{
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "M9KR7pRVFucG"
},
"source": [
" "
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "pOGWpDlgRp6l"
},
"source": [
"# Introduction to binary classification\n",
"\n",
"The data we're going to be working with for this example is the `oasis` data from Elizabeth Sweeney's R package. The data contain MR (magnetic resonance) images for lesion segmentation in multiple sclerosis (MS). MS is a disorder primarily caused by whtie matter lesions. This dataset is a collection of voxels from an image with radiologists labeling of whether or not a white matter lesion exists at that location. "
]
},
{
"cell_type": "code",
"execution_count": 26,
"metadata": {
"id": "G87LkI7xOQbh"
},
"outputs": [],
"source": [
"import numpy as np\n",
"import pandas as pd\n",
"import seaborn as sns\n",
"import matplotlib.pyplot as plt\n",
"## this sets some style parameters\n",
"sns.set()\n",
"from sklearn.metrics import accuracy_score, roc_curve, auc\n"
]
},
{
"cell_type": "code",
"execution_count": 27,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 187
},
"id": "sn3GZp8USNmd",
"outputId": "11bfa671-9698-4814-d6ff-083f448edc93"
},
"outputs": [
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
"
\n",
"
\n",
"
FLAIR
\n",
"
PD
\n",
"
T1
\n",
"
T2
\n",
"
FLAIR_10
\n",
"
PD_10
\n",
"
T1_10
\n",
"
T2_10
\n",
"
FLAIR_20
\n",
"
PD_20
\n",
"
T1_20
\n",
"
T2_20
\n",
"
GOLD_Lesions
\n",
"
\n",
" \n",
" \n",
"
\n",
"
0
\n",
"
1.143692
\n",
"
1.586219
\n",
"
-0.799859
\n",
"
1.634467
\n",
"
0.437568
\n",
"
0.823800
\n",
"
-0.002059
\n",
"
0.573663
\n",
"
0.279832
\n",
"
0.548341
\n",
"
0.219136
\n",
"
0.298662
\n",
"
0
\n",
"
\n",
"
\n",
"
1
\n",
"
1.652552
\n",
"
1.766672
\n",
"
-1.250992
\n",
"
0.921230
\n",
"
0.663037
\n",
"
0.880250
\n",
"
-0.422060
\n",
"
0.542597
\n",
"
0.422182
\n",
"
0.549711
\n",
"
0.061573
\n",
"
0.280972
\n",
"
0
\n",
"
\n",
"
\n",
"
2
\n",
"
1.036099
\n",
"
0.262042
\n",
"
-0.858565
\n",
"
-0.058211
\n",
"
-0.044280
\n",
"
-0.308569
\n",
"
0.014766
\n",
"
-0.256075
\n",
"
-0.136532
\n",
"
-0.350905
\n",
"
0.020673
\n",
"
-0.259914
\n",
"
0
\n",
"
\n",
"
\n",
"
3
\n",
"
1.037692
\n",
"
0.011104
\n",
"
-1.228796
\n",
"
-0.470222
\n",
"
-0.013971
\n",
"
-0.000498
\n",
"
-0.395575
\n",
"
-0.221900
\n",
"
0.000807
\n",
"
-0.003085
\n",
"
-0.193249
\n",
"
-0.139284
\n",
"
0
\n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" FLAIR PD T1 T2 FLAIR_10 PD_10 T1_10 \\\n",
"0 1.143692 1.586219 -0.799859 1.634467 0.437568 0.823800 -0.002059 \n",
"1 1.652552 1.766672 -1.250992 0.921230 0.663037 0.880250 -0.422060 \n",
"2 1.036099 0.262042 -0.858565 -0.058211 -0.044280 -0.308569 0.014766 \n",
"3 1.037692 0.011104 -1.228796 -0.470222 -0.013971 -0.000498 -0.395575 \n",
"\n",
" T2_10 FLAIR_20 PD_20 T1_20 T2_20 GOLD_Lesions \n",
"0 0.573663 0.279832 0.548341 0.219136 0.298662 0 \n",
"1 0.542597 0.422182 0.549711 0.061573 0.280972 0 \n",
"2 -0.256075 -0.136532 -0.350905 0.020673 -0.259914 0 \n",
"3 -0.221900 0.000807 -0.003085 -0.193249 -0.139284 0 "
]
},
"execution_count": 27,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"dat = pd.read_csv(\"https://raw.githubusercontent.com/bcaffo/ds4bme_intro/master/data/oasis.csv\")\n",
"dat.head(4)"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "Gb7bSjRHSJW9"
},
"source": [
"Note that we loaded `pandas` first. The various columns are voxel values from different kinds of MR images of the same patient. FLAIR (fluid attenuated inversion recovery), PD (proton density), T1 and T2. The latter two aren't acronyms, but instead named for the specific part of the *relaxation time* of the MR signal. Roughly, the relaxation time is related to the signal produced by protons snapping back into alignment in a strong magnetic field (recall *magnetic* resonance imaging). The `_10` and `_20` ending variables are local averages of the neighboring voxels. We're trying to predict `GOLD_Lesions`, which is the radiologist standard of whether or not there is a lesion at this voxel. (A voxel is a three dimensional pixel.) \n",
"\n",
"Note here we are doing *voxelwise segmentation* that is trying to predict whether there is a lesion at each specific voxel. This can be viewed as an image processing problem. Other classification problem consider, for example, whether a patient has any lesions (and then where as a followup). Approaching the problem that way is a *image level* segmentation approach. \n",
"\n",
"Let's plot it. I'm showing a couple of ways. I've been testing out plotting libraries in python, and I think that I like 'seaborn' (the second plot) the best. In the seaborn plots, I show both the marginal plot (without considering the gold standard) and then stratified by whether or not there was a lesion at that voxel."
