+y4KHdWqVSseQKhWrRpHjx4tMxgWqVOnTvC153ncfffdPP3006d0jlFRUTRq1Ihly5YFw5nnebz44oukpKSUaFv8oScREREkJyczb9483nzzzeDDRspzorBbkW34Q9eHnuelnmSYAUB9oK3neUeMMflAbeAoJVfL1T5pwVJltKzxHHPOxZ/pYHaGpAHNTtYI/82iFmjpnIsD0k9hrHrAH07hOBERETkHevXqxbRp0/A8jxUrVhAVFcVVV11FQkICe/bsYefOnQB89NFHtGjRAoDRo0fTvHlzZsyYQXp6OkeOHDmlsZOSksjOzqawsJCdO3eyZMkS2rdvX6pd9+7dmT17Nt999x0Au3fv5ptvvqnwODVr1mTu3LlMmzaNN954A4CUlBTGjRsXrD0vL4/9+/eXOjYjI4Nhw4aRkJBQ5v17kZGR/PjjjyXOqWjJ5eLFi7n88supW7duqfMuarNhwwYOHTpUtGsF0NkY0xTAGBNhjCnrZ7Yo4LtAMOsGXBPY/i3QwBhzmTGmFnDrCS7Jj0DkCfbJOaKZs3PMWusBkc65fdbafGAakAxchX/maGygXSLwN+Ag/n8pewO3OOe+KKvf48a4EpgB1MX/25G/O+f+ENjXG8gCCvF//g8CTYB2wBhrbRbwsHNu4Qm6/z2Q6Jw7AuCc+7bYuPnA60B3/FPpjwANgLuAS4FBzrlPgJeAeoFZxAPOuesDXXSx1j6C/4bUN51zjwT6/QuQin+mzgO6Oef2nuw6iIiISGmpqaksXryYXbt2ER0dzYgRI7jnnnuCD+QYPHgwPXv25L333qNp06ZEREQwZcoUAKpXr86oUaPo3r07nufRtm1b7r33XvLy8pg0aRIrV64kMjKSpKQksrKyGDFiRKXr69OnD8uXL6dVq1YYY3j22We58sor2bBhQ4l2LVq0ICsrix49enDs2DFq1KjBSy+9xDXXXHOCnkurU6cO7777LsnJydSpU4eMjAzy8/Np06YNnudRv3595s6dW+q4tm3bUrduXQYNGlRmv926deOZZ54hPj6eP/7xjwwfPpxBgwYRFxdHREQEr776aqljhgwZEmzTrFkzatf2T3B5nrfTGJMGzAiEK/Dfg3b8aqfpwHxjzCrAB2wIHH/EGDMS+D9gc9H2MrwMLDDG7NB9Z1XHlDd1LGdeGeFstnPuYWttY+AL4ErgCPA1MMA5t9haeweQDcSeKJwVX9Zora0NhAXGqAH8A3jWOfe+tXYN8KBz7hNrbXWgjnP/r717j7KyOvM8/i0hxDRqoh2ZATGoYDYiFIU0IiMoSAPeaSVEaRRLlKgd1GGMktUZElfQBWk1SeNl8BKhHZDGSwbQKMFGEEcZB0tKxMsTR0W5SAQNEVQCKZg/zqHkWlZpUeetU9/PWrWo87777P3ss6uo86v3UvFxbU4rTCkdAqwDfgacR+585okRMTu/fwXwcERcn1LqASwEboiIO/NzuCYieufn+mJEfHunvheS+83OMHK/tXkL6AWsB94DWkXEZymlg4HPIuKvtXrBP+//KOCd+fPn7/M0C0lqzM65bnahS1CGPHbb4EKXUNTWrFlD3759eeONNzjggPo/EW3VqlX0798f4OiIWFHvAyizPK2x8P4dIP+N9yegLbnTBj+NiIX5fQ8Bf65Dn82AW/JBrALoDJTl9z0N3JZSuh44LiI+rkO/XyN3EeoBEdETuBi4O6XUfqc2M/P/vkTub2XseFwBdPiC/h+OiG0R8WfgdaA98DEQwLSU0ijgoLoGM0mSpPrywAMP0LNnT26++eb9EszUtPkVVXibd/p8x6mGtbrtaQ3+G3Ao0DN/Xdgs8hd/RsQY4DJgC/BwPvDUSkR8CGwid+oiEfH/yIWwbjs125zfV7XzYz