]
},
{
"cell_type": "code",
"execution_count": 28,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 338
},
"id": "3jlT2wVkXDNy",
"outputId": "ba38743a-f439-4976-b112-ec2754a49a08"
},
"outputs": [
{
"data": {
"text/plain": [
"GOLD_Lesions\n",
"0 Axes(0.125,0.11;0.775x0.77)\n",
"1 Axes(0.125,0.11;0.775x0.77)\n",
"Name: FLAIR, dtype: object"
]
},
"execution_count": 28,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
"
"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"dat.groupby('GOLD_Lesions').FLAIR.hist(alpha= .5)"
]
},
{
"cell_type": "code",
"execution_count": 29,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 556
},
"id": "XbdVFgByj_mo",
"outputId": "45a2e3ca-3747-4236-bd91-e0b5b3b042ac"
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/bcaffo/miniconda3/envs/ds4bio_new/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
" with pd.option_context('mode.use_inf_as_na', True):\n"
]
},
{
"data": {
"image/png": "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",
"text/plain": [
"
"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/bcaffo/miniconda3/envs/ds4bio_new/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
" with pd.option_context('mode.use_inf_as_na', True):\n",
"/home/bcaffo/miniconda3/envs/ds4bio_new/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
" with pd.option_context('mode.use_inf_as_na', True):\n"
]
},
{
"data": {
"image/png": "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",
"text/plain": [
"
"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"\n",
"x0 = dat.FLAIR[dat.GOLD_Lesions == 0]\n",
"x1 = dat.FLAIR[dat.GOLD_Lesions == 1]\n",
"x2 = dat.FLAIR\n",
"\n",
"\n",
"sns.kdeplot(x2, fill = True, label = 'Marginal')\n",
"\n",
"plt.show()\n",
"\n",
"sns.kdeplot(x0, fill = True, label = 'Gold Std = 0')\n",
"sns.kdeplot(x1, fill = True, label = 'Gold Std = 1')\n",
"\n",
"plt.show()\n",
"\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "R8hBTU2Unxes"
},
"source": [
"# Classification\n",
"\n",
"Let's try creating the simplest possible classifier, a threshold. So here we want to pick the value of the threshold so that lower values are classified `GOLD_Lesion == 0` (i.e. no lesion) and higher values are `GOLD_Lesion == 1` (lesion at this voxel). We want to do this on labeled voxels so that we can pick a meaningful threshold on voxels without a gold standard labeling. That is, for new patients we want to automatically label their images one voxel at a time with a simple thresholding rule. We're going to use our **training data** where we know the truth to develop the threshold. \n",
"\n",
"*Note the idea behind doing this is only useful if the new images without the gold standard are in the same units as the old one, which is not usually the case for MRIs. The technique for trying to make images comparable is called normalization*. \n",
"\n",
"Let's first just try each of the datapoints itself as a threshold and pick which one does best.\n",
"However, I'm going to break the data into a training and testing set. The reason for this is that I want to make sure that I don't overfit. That is, we're going to test our algorithm on a dataset that wasn't used to train the algorithm. \n"
]
},
{
"cell_type": "code",
"execution_count": 30,
"metadata": {
"id": "Qykc4u2KMrWv"
},
"outputs": [],
"source": [
"x = dat.FLAIR\n",
"y = dat.GOLD_Lesions\n",
"n = len(x)\n",
"trainFraction = .75\n",
"\n",
"## Build a training and testing set\n",
"## Prob of being in the train set is trainFraction\n",
"sample = np.random.uniform(size = n) < trainFraction\n",
"\n",
"## Get the training and testing sets\n",
"xtrain = x[ sample]\n",
"ytrain = y[ sample]\n",
"xtest = x[~sample]\n",
"ytest = y[~sample]\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 31,
"metadata": {
"id": "FkxeXhYgbBou"
},
"outputs": [],
"source": [
"## Starting values, just set it to \n",
"## 0 so that it improves on the first\n",
"## try\n",
"bestAccuracySoFar = 0\n",
"\n",
"for t in np.sort(xtrain):\n",
" ## Strictly greater than the threshold is\n",
" ## our algorithm\n",
" predictions = (xtrain > t)\n",
" accuracy = np.mean(ytrain == predictions)\n",
" if (accuracy > bestAccuracySoFar):\n",
" bestThresholdSoFar = t \n",
" bestAccuracySoFar = accuracy \n",
"\n",
"threshold = bestThresholdSoFar\n",
" "
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "DuHaFEoLd_VQ"
},
"source": [
"Now let's test our our \"algorithm\", on the test set. We'll look at the test set accuracy, but also how it breaks up into the sensisitivity and specificity.\n",
"\n",
"## Definitions\n",
"test set **accuracy** = proportion of correct classifications on the test data\n",
"\n",
"test set **sensitivity** = proportion declared diseased among those that are actually diseased. (In this case lesion = disease)\n",
"\n",
"test set **specificity** = proportion declared not diseased among those that are actually not diseased.\n",
"\n",
"To interpret the sensitivity and specificity, imagine setting the threshold nearly to zero. Then we'll declare almost every voxel a lesion and we'll have nearly 100% sensitivity and nearly 0% specificity. If we declare a voxel as a lesion it's not that interesting. If we declare a voxel as not lesions, then it's probably not a lesion.\n",
"\n",
"If we set the threshold really high, then we'll have nearly 0% sensitivity and 100% specificity. If we say a voxel is not lesioned, it's not that informative, since we declare nearly everything not a lesion. But if we declare a voxel a lesion, it usually is.\n",
"\n",
"\n",
"So, if you have a high sensitivity, it's good for ruling diseases out. If you have a high specificity it's good for ruling diseases in. If you have a high both? Then you have a very good test.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 32,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"
\n",
"\n",
"
\n",
" \n",
"
\n",
"
\n",
"
Threshold
\n",
"
Accuracy
\n",
"
Specificity
\n",
"
Sensitivity
\n",
"
\n",
" \n",
" \n",
"
\n",
"
0
\n",
"
1.4632
\n",
"
0.705882
\n",
"
0.833333
\n",
"
0.636364
\n",
"
\n",
" \n",
"
\n",
"
"
],
"text/plain": [
" Threshold Accuracy Specificity Sensitivity\n",
"0 1.4632 0.705882 0.833333 0.636364"
]
},
"execution_count": 32,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"## Let's test it out on the test set\n",
"testPredictions = (xtest > threshold)\n",
"\n",
"## The test set accuracy\n",
"testAccuracy = np.mean(testPredictions == ytest)\n",
"\n",
"## Let's see how it specifically does on the\n",
"## set of instances where ytest == 0 and ytest == 1\n",
"## The % it gets correct on ytest == 0 is called\n",
"## the specificity and the percent correct when \n",
"## ytest == 1 is called the sensitivity.\n",
"sub0 = ytest == 0\n",
"sub1 = ytest == 1\n",
"\n",
"testSpec = np.mean(ytest[sub0] == testPredictions[sub0])\n",
"testSens = np.mean(ytest[sub1] == testPredictions[sub1])\n",
"\n",
"pd.DataFrame({\n",
" 'Threshold': threshold,\n",
" 'Accuracy': testAccuracy, \n",
" 'Specificity': testSpec, \n",
" 'Sensitivity': testSens}, index = [0])\n",
"\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 33,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 285
},
"id": "j2GagXHug8T3",
"outputId": "0946a96e-90b1-4518-f795-d764e9cbed49"
},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/bcaffo/miniconda3/envs/ds4bio_new/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
" with pd.option_context('mode.use_inf_as_na', True):\n",
"/home/bcaffo/miniconda3/envs/ds4bio_new/lib/python3.10/site-packages/seaborn/_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.\n",
" with pd.option_context('mode.use_inf_as_na', True):\n"
]
},
{
"data": {
"image/png": "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",
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"sns.kdeplot(x0, fill = True, label = 'Gold Std = 0')\n",
"sns.kdeplot(x1, fill = True, label = 'Gold Std = 1')\n",
"plt.axvline(x=threshold)\n",
" \n",
"plt.show()\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "qZH4t-52g8nz"
},
"source": [
"OK, so out plot has better sensitivity than specificity and a test set accuracy of around 68%. The lower specificity is because there's a lower percentage of blue below the line than orange above the line. Recall, we're saying above the threshold is a lesion and orange is the distribution for voxels with lesions.\n",
"\n",
"So, for this algorithm, the high sensitivity says that all else being equal, if you declare a voxel as not being a lesion, it probably isn't. In other words, if you're out in the lower part of the orange distribution, there's a lot of blue there.\n",
"\n",
"However, all else isn't equal. Most voxels aren't lesions. This factors into our discussion in a way that we'll discuss later.\n",
"\n"
]
},
{
"cell_type": "code",
"execution_count": 34,
"metadata": {
"id": "W4cTGDUnvWOH"
},
"outputs": [],
"source": [
"fpr, tpr, thresholds = roc_curve(ytest, xtest)\n",
"roc_auc = auc(fpr, tpr)"
]
},
{
"cell_type": "code",
"execution_count": 35,
"metadata": {
"colab": {
"base_uri": "https://localhost:8080/",
"height": 301
},
"id": "Qq-AIu3owmqa",
"outputId": "238aa73e-9beb-4290-b2c2-5fcf8f9fb511"