6fW032eC3y/ZwETCIXXCtSSqV7e7IkSdL+NmLECFauXMnQoUMLXYqKkOEsm94AvpFSOgUgpfQ9chd51ta3gPcjYnNK6Qhy16uR7ytFxCsR8a/kQlaP/K6PaznGDOD0fF+tgK7kTsesi4+Bv0kpfeE1j/nTGA+PiGci4mf5sTrXcTxJkiQp8wxnGRQRfyF37dWdKaX/S+5mHe/VoYtJwMkppaXA/wDm77RvYkppef5mHAOAX+S33wOMSyktTSn9fQ19/zNwTkppeb7ff46IfV1YulcR8RG5i1ZfSSk9/wXNvwnMSikty4+5FvhtXcaTJEmSGgNvCNJI5G+2cXZt7taoPXlDEEnFzhuCaGfeEKRx84YgTZdHziRJkiQpA/w7Z41ERBwFkFJ6kT3X7f9ExJX1NVZK6XJyf/9sd+URUVlf40iSJEn6nOGskYmIv2uAMe4D7tvf40iSJEn6nKc1SpIkSVIGGM4kSZIkKQMMZ5IkSZKUAYYzSZIkScoAw5kkSZIkZYDhTJIkSZIywHAmSZIkSRlgOJMkSZKkDDCcSZIkSVIGGM4kSZIkKQMMZ5IkSZKUAYYzSZIkScqA5oUuQJIkfTVbtlbx2G2DC12GMmTL1ipafK1ZocuQVEceOZMkqZHzTbh259eE1DgZziRJkiQpAwxnkiRJkpQBhjNJkiRJygDDmSRJkiRlgOFMkiRJkjLAcCZJkiRJGWA4kyRJkqQMMJxJkiRJUgYYziSpkdq2ZUuhS1BG+LUgScWheaELkCR9OQe0aMFzg4cUugxlwMmzHy10CZKkeuCRM0mSJEnKAMOZJEmSJGWA4UySJEmSMsBwJkmSJEkZYDiTJEmSpAwwnEmSJElSBhjOJEmSJCkDDGeSJEmSlAGGM0mSJEnKAMOZJEmSJGWA4UySpCI2cuRIWrVqRefOnffZZvv27VxzzTV06NCB0tJSXnrppV32f/zxxxxxxBGMHj26TmMfddRRrF+//kvVXZOFCxdy9tlnAzBnzhwmTpwIQHl5OY888kit+6lr+1//+td8+umndSt2JzvXXZNhw4ZRWlrKr371qzqPsWLFiuq1rqys5IknnqhzH5IKx3AmSVIRKy8vZ+7cuTW2efLJJ3nzzTd58803ueeee7jqqqt22T9u3DhOPfXU/Vnml3buuefy4x//uEHG+qrhrDbWrl3L888/z7JlyxgzZsxX6stwJjU+hjNJkorYKaecwmGHHVZjm9mzZzNixAhKSko46aST2LBhA++//z4AFRUV/PGPf2TgwIHV7d99912OPfZY1q9fz7Zt2+jTpw/z5s2rcYxp06Zx4oknUlZWxhVXXEFVVRVVVVWUl5fTuXNnunTpUn2kaNKkSXTq1InS0lIuvPDCGvudOnXqXo/ojRs3jvLycrZt20ZFRQWnnnoq3bt3Z9CgQdVz22H+/Pmcd9551Y+feuopzj///F3aTJo0iTVr1tCvXz/69esHwIwZM+jSpQudO3dm7Nixe61v7ty5dOzYkd69e/Pb3/62evsnn3zCyJEj6dGjB926dWP27NkADBw4kA8++ICysjKeffZZ7r33Xnr06EHXrl0ZMmRIdTjc/ajfQQcdtMu4W7Zs4ac//SkzZ86krKyMmTNn1vg6SsoGw5kkSU3c6tWrOfLII6sft23bltWrV7Nt2zauu+46brnlll3at2vXjrFjx3LllVdy22230alTp13C2+5ef/11Zs6cyXPPPUdlZSXNmjVj+vTpVFZWsnr1apYvX84rr7zCpZdeCsDEiRNZunQpy5YtY/LkyXWezw033MAHH3zAlClTqKqq4uqrr+aRRx6hoqKCkSNH8pOf/GSX9qeddhqvv/4669atA2DKlCnVtexwzTXX0KZNGxYsWMCCBQtYs2YNY8eO5emnn6ayspIlS5Ywa9asXZ6zefNmRo0axWOPPcazzz7L2rVrq/fdfPPNnHbaaSxZsoQFCxZw/fXX88knnzBnzhzat29PZWUlffr04fzzz2fJkiW8/PLLHHfccfzmN7+p1WvQokULfv7zn3PBBRdQWVnJBRdcUOfXUVLDM5xJktTEbd++fY9tJSUl3HXXXZx55pm7BLcdLr/8cjZu3MjkyZO59dZba+x//vz5VFRU0KNHD8rKypg/fz5vv/02xxxzDG+//TZXX301c+fO5ZBDDgGgtLSU4cOHM23aNJo3b16nuYwfP54NGzZw9913U1JSQkSwfPlyBgwYQFlZGTfddBOrVq3aY64XX3wx06ZNY8OGDSxevJgzzjijxnGWLFlC3759Ofzww2nevDnDhw9n0aJFu7R54403OProozn22GMpKSnhoosuqt43b948Jk6cSFlZGX379mXz5s289957e4yzfPly+vTpQ5cuXZg+fTqvvvpqnV4PSY1L3f7HkyRJRadt27asXLmy+vGqVato06YNixcv5tlnn+Wuu+5i06ZNbNmyhYMOOoiJEyfy6aefVoecTZs2cfDBB++z/+3bt3PJJZcwYcKEPfa9/PLL/P73v+fOO+/koYce4v777+d3v/sdixYtYs6cOYwfP55XX3211iGtR48eVFRU8NFHH3HYYYexfft2jj/+eBYvXlzj8y699FLOOeccDjzwQIYOHfqF4+0t0O5NSUnJPp//6KOPklLaZfuKFSt2eVxeXs6sWbPo2rUrU6dOZeHChQA0b96cbdu2Vfe1ZcuWWtUjKdsMZ01MSul0YALQDFgHXAG0Bf4VeAnoCvwVKI+I1/LPuQT4J3JfL38GroqISCmVA/8I/AnoDGwAhkTEWvYhpXQrcCrQAlgPjIyId/P7zgZuBL4GbAMuiYhlKaVewC3Ajp/810dEzRc3SJJq7dxzz+WOO+7gwgsv5IUXXuCb3/wmrVu3Zvr06dVtpk6dyosvvlh9Z8SxY8cyfPhw2rVrx6hRo3j88cf32X///v0ZPHgwY8aMoVWrVnz00Uds3LiRli1b0qJFC4YMGUL79u2rrxFbuXIl/fr1o3fv3jz44INs2rSJb33rW7Way+mnn86gQYM466yzmDdvHikl1q1bx+LFi+nVqxdbt27lD3/4A8cff/wuz2vTpg1t2rThpptu4qmnntpr3wcffDAbN27k29/+Nj179uTaa69l/fr1HHroocyYMYOrr756l/YdO3bknXfe4a233qJ9+/bMmDGjet+gQYO4/fbbuf322ykpKWHp0qV069ZtjzE3btxI69at2bp1K9OnT+eII44AcnfCrKio4Pvf/z6zZ89m69at+6xXUuPhaY1NSEqpFfA/geERUQo8COz4yVsKTI2IE4A7gQf3A+V1AAAKEklEQVTyz+kDfB84JSK6kwtJ9+/UbQ/gRxFxPPAasOtPpj1NjIgeEdEVmAH8Ij/Od4H7gGH5fScB76SUDgP+F3BDfvsJwJKv8DJIUpMybNgwevXqRUTQtm3b6muWJk+eXH0915lnnskxxxxDhw4dGDVqFHfddVeNfT7zzDMsWbKkOqC1aNGCKVOm7LN9p06duOmmmxg4cCClpaUMGDCA999/n9WrV9O3b1/KysooLy9nwoQJVFVVcdFFF9GlSxe6devGmDFjah3Mdhg6dCijRo3i3HPPpaqqikceeYSxY8fStWtXysrKeP755/f6vOHDh3PkkUfSqVOnve7/wQ9+wBlnnEG/fv1o3bo1EyZMoF+/fnTt2pUTTjiBwYMH79L+wAMP5J577uGss86id+/etGvXrnrfuHHj2Lp1K6WlpXTu3Jlx48btdczx48fTs2dPBgwYQMeOHau3jxo1imeeeYYTTzyRF154gZYtW+7x3H79+vHaa695QxCpESmp7WF5NX4ppXOAayPi7/OPDwA+A74H/DIijt1t++HAfyd3dGzHH6opAQ6NiO/kj5x9LyLOzj/vMmBAROzz1loppYuBHwIHkT9yGxEdU0o/BLpHxMjd2p8FXBcRp33FuR8FvDN//nzatm37VbqSMuW5wUMKXYIy4OTZjxa6hKIwevRounXrxmWXXVboUtTErVq1iv79+wMcHRErClyOGpCnNTYtJUBd03gJcH9E/HQf+zfv9HkVNXxNpZTaAb8CekTEOyml/0Lu6N2OcfY1viRJ+1X37t1p2bIlt912W6FLkdSEeVpj07IYKEsp7Tgv4hJgKbAR6JA/hRFyR8peiYiPgceAESmltgAppWYppe5fcvxDgC3A2vzRuSt32vd74MyU0o6jd19PKR0MPA90yl93tmP8Q7/k+JIk7VVFRQWLFi3i61//eqFLkdSEGc6akIhYB1wMPJhSWgZclP8AqASGpZQqgGuAEfnnLAJ+AsxJKb0MLAcG7953Lcd/BXgYeBV4Gnhnp31vAqOAmflxFgNHRcRHwPnAL/M1VwBfNhxKkiRJmeU1ZyKl1Be4NSL+rtC17C9ec6Zi5TVnAq85k4qN15w1XR45kyRJkqQM8IYgIiIWAvV21CyldDkwei+7yiOisr7GkSRJkoqJ4Uz1LiLuI/c3yyRJkiTVkqc1SpIkSVIGGM4kSZIkKQMMZ5IkSZKUAYYzSZIkScoAw5kkSZIkZYDhTJIkSZIywHAmSZIkSRlgOJMkSZKkDDCcSZIkSVIGGM4kSZIkKQMMZ5IkSZKUAYYzSZIkScqA5oUuQJL05WzbsoWTZz9a6DKUAdu2bOGAFi0KXYYk6SvyyJkkNVK+GdcOfi1IUnEwnEmSJElSBhjOJEmSJCkDDGeSJEmSlAGGM0mSJEnKAMOZJEmSJGWA4UySJEmSMsBwJkmSJEkZYDiTJEmSpAwwnEmSJElSBhjOJEmSJCkDDGeSJEmSlAGGM0mSJEnKAMOZJEmSJGWA4UySJEmSMsBwJkmSJEkZYDiTJEmSpAwwnEmSJElSBhjOJEmSJCkDDGeSJEmSlAGGM0mSJEnKAMOZJEmSJGWA4UySJEmSMqB5oQuQGkgzgLVr1xa6DkmSpBrt9H6lWSHrUMMznKmpaA0wfPjwQtchSZJUW62BtwpdhBqO4UxNxRKgD/A+UFXgWiRJkmrSjFwwW1LoQtSwSrZv317oGiRJkiSpyfOGIJIkSZKUAYYzSZIkScoAw5kkSZIkZYDhTJIkSZIywHAmSZIkSRlgOJMkSZKkDDCcSZIkSVIGGM4kSZIkKQMMZ5IkSZKUAYYzSZIkScqA5oUuQFL9SSndCfQH/gJsAq6NiBf30XYcUJ5/ODUixjdIkftJSuki4AagE/BfI+KOfbTrCzwB/CG/6S8R0bNBitwPajvvfNtRwFigBHgSuCYitjVIoftBSulvgClAd+CvwI8i4vG9tOtLEax5Sum7wL8Bfwt8CIyIiDd3a9MMmAScDmwHJkbEfQ1da32r5dxvBP4JWJPf9FxE/LAh66xvKaVbgSHAUUCXiFi+lzbFuua1mfuNFNmaSx45k4rLk+R+iHUFJgAz99YopXQKMBTonP8Ymt/WmFUCFwIP1qLtaxFRlv9odG/Sd1OreaeUjgZ+BvQCjs1/XLTfq9u/fgRsjIgOwDnAfSmlg/bRthjWfDJwZ0R8F7gTuHsvbYYDHcitby/gxpTSUQ1W4f5Tm7kDPLDTOhfDm/RZwCnAuzW0KdY1r83cofjWXE2c4UwqIhHxeERszT9cDLRNKe3t+/wCcj/QPouIz4AH8tsarYhYHhGvAY32SNCXUYd5fw+YFRHr8kfL7qWRrzm5+icD5I+ivAicUdCK9pOUUivgBGBGftMM4ISU0uG7Nb0AuDcitkXEOnJvcIc2XKX1rw5zLzoR8b8jYuUXNCu6NYdaz10qOoYzqXiNBn63j9PWvsOuv418DziyQarKhu+mlF5KKb2QUrqk0MU0kGJc87rMqbGv+ZHA6oioAsj/u4Y951uM61zbuQNcmFJallKal1Lq1ZBFFlAxrnldNMU1VxHzmjOpEUkpvUTuB/He/Kcdb15SShcC/0julJCiUNu518JLwJER8ef8qX7/kVJaHRH/US+F1rN6nHej80Vzr0NXjWrN9aVNBm6OiK0ppQHA7JTScRHxYaEL037jmqvoGM6kRiQiTviiNiml84Cbgf4R8cd9NHsPaLfT4+8AmT59pDZzr2U/H+/0+TsppVnAyUAm36jX17wpwjVPKe2Y07r8pu8AC/bST6Na831YCRyRUmoWEVX5m0C0Yc813PGaLMk/3v2oSmNUq7lHxNqdPn8qpbSS3DW1zzRotQ2vGNe8VprwmquIeVqjVERSSmcDvwQGRcSKGpo+DIxIKX0jpfQNYATwUAOUWHAppdYppZL854cBA8ndVKPYPQr8Q0rp8Px1iKNo/Gv+MHAFQErpWKAHMHf3RsWw5hHxAbmah+U3DQOW5q8x2tnDwKiU0gH5a7L+gdzaN1q1nXtK6YidPi8jd5e/aKAyC6no1ry2mvCaq4iVbN++vdA1SKonKaV1wBY+P5IAuSNoH6aU7gPmRMScfNsbgYvJ3Vb9gYi4sYHLrVcppWHALcCh5F6DT4CBEfFaSunnwJqImJxSGg1cBWwld/bAAxHxL4Wq+6uq7bzzba8gd9t9gHnA6MZ8WmRKqSUwFegGVAE3RMTs/L6iW/OUUkdyt5M/FPgTudvJR0rpCeCnEfFi/qjSHeQCKMAvIuKewlRcf2o5938j92cVqsh9L/wsIp4oWNH1IKU0CTgf+M/AeuDDiDi+iax5beZedGsuGc4kSZIkKQM8rVGSJEmSMsBwJkmSJEkZYDiTJEmSpAwwnEmSJElSBhjOJEmSJCkDDGeSJEmSlAGGM0mSJEnKAMOZJEmSJGXA/wfRW4jV+NwGAAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lmp.local_logistic_plot(loan_sample,\n",
    "                        feature_names=ln_X_test.columns,\n",
    "                        logistic_model=lr,\n",
    "                        classes_label=[\"not default\", \"default\"]);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Conclusions <a name=lrc>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Like linear models, logistic regression is pretty intuitive and easy to interpret, in industries where interpretability is required by law, it is being used as a standard for explainability."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:iml-all]",
   "language": "python",
   "name": "conda-env-iml-all-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}