},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAj4AAAHKCAYAAAAD2UE+AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAACEq0lEQVR4nO3dd1gUZ9cG8HuXLrAUxYaCioIFa1RAURS7YOwlEVtU7NhjjSX2GkUsiMQWo8befWPvRBM1ttiwN1Ta0svufH/wsbpSZIFlKPfvunIZzs7Mnj0s7OF5npmRCIIggIiIiKgIkIqdABEREVFeYeNDRERERQYbHyIiIioy2PgQERFRkcHGh4iIiIoMNj5ERERUZLDxISIioiKDjQ8REREVGWx8iIiIqMhg40Ma2bt3LxwcHFT/Va9eHa6urhg7diyePXsmdnoAAHd3d0yePFnsNAqVbdu2Ye/evWnir169goODQ7qP5Qep+QUGBoqdisbi4uKwatUq/PXXX1o5/l9//QUHBweNj19Q3wv5kYODA1atWiV2GkWOrtgJUMG0YMECVKpUCQkJCbh+/TrWrVuHv/76C8eOHYOZmZmoufn5+cHExETUHAqb7du3w8LCAl26dFGLlyxZEjt37oSNjY1ImRVecXFx8PPzw8iRI+Hk5JTrx69RowZ27tyJypUra7Qf3wtU0LHxoWypUqUKatasCQBwcnKCQqHAqlWrcPLkSXTt2lXU3KpXr57nz6lQKKBQKKCvr5/nz60pQRCQkJAAQ0PDHB9LX18fderUyXlSBVx8fDwMDAwgkUjETuWrkpKSIJFIYGJikqvfO74XqKDgVBflitQmKDQ0VC1++/ZtDB06FA0bNkTNmjXRqVMnHD16NM3+ISEh+Omnn+Dm5gZHR0e4urrCx8cHHz9+VG0THR2NRYsWwd3dHY6OjmjSpAnmzZuH2NhYtWN9PtUVFhYGR0dHrFixIs1zBgcHw8HBAVu2bFHFPnz4gBkzZqBp06ZwdHSEu7s7/Pz8kJycrNomdUg/ICAAa9asgbu7O2rWrImgoKAM65OQkIBly5ap5T579mzI5fI0uQ8ZMgQnTpxAhw4dULNmTbRo0UItR03r4eDggJ9//hnbt29Hu3btULNmTezbtw9AyuhY9+7d0bBhQ9SrVw+dO3fGrl278Pm9i93d3fHo0SNcvXpVNcXp7u6uVovPpzdWrVoFBwcHPHr0COPGjcM333yDRo0aYcqUKYiKilLLTS6XY+rUqWjYsCHq1q0Lb29vvHz5MstTAHK5HAsXLkSLFi3g6OgIFxcXDB48GMHBwWm23bhxI9zd3VG3bl307NkTN2/eVHv89u3bGDt2LNzd3VGrVi24u7tj3LhxeP36tdp2qdO9Fy9exJQpU+Ds7IzatWsjMTERz58/x5QpU9C6dWvUrl0bTZo0wdChQ/HgwQONcn/16hVcXFxU36PUun8+hfvs2TOMHz8eLi4ucHR0RLt27bBt2za150idztq/fz8WLlyIJk2aoGbNmnj+/Hm6U10vX77E2LFj4erqCkdHRzRq1Aj9+vXDf//9B0Dz9wKQ8nM2btw4NGrUCI6OjmjWrBl+/PFHJCYmZvq9TUxMxJo1a9C2bVs4OjrC2dkZU6ZMQVhYmGqb9evXo2rVqjh9+rTavpMnT0bt2rVVdU9ISMDChQvRsWNHfPPNN2jYsCF69uyJkydPpnne1J+XPXv2oE2bNqhVqxa6dOmCmzdvQhAEbNiwQfU+6tu3L54/f662f58+feDp6Ym///4bPXr0QK1atdCkSROsWLECCoUi09cMZO13EOUMR3woV7x69QoAUKFCBVUsKCgIgwYNQu3atTFr1iyYmpri6NGjGDt2LOLj41VD5SEhIejatSuSk5MxdOhQODg4IDw8HBcvXkRkZCRKlCiBuLg4eHl54d27d6ptHj16BF9fXzx8+BCbNm1K969tS0tLNGvWDPv374ePjw+k0k+9/t69e6Gnp4cOHToASPmF0717d0ilUowYMQI2Nja4ceMG1q5di9evX2PBggVqx966dSsqVKiASZMmwcTEBLa2tunWRhAEDB8+HEFBQfD29kb9+vXx4MEDrFq1Cjdv3sTOnTvVRor+++8/zJ8/HyNHjkSJEiVw6NAhzJs3D0lJSRg4cCAAaFyPkydP4u+//8aIESNQokQJFC9eHADw+vVr9OzZE2XLlgUA3Lx5E3PnzkVISAhGjhwJIOWD18fHB6amppg5cyYAZGlka9SoUWjfvj26deuGhw8fYtmyZQCgqqNSqcTQoUNx584djBw5EjVq1MCNGzcwaNCgrx4bSGn8vv/+e7x+/Vr1PouNjcW1a9fw4cMH2NnZqbbdtm0bKlWqhKlTpwIAVq5cCW9vb5w6dQqmpqaqWlSsWBEeHh4wMzPDhw8fsH37dnTr1g1HjhyBpaWl2vNPnToVzZo1w+LFixEXFwddXV28f/8e5ubmGD9+PCwtLREZGYl9+/ahR48e2LdvHypVqpSl3OvVq4cNGzZg0KBB6NatG7p37w4AqhweP36MXr16oUyZMpg0aRKsrKxw8eJFzJ07F+Hh4arvXarly5ejTp06mD17NqRSKYoXL672R0WqwYMHQ6lUYuLEiShbtizCw8Nx48YNVYOu6Xvh/v37+O6772BhYQEfHx/Y2triw4cPOH36NBITEzPcV6lUYvjw4fjnn38wcOBA1KtXD69fv8aqVatw69Yt7NmzB4aGhhg8eDD+/vtvTJ48Gfv27YO1tTX27NmDffv2Ye7cuXBwcACQ0kRFRkbihx9+QKlSpZCUlITLly9j1KhRWLBgATp16qT2/GfPnsW9e/cwYcIESCQSLFmyBEOGDEGnTp3w8uVLzJgxA1FRUVi4cCFGjRqFAwcOqP28ffjwAWPHjoW3tzd8fHxw9uxZrF27FnK5HDNmzMiwXpr+DqJsEog0sGfPHsHe3l64efOmkJSUJERHRwvnz58XGjduLPTu3VtISkpSbdu2bVuhU6dOajFBEIQhQ4YIjRs3FhQKhSAIgjBlyhShRo0awuPHjzN8Xn9/f6Fq1arCrVu31OLHjx8X7O3thbNnz6pizZs3FyZNmqT6+tSpU4K9vb1w8eJFVSw5OVlwdXUVRo0apYr99NNPQp06dYTXr1+rPUdgYKBgb28vPHr0SBAEQXj58qVgb28vtGzZUkhMTPxqzc6fPy/Y29sLAQEBavEjR44I9vb2ws6dO9Vyd3BwEP777z+1bQcMGCDUq1dPiI2N1bge9vb2wjfffCNERERkmqdCoRCSkpIEPz8/oWHDhoJSqVQ95uHhIXh5eaXZJ7UWe/bsUcV8fX3Tfb2zZs0SatasqTru2bNnBXt7e+H3339X287f31+wt7cXfH19M83Xz89PsLe3Fy5dupThNqn5eXp6CsnJyar4v//+K9jb2wuHDx/OcN/k5GQhJiZGqFOnjrB582ZVPPVn4Mcff8w0v9RjJCYmCq1btxbmz5+vUe6hoaEZ1uGHH34QmjZtKkRFRanFf/75Z6FmzZqq73VQUJBgb28v9O7dO80xUh8LCgoSBEEQwsLCBHt7e2HTpk2ZviZN3gt9+/YV6tevL4SGhmZ6zC8dPnxYsLe3F/73v/+pxW/duiXY29sL27ZtU8XCwsKEpk2bCt26dRPu3r0r1K5dW5gwYUKmx09OThaSkpKEqVOnCp06dVJ7zN7eXmjcuLEQExOjip04cUKwt7cXOnbsqPZzsWnTJsHe3l64f/++Kubl5SXY29sLJ0+eVDvu9OnThapVq6r9fvny+5vV30GUM5zqomzp0aMHatSogXr16mHQoEGQyWRYs2YNdHVTBhGfP3+OJ0+eqEZTkpOTVf81bdoUHz58wNOnTwEA58+fh5OTk9pf6F86c+YMqlSpgmrVqqkdy9XVFRKJBFevXs1w36ZNm8LKykptCP7ixYt4//692nqks2fPwsnJCSVLlkyTL4A0z+Hu7g49Pb2v1ip1CuzLxaDt2rVDsWLFcOXKFbV4lSpVULVqVbWYp6cnoqOjcffu3WzVw9nZOd1F51euXEH//v3xzTffoFq1aqhRowZ8fX0RERGRZtpSU6lTIKkcHByQkJCgOm5qju3atVPbzsPDI0vHv3DhAipUqIBGjRp9ddtmzZpBR0dH9XVqfT+fxoqJicGSJUvQqlUrVK9eHdWrV0fdunURGxub7tRZ69at08SSk5Oxbt06tG/fHo6OjqhevTocHR3x7NkztWNokvuXEhISEBQUhFatWsHQ0DDNezUhISHNNF56uX7J3NwcNjY2CAwMxMaNG3Hv3j0olUqN80sVFxeHa9euoV27dmlGy77mzJkzkMlkaN68udrrq1atGqysrNTe3xYWFvjll19w79491SjY7Nmz0xzz2LFj6NWrF+rWrYvq1aujRo0a2L17d7rfWycnJxQrVkz1dervpqZNm6qN7KTG37x5o7a/sbExWrRooRbz9PSEUqnEtWvXMnzdmv4OouzhVBdly6JFi2BnZ4eYmBgcPXoUO3fuxLhx47BhwwYAUA2jL1q0CIsWLUr3GOHh4ap/S5UqlenzhYaG4vnz56hRo0amx0qPrq4uvv32W/z222+Qy+WQyWTYu3cvrKys4OrqqvYcZ86cyfJzWFlZZZpzqoiICOjq6qb55S+RSFCiRAlERESoxUuUKJHmGKmx1G01rUd6ud66dQsDBw5Ew4YNMWfOHJQuXRp6eno4efIk1q1bh/j4+Cy9voyYm5urfZ06rZF63NS6fLldeq8/PWFhYShTpkyOcklISFDFxo8fj6CgIAwfPhw1a9aEsbExJBIJvL291bZLlV5NFy5ciG3btmHw4MFo0KABzMzMIJFIMH36dLVjaJL7lyIiIpCcnIytW7di69at6W6TnfeqRCLBpk2bsHr1amzYsAELFy6Eubk5OnTogDFjxmh8pqRcLodCofjqz3Z6QkNDIZfL4ejomO7jX76+2rVro3Llyqqptc+bFgD4888/MWbMGLRt2xaDBg1CiRIloKOjg+3bt2PPnj1pjv/lHwmpf+BkFP/y/ZGVn+H0aPo7iLKHjQ9li52dnWpBs7OzM5RKJXbt2oXjx4+jbdu2sLCwAAAMGTIErVq1SvcYFStWBJDyF1tISEimz2dhYQEDAwPMnz8/w8cz07VrVwQGBuLIkSNo3749Tp8+jX79+qmNAlhYWMDBwQFjxoxJ9xglS5ZU+zqrZ/CYm5sjOTkZYWFhas2PIAj4+PGjqo6p0lt7kRpL/QDXtB7p5XrkyBHo6urC398fBgYGqnh6Cz61IbUuERERao3Jhw8fsrS/paUl3r17lyu5REVF4ezZsxg5ciS8vb1V8dS1IelJr6YHDx5Ep06dMG7cOLV4eHg4ZDJZruQuk8mgo6ODjh074vvvv093m3Llyn011/RYW1ur3lNPnz7FsWPH4Ofnh8TERPz8888a5WlmZgYdHZ2v/mynx8LCAubm5qo/pL5kbGys9nXq2rbUEcvmzZujfPnyqscPHjyIcuXKYcWKFWq12Lx5s8a5ZUVWfobTo+nvIMoeNj6UKyZOnIg///wTvr6+aN26NSpVqoQKFSrg/v37aT4EvtS0aVMcPHgQT548US3+/FKzZs3g7+8Pc3NztV9oWWVnZ4fatWtj7969UCqVSExMTDP11KxZM5w7dw42Nja5ei0iFxcXbNiwAQcPHkT//v1V8f/973+IjY1Vnb2T6tGjR7h//77adNfhw4dhbGys+kswp/UAUj4MdXR01BZ8x8fH4+DBg2m21dfXz/EI0JcaNmyIDRs24OjRo2of4Omd9ZeeJk2awNfXF1euXElTQ01JJBIIgpBmse2uXbuydCbO58f5cvrz7NmzCAkJUVv8npXcvxwhS2VkZAQnJyfcu3cPDg4OWruEQsWKFTF8+HD8+eefuHfvnlpeWXkvGBoaokGDBjh+/DjGjBmj0XRXs2bNcOTIESiVStSuXTvTbS9duoT169dj2LBh6NevHzp16oQxY8Zg+/btqtqkfl++XIB86tSpLOekiZiYGJw6dUptuuvw4cOQSqVo0KBBhvtp63cQqWPjQ7nCzMwM3t7eWLJkCQ4dOoSOHTti9uzZGDx4MAYOHIjOnTujVKlSiIyMRHBwMO7evQtfX18AwOjRo3H+/Hl4eXlhyJAhsLe3R1RUFC5cuID+/fvDzs4O/fr1w59//gkvLy/0798fDg4OUCqVePv2LS5evIgffvjhq78gu3btihkzZuD9+/eoW7dumibLx8cHly9fRq9evdCnTx9UrFgRiYmJePXqFc6fP4/Zs2ejdOnSGtemcePGcHV1xdKlSxEdHY169erhwYMH8PX1RfXq1dGxY0e17UuWLIlhw4Zh5MiRsLKywsGDB3Hp0iVMmDABRkZGAJAr9XBzc8PGjRsxfvx49OzZExEREQgMDEz3g9Te3h5HjhzB0aNHUa5cORgYGKjOmMmuJk2aoF69eli0aBGio6Ph6OiIGzdu4MCBAwC+PkrRr18/HDt2DMOHD4e3tzdq1aqF+Ph4XLt2Dc2aNYOzs3OWczExMUGDBg0QGBgICwsLWFtb4+rVq9i9e7faSM3XNGvWTHX2loODA+7evYvAwMA075us5G5iYgJra2ucOnUKLi4uMDMzg4WFBcqVK4dp06bh+++/R+/evfHdd9/B2toaMTExePHiBU6fPp3u5Q++5v79+5gzZw7atm0LW1tb6OnpISgoCA8ePFAbBdPkvTBlyhR899136NGjB7y9vWFjY4PQ0FCcPn0as2fPznD6zMPDA4cOHYK3tzf69OmDWrVqQU9PD+/evcNff/2FFi1aoFWrVnj//j0mTpyIBg0aYOTIkZBKpfjll1/g5eWFJUuWYNq0aarvy59//olZs2ahTZs2ePfuHdasWYOSJUtq5Yrz5ubmmDVrFt6+fYsKFSrg3Llz+OOPP/Ddd9+pzqBMj7Z+B5E6Nj6Ua/r06YNt27ZhzZo18PT0hLOzM3bt2oV169Zh/vz5kMvlMDc3h52dndqC1lKlSmH37t3w9fVFQEAAIiIiYGFhgW+++UY1LFysWDFs27YN69evx86dO/Hq1SsYGhqiTJkyaNSoEaytrb+an4eHB+bPn493796lOd0XSGk4du/ejTVr1iAwMBAhISEwNjaGtbU1mjRpotEH4OckEgnWrFmDVatWYe/evVi3bh3Mzc3RsWNHjBs3Lk2jUa1aNXTp0gWrVq3Cs2fPULJkSUyZMkVttCg36uHi4oL58+cjICAAQ4cORalSpdCjRw9YWlqqPjBSjRo1Ch8+fMD06dMRExMDa2vrNNdO0ZRUKsW6deuwcOFCBAQEICkpCfXq1cOSJUvQo0ePr9bbxMQEv//+O1atWoU//vgDq1evhkwmQ82aNdGjRw+N81m2bBnmzZuHJUuWIDk5GfXq1cPGjRsxZMiQLB9j2rRp0NXVxfr16xEbG4vq1atj1apVWLlyZbZynzdvHhYvXoxhw4YhMTERnTt3xsKFC1G5cmXs3bsXa9aswYoVKxAWFgZTU1PY2trCzc1N49cOpKwDsrGxwe+//66ahitfvjwmTZqEPn36qLbT5L1QtWpV1c/2smXLEBMTAysrKzg7O2c6UqWjo4O1a9diy5YtOHDgANavXw8dHR2ULl0aDRo0gL29PRQKBcaPHw+JRIJly5apRi7r1KmDsWPHYvHixXByckLLli3RtWtXhIaGYseOHdizZw/Kly8Pb29vvHv3Dn5+ftmqV2asrKwwY8YMLFq0CA8fPoSZmRmGDh2KUaNGZbqftn4HkTqJIHx2pTIiEpW7uzuqVKkCf39/sVMRzaFDhzBhwgRs374d9erVEzsdIo306dMH4eHhOHz4sNipUAY44kNEojl8+DBCQkJgb28PqVSKmzdvIjAwEA0aNGDTQ0RawcaHiERjbGyMI0eOYO3atYiLi4OVlRW6dOmS4VktREQ5xakuIiIiKjJ45WYiIiIqMtj4EBERUZHBxoeIiIiKDDY+REREVGTwrK4vCIIApVI7672lUonWjk2fsM55g3XOG6xz3mCd8442ai2VSrJ8Tzo2Pl9QKgWEhcXk+nF1daWwsDCGXB6L5GRlrh+fUrDOeYN1zhusc95gnfOOtmptaWkMHZ2sNT6c6iIiIqIig40PERERFRlsfIiIiKjIYONDRERERQYbHyIiIioy2PgQERFRkcHGh4iIiIoMNj5ERERUZLDxISIioiJD9Mbn+fPnmDFjBjp27Ijq1avD09Mzy/vu27cPbdu2Rc2aNeHp6Yljx45pMVMiIiIq6ERvfB49eoRz587B1tYWdnZ2Wd7v+PHjmDx5Mlq1aoWAgAA4Oztj7NixuHjxohazJSIiooJM9Ht1ubu7o2XLlgCAyZMn486dO1nab+XKlWjbti3Gjx8PAHB2dsbTp0/h6+sLV1dXreVLREREBZfoIz5SqeYpvHz5Ek+ePEkzLebp6Ylbt24hLCwst9IjIiKiQkT0EZ/sePLkCQCgUqVKanE7OzsIgoAnT57A0tIy28fX1c39flBHR6r2L2kH65w3WOe8wTrnDdZZ+xQKJTZt+hf9Gj8FnqyCQY0xgG1HUXIpkI1PZGQkAEAmk6nFzczM1B7PDqlUAgsL4+wn9xUymZHWjk2fsM55g3XOG6xz3mCdtSMkJBpeXvtx8uQT3HN9gl87XUOxxDkoVud7UfIpkI1PKolEova1IAjpxjWhVAqQy2NzlFd6dHSkkMmMIJfHQaFQ5vrxKQXrnDdY57zBOucN1lm7rl17hdOnnwIANl+qgLFOJVHDWI7I8Jhcew6ZzCjLI3YFsvH5fGSnRIkSqrhcLgeQdiRIU8nJ2nvjKxRKrR6fUrDOeYN1zhusc95gnbXDycka48c7Y/PmW9j+/Q7ULPMeSqGsaLUukBOaqWt7Utf6pAoODoZEIkmz9oeIiIjyxsePsVAqBbXYuHFOOHeuL5rZvxUpq08KZONTvnx5VKpUCUePHlWLHz58GLVq1crRwmYiIiLKnjNnnqFp081Yu/YftbiOjhTFi+ePNVSiT3XFxcXh3LlzAIDXr18jOjoax48fBwA0bNgQlpaWmDp1Kvbv34979+6p9vPx8cHYsWNhY2ODRo0a4dSpU7h06RI2bNggyusgIiIqqpKTlVi06DJWrrwKAJg37yKcnMqifv2yImeWluiNT2hoKEaPHq0WS/16y5YtcHJyglKphEKhUNumXbt2iI+Px7p16xAYGAhbW1v88ssvvHghERFRHnrzJgpDhhzFX3+9VsXc3GxQsaKFiFllTCKkngpFAFIWt4WF5d5K81S6ulJYWBgjPDyGi+e0iHXOG6xz3mCd8wbrnH0nTjzBqFHHERYWDyClllOnNsbw4fUhlaY9w7r47qqQxr6BslhZhHa7n2t5WFoaF+6zuoiIiEg8SUkKzJt3EWvWfFrLU66cKfz9PdCgQf6b3vocGx8iIiLKsnfvojFgwEH88887VaxtWzusXNkaFhb5YwFzZtj4EBERUZaZmOgjIiIBAKCnJ8WMGU3h7V03RxcPzksF8nR2IiIiEoeJiT4CAjxhb2+Jw4d7YciQegWm6QE44kNERESZePYsAjo6UpQv/+muCI6OVjh/vl+6C5jzO474EBERUboOHXqIFi1+w+DBh5GYqH5ZmYLY9ABsfIiIiOgL8fHJmDTpFAYOPIyoqERcv/4Oq1f/LXZauYJTXURERKTy5Ek4Bg06jDt3PqhinTs7YNCgOuIllYvY+BAREREAYO/e+xg//gRiYpIAAIaGOpg3rzm8vGoWqAXMmWHjQ0REVMTFxSVh+vSz2Lr1tipWubIFAgI8UaOGlYiZ5T42PkREREVYXFwS2rXbjnv3Pqpi3btXw6JFLWBioi9iZtrBxc1ERERFmJGRHtzcbP///3WxcmVr+Pm1LZRND8ARHyIioiJv2jRXhIbGYeTI+qhatYTY6WgVGx8iIqIi5P79j3jwIBQdOzqoYvr6OvDzaytiVnmHjQ8REVERIAgCtm+/iylTTkOpFGBnZwlHx8K1cDkruMaHiIiokIuOTsTw4ccwZsyfiItLRkKCAsuXB4mdlig44kNERFSI3bnzAYMHH0ZwcLgq1rdvLcyZ4yZiVuJh40NERFQICYKALVtuY/r0M0hISLnPlomJPpYvb4VOnRy+snfhxcaHiIiokImKSsC4cSdw4MBDVaxmzZIICPBApUoWImYmPjY+REREhUz//odw4cIL1dcDB9bBzJlNYWjIj30ubiYiIipkJk9uBF1dKWQyAwQGemLBAnc2Pf+PVSAiIipkGjQoi1Wr2uCbb8qgQgVzsdPJV9j4EBVg+s/2wfjfeZAkRefp80oAQCqBmVKAkKfPXLSwznmjoNf56jMr+J+vjvVe56Ej/fQKBgPA3///Xz4hiXsndgpsfIgKMuN/50E38uHXN9QSzpXnDdY5bxS0OgsCsOKCMyYdaYUkhQ4qyt5gZutzYqeVJYKeiWjPzcaHqABLHekRJFIojUrn3fMCkEolUBbQv5ALCtY5bxTEOofFGGDgVjcculVBFTsRXBWTDR5DVyf/vgoJAKmhDHG1p4mWAxsfokJAaVQaYd3u59nz6epKYWFhjMjwGCQnK/PseYsa1jlvFLQ6X736BkOGHMHr11Gq2MiR9TFlSmPI9ZaImNnXpdY6KTwGEKnWbHyIiIgKAKVSwOrVf2P+/ItQKFJGdSwtDeHn1xYtW1YSObuCg40PERFRPvfxYyxGjTqOU6eeqWJOTtbw92+PsmVNxUusAGLjQ0RElM+tWfO3qumRSIDRoxvixx9TrtVDmmHjQ0RElM9NnOiCU6ee4cOHGKxe3Q7Nm1cQO6UCi40PERFRPqNQKKGj82k0x8hIDxs3dkCxYnooXVq8U8ELA46RERER5SMXLrxA48abEBwcrhavVMmCTU8uYONDRESUDygUSixefBnduu3GkycRGDToMOLjk8VOq9DhVBcREZHIQkKiMWzYMVy8+FIVK1GiGOLiknhz0VzGahIREYnozJlnGDHiGD5+jAOQchXpyZMbwcenIaRSicjZFT5sfIiIiESQnJwytbVy5VUI/3+XidKljeHv7wEXl3LiJleIsfEhIiLKY2/eRGHo0KMICnqtirVoUQGrVrVFiRLFRMys8GPjQ0RElMcePw7HX3+lND06OhJMneqKESPqc2orD/CsLiIiojzWtKkNxo51grW1KQ4c6IlRoxqw6ckjHPEhIiLSspCQGJQsWQwSyafmZsIEFwwd+g3MzQ1FzKzo4YgPERGRFh0/HowmTTZh3brranFdXSmbHhGw8SEiItKCxEQFfvrpLPr2PYCIiATMmXMB16+/FTutIo9TXURERLns+fNIeHsfxo0bIapYmzaVYGdnIWJWBLDxISIiylWHDz/CmDF/Qi5PAADo6+tg9mw3/PBDbbU1PiQONj5ERES5ID4+GbNnn0dg4E1VrEIFM2zY4IlatUqJlxipYeNDRESUQ69fR6Fv3wO4ffu9KtapkwOWLWsJU1MDETOjL7HxISIiyiGZTB/R0YkAAAMDHcyb1xx9+tTk1FY+xLO6iIiIcsjU1AAbNniiRg0rHD/+Pfr2rcWmJ5/iiA8REZGGHj8Og6GhLsqVk6liNWuWxKlTXrwCcz7HER8iIiIN7Np1Dy1bboO39xEkJSnUHmPTk/+x8SEiIsqC2NgkjBnzP4wYcRyxsUn4+++3WL/+hthpkYY41UVERPQVDx6EYtCgw3jwIFQV+/77GhgwoLaIWVF2sPEhIiLKgCAI2LHjLiZPPo24uGQAQLFieli8uAV69KgucnaUHWx8iIiI0hEdnYhJk05h167/VLFq1UpgwwZPVKliKWJmlBP5Yo3P06dPMXDgQNSpUwcuLi6YO3cu4uPjv7pfbGwsli5dipYtW6J27dpo3bo1Vq1ahcTExDzImoiICquYmCS0afO7WtPTt28tHD/+HZueAk70ER+5XI5+/fqhbNmy8PX1RVhYGBYsWICIiAgsXbo0031nzZqFkydPYuzYsahSpQpu3boFX19fREZGYvr06Xn0CoiIqLAxNtaDu3sFPHoUBhMTfSxb1hKdO1cVOy3KBaI3Pjt27IBcLsf+/fthaZnSRevo6GDChAkYNmwY7Ozs0t0vOTkZx48fx6BBg9CnTx8AgLOzM968eYOjR4+y8SEiohz56acmiIpKgI9PQ1SqxLuqFxaiT3WdP38eLi4uqqYHANq0aQN9fX2cO3cuw/0EQYBCoYCpqalaXCaTQRAEreVLRESFz40bb7F//wO1mL6+DlasaMOmp5ARvfEJDg5OM6qjr68PGxsbBAcHZ7ifnp4eunTpgq1bt+Lff/9FTEwMgoKC8Mcff6B3797aTpuIiAoBQRCwYcMNODsHYsSIY/jvv49ip0RaJvpUl1wuh0wmSxOXyWSIjIzMdN9Zs2Zh5syZ6NGjhyrWp08fjBw5Mkc56ermfj+ooyNV+5e0o6jVWfLZv9p432akqNVZLKyzdkVGxsPH5384dOiRKubndw3+/h4iZlW45Yf3tOiNT0YEQfjqDd6WLl2Ks2fPYs6cOahYsSLu3r0LX19fyGQy+Pj4ZOt5pVIJLCyMs7VvVshkRlo7Nn1SZOr8/5fH1/b7NiNFps4iY51z37Vrr9Gz5248fRqhio0Z44RFi1pBX19HvMSKCDHf06I3PjKZDHK5PE08Kioqw4XNAPDw4UP8+uuvWLNmDVq0aAEAaNCgASQSCRYvXozevXujePHiGuejVAqQy2M13u9rdHSkkMmMIJfHQaFQ5vrxKUVRq7OZUoAUKe/byPCYPHveolZnsbDOuU8QBKxbdx2zZp1DUlJKTc3MDLB5cyc0b26LmJh4xOTdj1KRo633tExmlOVRJNEbHzs7uzRreRITE/HixQt07do1w/0eP34MAKhWrZpavFq1akhOTsbr16+z1fgAQHKy9n7BKBRKrR6fUhSVOguf/SvG6y0qdRYb65w7wsPjMHr0nzh+/NNnzjfflMGvv3qiVq2yCA+PYZ3ziJjvadEnjps2bYqgoCCEh4erYidOnEBiYiLc3Nwy3M/a2hoAcPfuXbX4nTt3AADlypXTQrZERFQQCYKAvn0PqjU9I0bUx8GDPVC+vJmImVFeE73x6dWrF0xNTTF8+HBcuHAB+/fvx5w5c9ChQwe1qa6pU6eievVP90VxdHRErVq1MHPmTGzfvh1BQUEICAjAqlWr0L59e7XT44mIqGiTSCSYPt0VOjoSWFoaYtu2Tpg5syn09Liep6gRfapLJpNh8+bNmDt3LkaNGgVDQ0N4enpiwoQJatsplUooFArV1zo6Oli3bh1WrlyJgIAAfPz4EWXKlIGXlxeGDh2a1y+DiIjyOScna6xe3Q7OztYoW9b06ztQoSQReLU/NQqFEmFhub+yTVdXCgsLY84ha1lRq7Pl7qrQiX0DRbGyCOt2P8+et6jVWSysc/YFBb3Ctm13sHJlG0ilmZ8hzDrnHW3V2tLSuOAsbiYiIsotSqUAX9+rWLToMhQKAZUqWWDsWCex06J8RPQ1PkRERLnhw4dY9Oy5F/PnX4JCkTKZcenSS14KgNRwxIeIiAq8S5deYujQowgJSVmqIJEA48c7Y/x4Z175mtSw8SEiogJLoVBi+fK/sGxZEJTKlFGekiWNsXZtOzRpYiNydpQfsfEhIqICKSQkBsOHH8WFCy9VMTc3W6xe3RYlS+b9LVyoYGDjQ0REBZKf3zVV0yOVSjBpUiOMHt3wq2dxUdHGxoeIiAqkKVMa4+zZ54iMjIe/vwdcXHjFfvo6Nj5ERFQgJCcroav7aaFysWJ62Lz5W8hkBihRopiImVFBwqXuRESU75069RQuLhvx5Em4WrxSJQs2PaQRNj5ERJRvJSUp8PPP5/Hdd/vw/HkkvL2PICEhWey0qADjVBcREeVLr17J4e19BH///VYVK1PGBAkJChgY8OOLsofvHCIiyneOHw+Gj89xREQkAAD09KT46acmGDKkHiQSnrVF2cfGh4iI8o3ERAXmzLkAf//rqpiNjQzr13ugXr0yImZGhQUbHyIiyheeP4/EkCFHcP36O1XMw6MyVqxoDTMzQxEzo8KEjQ8REeULT56Eq5oefX0dzJ7dFD/8UIdTW5SreFYXERHlC82bV8Do0Q1RoYIZjh7thYED67LpoVzHER/Kd/Sf7YPxv/MgSYrWeF8JAEglMFMKEHI9s/xHGvfu6xsR5VNv30ahdGkTteZm0qRG8PFpAFNTAxEzo8KMjQ/lO8b/zoNu5MMcHaOoDWUKeiZip0CkkQMHHmDs2BOYMqURBg+up4rr6krZ9JBWsfGhfCd1pEeQSKE0Kq3Zvki5WaGyiIz4AClNT0yd6WKnQZQlcXFJ+Omnc9iy5RYAYNas83ByskatWqVEzoyKCjY+lG8pjUojrNt9jfbR1ZXCwsIYkeExSE5WaikzIsqOx4/DMGjQYdy791EV+/Zbe1SqZCFiVlTUsPEhIiKt27XrHiZOPIXY2CQAgJGRLhYscMd339XgAmbKU2x8iIhIa2JjkzB16mn8/vtdVcze3hIbNniiatUSImZGRVW21oAGBwdj3LhxcHV1haOjI+7eTXlD+/n5ISgoKFcTJCKigunZswi0bfu7WtPz3Xc18L//9WbTQ6LRuPH577//0K1bN1y9ehUNGzaEQqFQPRYTE4MdO3bkaoJERFQwWVoaIS4u5U7qxYrpws+vLVaubANjYz2RM6OiTOPGZ+nSpXBwcMCJEyewePFiCMKnc2dq1aqF27dv52qCRERUMMlkBggI8ECdOqVw4oQXevSoLnZKRJo3PtevX8egQYNgZGSUZkFaiRIl8PHjxwz2JCKiwuzevQ948yZKLVanTmn873/fo0oVS5GyIlKXrTU+enrpD1NGRkZCX18/RwkREVHBIggCtm69hbZtf8eQIUfTXEqCZ21RfqJx4+Pg4ICTJ0+m+9iFCxdQo0aNHCdFREQFQ1RUAoYOPYrx408iPl6Bv/56jV9/vSl2WkQZ0vh09r59+2L8+PEwMjJCx44dAQBv375FUFAQ9uzZA19f31xPkoiI8p/bt99j0KDDePo0QhUbMKA2+vatJV5SRF+hcePTvn17vHjxAn5+fti6dSsAYNSoUdDR0YGPjw/c3d1zPUkiIso/BEHAxo3/YsaMc0hMTDmz19RUHytWtEaHDvYiZ0eUuWxdwHDo0KHo1KkTLly4gNDQUFhYWMDV1RXW1ta5nR8REeUjcnkCxo79E4cOPVLF6tQphfXrPVChgrl4iRFlkcaNz7Vr11C9enWULl0a3bt3V3ssJiYG9+7dQ4MGDXItQSIiyh+iohLQosVveP48UhUbMqQepk93hYEBbwRABYPGi5v79u2L4ODgdB97+vQp+vbtm+OkiIgo/zE1NUDLlhUBAGZmBti8+VvMmdOMTQ8VKBq/Wz+/YOGXkpOTIZVm6wx5IiIqAGbNaorERAXGjHFC+fIysdMh0liWGp/o6GjI5XLV1x8+fMCbN2/UtomPj8e+fftQogTvv0JEVBj8/fcbvHkTjW+//bRg2cBAF8uWtRIxK6KcyVLjs2nTJqxevRpAyoWoRo4cme52giBgyJAhuZcdERHlOaVSwNq1/2DevIvQ15fCwaE4HByKi50WUa7IUuPTuHFjFCtWDIIgYMmSJfDy8kLZsmXVttHX14e9vT0aNmyolUSJiEj7wsLiMGrUcZw48RQAkJysxNq1f2PFijYiZ0aUO7LU+NStWxd169YFAMTFxaF79+4oVaqUVhMjIqK8FRT0GkOHHsGbN9Gq2OjRDTFpUiMRsyLKXRovbs5omouIiAompVLAqlXXsHDhJSgUKSewlChhBD+/dnB3ryBuckS5LFvnICoUCpw/fx7BwcGIj49Xe0wikWDEiBG5khwREWnXhw+xGDHiGM6efa6KNW5cDmvXtkfp0iYiZkakHRo3PuHh4ejduzeePHkCiUSiOr3987vvsvEhIsr/BEGAl9c+3LgRAgCQSIBx45wxYYIzdHR4aRIqnDR+Z//yyy8wMDDAmTNnIAgC/vjjD/z555/o378/KlSogLNnz2ohTSIiym0SiQSzZrlBKpXAyqoYdu3qhkmTGrHpoUJN43d3UFAQ+vfvj5IlS6YcQCqFjY0NJk2ahEaNGmHRokW5niQREWmHi0s5rFnTDqdP90HTpjZip0OkdRo3Pu/evYO1tTV0dHQglUoRFxeneqx58+a4dOlSriZIRES549y55xg58jiUSvUr8HfpUhWlShmLlBVR3tK48bGwsEB0dMqpjiVLlsTDhw9Vj0VGRkKhUORedkRElGPJyUosXHgJPXrswR9/3IOf3zWxUyISjcaLm2vUqIFHjx6hWbNmaNq0KdasWQMTExPo6elh+fLlqF27tjbyJCKibHj7NgpDhx7FlSuvVbGrV99AqRQglUoy2ZOocNK48fHy8sKLFy8AAGPGjMG///6LSZMmAQBsbGwwbdq03M2QiIiy5fTppxgx4jhCQ1OWJOjoSDBlSmOMHNmATQ8VWRo3Po0aNUKjRilX8bS0tMT+/fvx8OFDSCQSVKpUCbq62bo0EBER5ZKkJAUWLboMX99PU1ply5rA398DTk7WImZGJL4cdykSiQQODg4AUq4JceDAAXTs2DHHiRERkeZev46Ct/cRXLv2RhVr3boSfH3bwNLSSMTMiPKHXLtYw9GjR+Hh4YHJkyfn1iGJiEhDq1ZdVTU9urpSzJ7thq1bO7LpIfp/WW581q9fD3d3d9SuXRudOnXC+fPnAQDXr19Hx44dMX78eMjlcvz0009aS5aIiDI3fXoTVKliifLlZTh0qCeGDftG7cr6REVdlqa6tm3bhuXLl8PU1BT29vZ4+/YtRowYgZ9++gk///wzdHV1MWLECPzwww8oVqyYtnMmIqL/l5SkgJ6ejuprExN91QiPubmhiJkR5U9Zanz27NmDb775Bv7+/jAxMYFCocCsWbMwc+ZMWFtbIzAwELa2ttrOlYiIPnP06GPMmHEWu3d3Q4UK5qp4pUoW4iVFlM9laarr6dOnGDBgAExMUu7Uq6Ojg2HDhkEQBIwePZpNDxFRHkpISMa0aWfQv/9BvHghh7f3ESQkJIudFlGBkKURn7i4ONW9uVKVKlUKANj0EBHloadPI+DtfQT//huiitnYmCEpSQkDAxETIyogcnxWV25ct+fp06cYOHAg6tSpAxcXF8ydOxfx8fFZ2jciIgKzZs2Cq6sratasiTZt2mDHjh05zomIKL85ePAhWrb8TdX0GBjoYPHiFggI8ICJib7I2REVDFnuWhYtWgRTU9M08fnz56umwICU6/qsXbs2ywnI5XL069cPZcuWha+vL8LCwrBgwQJERERg6dKlme4bExODPn36wMDAAFOnTkXx4sXx/PlzJCUlZfn5iYjyu/j4ZEydegabNv2rilWqZI6AAE/UrFkykz2J6EtZanzKli2Lt2/f4u3bt2nib968UYtpetrkjh07IJfLsX//flhaWgJIWUM0YcIEDBs2DHZ2dhnu6+/vj/j4eOzatQuGhilnLzg5OWn0/ERE+dnDh6Ho2nUn7tz5oIp16VIVS5e25CgPUTZkqfE5ffq01hI4f/48XFxcVE0PALRp0wZTp07FuXPnMm189uzZg759+6qaHiKiwiY4OEzV9Bga6mDBAnd8/70jr81DlE2i31grODgYXbt2VYvp6+vDxsYGwcHBGe738uVLfPz4ETKZDEOGDMGlS5dgbGyM9u3bY9KkSTlqhnR1c+2C1io6OlK1fyljks/+1fR7wTrnDdY5b+joSNGuXRWMHt0Qx449xq+/dkD16lZip1Xo8P2cd/JDrUVvfORyOWQyWZq4TCZDZGRkhvt9/PgRALB48WK0bdsWAQEBePz4MZYvX46kpCTMnTs3W/lIpRJYWBhna9+skMl42fiv+v+7Rufke8E65w3WOfe9eBGJ8uVlaiM6S5a0xrx5LWBszKktbeL7Oe+IWWvRG5+MCIKQ6VCuUqkEANjZ2WHBggUAABcXFyQnJ2Px4sUYPXo0rKw0/8tIqRQgl8dmL+lM6OhIIZMZQS6Pg0KhzPXjFyZmSgFSpHwvIsNjNNqXdc4brLN2bN9+BxMnnsTs2W4YOLCuqs5xcYlQKJRITOSJG9rA93Pe0VatZTKjLI8iid74yGQyyOXyNPGoqKhM1/eYm5sDAJydndXizs7OUCqVCA4OzlbjAwDJydp74ysUSq0evzAQPvs3u7VinfMG65w7YmKSMHnyKezceQ8AMG3aWdSvXxa1a6dcL411zhusc94Rs9aiT2ja2dmlWcuTmJiIFy9eZNr4lC9fHnp6emnigpDysSmViv7SiIi+6r//PqJ1622qpgcAevasjkqVzMVLiqgQE707aNq0KYKCghAeHq6KnThxAomJiXBzc8twP319fTRu3BhXrlxRi1+5cgW6urqoXLmy1nImIsopQRDw22+30abNNjx6FAYAMDbWw9q17bBsWSsYGaX9w46Ici7bjU9UVBQuXLiAgwcPZroI+Wt69eoFU1NTDB8+HBcuXMD+/fsxZ84cdOjQQW3EZ+rUqahevbraviNGjMCDBw/w448/4uLFi9i0aRNWrVqF3r17q50eT0SUn0RHJ2LYsGMYN+4E4uMVAIAaNaxw8qQXunatJnJ2RIVbttb4rF69GgEBAYiPj4dEIsHu3bthZmaGfv36oXHjxvD29s7ysWQyGTZv3oy5c+di1KhRMDQ0hKenJyZMmKC2nVKphEKhUIvVqlUL/v7+WLZsGYYOHQpzc3N4eXlh9OjR2XlZRERaFxwcjt699+HJkwhVbMCA2pg92w2GhqIvuyQq9DT+Kdu2bRtWr16N77//Hk2aNMGQIUNUjzVv3hx//vmnRo0PAFSsWBGBgYGZbrNw4UIsXLgwTbxx48Zo3LixRs9HRCSWEiWMVIs6TU318csvrfHtt/YiZ0VUdGSr8enfvz9+/PHHNCMwtra2eP78ea4lR0RU2JiZGWL9eg9Mn34Wq1e3Q8WK5mKnRFSkaNz4vHz5Ek2aNEn3MWNj43RPTSciKqr+/TcEpUoZo3TpTzdzrlevDI4c6cXbThCJQOPFzaampqqrJn/p9evXKF68eI6TIiIq6ARBQEDAdbRvvx1Dhx5Nc80SNj1E4tC48XFxccGGDRsQG/vp6sYSiQTJycnYvn07XF1dczVBIqKCJiIiHv37H8S0aWeRlKTE5cuvsHXrbbHTIiJkY6rLx8cH3bp1g4eHB1q2bAmJRILffvsN//33H968eYMVK1ZoIU0iooLhn3/ewtv7CF6+/DTtP2zYN+jd21HErIgolcYjPra2tti+fTsqVaqE7du3QxAEHDhwABYWFvj9999RtmxZbeRJRJSvCYKANWv+RocOO1VNj4WFIX77rRNmz3aDvr6OyBkSEZDN6/hUrlwZgYGBSExMRHh4OMzMzGBoaJjbuRERFQhhYXHw8fkf/vzziSrWsGFZ+Pt7wNraVMTMiOhLGo/4nDlzRnVndH19fZQqVYpNDxEVWRER8WjR4je1psfHpwH27evOpocoH9K48Rk2bBiaNm2KJUuWpLm5KBFRUWNubohWrSoBAIoXN8KOHZ0xfXoT6OlxaosoP9J4qsvf3x/79u3D1q1b8euvv6JWrVro2rUr2rdvDxMTk68foIjSe7oX2D8fZvFyCGInk89J496JnQKRRn7+2Q2CIGDcOCeUKcNRHqL8TCIIQrY+h+VyOQ4dOoT9+/fj9u3bMDQ0RKtWrdClSxe4uLjkdp55RqFQIiwsJtePa3mgPnQiH+b6cQuzZDN7hHf8W6N9dHWlsLAwRnh4TJrrplDuKcp1vnz5JUJD49Chg/ZvM1GU65yXWOe8o61aW1oaQ0cna5NY2b4jnkwmQ+/evdG7d28EBwdjz5492L9/P44cOYJ79+5l97CFliQpGgAgSKRQGpUWOZv8T9AzQUyd6WKnQaSiUCixYsVVLFlyBYaGuqhatQSqVLEUOy0i0lCObwUsCALevn2Ld+/eITo6GtkcQCoyBKPSCOt2X+w0iEgDISExGD78GC5ceAEAiI1NwoYNN7BoUQuRMyMiTWW78Xn+/Dn27t2LAwcOICQkBCVLlsSAAQPQpUuX3MyPiEhU58+/wLBhR/HhQ8rV6qVSCSZOdMGYMQ1FzoyIskPjxmfPnj3Yu3cvrl+/Dj09Pbi7u6NLly5wdXWFVKrxSWJERPmSQqHE0qVBWL48CKkD2aVLG2PduvZo1Ki8uMkRUbZp3PhMmzYN1atXx7Rp09ChQweYmZlpIy8iItG8exeNoUOP4vLlV6pY8+a2WL26HUqUKCZiZkSUUxo3Pvv370fVqlW1kQsRkegEQcB33+3D3bsfAAA6OhJMmdIYI0c2gFTKO6oTFXQaz02x6SGiwkwikWDOHDdIpRKULWuC/ft7wMenIZseokIiSyM+fn5+6N69O0qVKgU/P79Mt5VIJBgxYkSuJEdEJAZXVxusW9ceTZvawNLSSOx0iCgXZbnxadq0KRsfIip0Tpx4ggMHHsLXt43aqE6nTg4iZkVE2pKlxuf+/fvp/j8RUUGVmKjAvHkXsXbtPwCAatVKYMSI+iJnRUTaluMLGBIRFTQvXkRiyJAj+OefT/eFu3HjHQRBgETCtTxEhZnGi5urVauGW7dupfvYnTt3UK1atRwnRUSkLUePPkaLFr+pmh49PSnmzWuGgAAPNj1ERYDGIz6Z3ZJCqVTyFwcR5UsJCcn4+ecLCAi4oYrZ2pohIMADderw/nlERUWuTnXdvXsXpqamuXlIIqIce/o0At7eR/DvvyGqWIcOVfDLL60hkxmImBkR5bUsNT6bN2/Gli1bAHw6a0tfX19tm4SEBISGhqJNmza5nyURUQ74+V1TNT0GBjr4+edm6N+/FkeoiYqgLDU+xYsXR5UqVQAAr1+/Rvny5SGTydS20dfXh729Pfr27Zv7WRIR5cDs2W64fPkVBEFAQIAnatYsKXZKRCSSLDU+np6e8PT0BAD06dMHs2bNgp2dnVYTIyLKrsREBfT1dVRfm5joY9u2TihZ0hgmJvqZ7ElEhZ3GZ3Vt3bqVTQ8R5Vt7996Hs/OvePEiUi1eqZIFmx4iytqIz5s3b2BlZQU9PT28efPmq9uXLVs2x4kREWkiLi4J06adwW+/3QEADBlyBAcO9FQb+SEiylLj06JFC+zcuRO1atWCu7v7VxcE/vfff7mSHBFRVjx8GIrBgw/jv/9CVTE7O0skJyvZ+BCRmiw1PvPnz0f58uVV/88zIYgov9ix4y4mTz6F2NhkAECxYrpYuLAFevWqIXJmRJQfZanx6dy5s+r/u3TporVkiIiyKiYmCZMnn8LOnfdUsapViyMgwBMODsVFzIyI8jONFzenJyEhAcHBwVAoFLlxOCKiTP3330e0abNNrenx8nLE8ePfs+khokxl66yu1atXq76+c+cO3Nzc4OnpiTZt2uDt27e5miAR0ZdevIjEw4dhAABjYz2sXdsOy5e3RrFieiJnRkT5ncaNz65du9QuXrh06VKYmZlhypQpEAQBa9euzdUEiYi+1KaNHYYO/QY1aljh5EkvdO3KmyMTUdZofK+ut2/folKlSgCA6Oho/P3331i+fDlat24NmUwGX1/fXE+SiIq2588jYWMjUzuxYvp0VyiVAgwNc/WWg0RUyGk84pOYmAhd3ZRfNDdv3oRSqUSjRo0AAOXKlcPHjx9zN0MiKrIEQcCmTf/C1XUTtmy5rfaYvr4Omx4i0pjGjU+ZMmXw999/AwBOnTqFqlWrwsTEBAAQFham+n8iopyQyxPg7X0EP/54CgkJCkyffgb37/MPKyLKGY3/XPr222+xevVqnDp1Cvfv38ePP/6oeuzOnTuoUKFCbuZHREXQv/+GYPDgw3j27NNtJ/r0qYmKFc3FS4qICgWNG59hw4ZBV1cX169fR8uWLdGnTx/VYw8fPkTr1q1zNUEiKjoEQUBg4E3MmnUeiYkpl8eQyQywYkVreHpWETk7IioMNG58JBIJvL29031s3bp1OU6IiIqmiIh4jBnzJ44efayK1atXGv7+HrC1NRMxMyIqTLK9MjA6Oho3b95EREQELCwsULt2ba7vIaJsuX//I7y89uPFC7kqNmzYN5g2zZX32iKiXJWtxicwMBB+fn6Ij4+HIAiQSCQwNDSEj48PBgwYkNs5ElEhV6qUMZRKAQBgYWEIX982aNPGTuSsiKgw0rjx2b9/P5YsWYKmTZuic+fOKFmyJN6/f4/9+/dj8eLFsLCwQKdOnbSQKhEVVhYWRvD398C8eRfh59cW5crJvr4TEVE2aNz4bNq0CZ6enli6dKlavF27dpgwYQI2b97MxoeIMnX16hvY2spQqtSn6fEGDcpi377uahcpJCLKbRpfx+fJkyf49ttv033s22+/RXBwcI6TIqLCSakU4Ot7FR077sSwYcegUCjVHmfTQ0TapnHjY2hoiMjIyHQfi4yMhKGhYY6TIqLC5+PHWPTuvQ9z516EQiHg4sWX2LHjrthpEVERo3Hj880338DPzw8hISFq8Q8fPmD16tWoX79+riVHRIXDlSuv4O6+FadOPQMASCTAuHFO6NmzhriJEVGRo/Ean3HjxqFnz55o3bo1XFxcYGVlhQ8fPiAoKAi6urrw8/PTRp5EVAApFEqsXHkVixdfUZ21VaJEMaxd2w5ubrYiZ0dERZHGjU+VKlWwe/du+Pn54a+//kJERATMzc3RokULjBw5EhUrVtRGnkRUwLx/H4Phw4/h/PkXqliTJuWxZk07tUXNRER5SaPGR6FQICwsDOXKlcPy5cu1lRMRFXAfP8aiefOt+PAhFgAglUowYYIzxo51go6OxjPsRES5Jku/gQRBwLJly1C/fn00bdoU9erVw7hx4xAdHa3t/IioACpRohjatk25AGGpUsbYs6cbJkxwYdNDRKLL0ojPli1bEBAQgPLly6NGjRp48eIFjh49Cj09PSxatEjbORJRATR3bjPo6UkxfrwLrKyKiZ0OERGALI747N27F25ubjh27BhWrFiBvXv3YvDgwTh69CgSEhJynMTTp08xcOBA1KlTBy4uLpg7dy7i4+M1OsaJEyfg4OAAT0/PHOdDRJo5c+YZDh16qBYzMtLDwoUt2PQQUb6Spcbn2bNn6NWrF3R1Pw0Q9enTB0lJSXj16lWOEpDL5ejXrx9iYmLg6+uLSZMm4dChQ5g+fXqWjxEfH48FCxagRIkSOcqFiDSTnKzEvHkX0bPnXvj4/A/BweFip0RElKksTXUlJCSgePHiarHUr3M64rNjxw7I5XLs378flpaWAAAdHR1MmDABw4YNg53d129U6O/vj7Jly6JcuXK4c+dOjvIhoqx59UqO7t3/QFDQawBATEwSNm++hZ9/dhM5MyKijIm+0vD8+fNwcXFRNT0A0KZNG+jr6+PcuXNf3f/FixfYuHGjRiNERJQzJ048QZ0661RNj66uFDNnNsWsWU1FzoyIKHNZPp19woQJMDAwSBMfO3Ys9PX1VV9LJBIcPHgwywkEBweja9euajF9fX3Y2Nhk6b5f8+bNQ8eOHVG1atUsP+fX6OpqoR+UfPpXK8cnAFCdNcSzh7QjKUmBuXMvYtWqa6pYuXKm2LChAxo2LCtiZoUT3895g3XOO/mh1llqfBo0aKBRXBNyuRwymSxNXCaTZXhPsFSnT5/GjRs3cPz48RznkUoqlcDCwjjXjqfy/zdflEq0dHxSI5MZiZ1CofP8eQR69dqDoKBP6/o6dnTAr792hKUl661NfD/nDdY574hZ6yw1Plu3btV2HmkIgpDpnZoTEhIwf/58jBo1Sm2aLKeUSgFyeWyuHS+VmSBACkApCIgMj8n141MKHR0pZDIjyOVxae78TdmnVApo3XorHj4MAwDo6UmxZEkr9O9fC0qlEuF8T2sF3895g3XOO9qqtUxmlOVRJI1vWZHbZDIZ5HJ5mnhUVFSmC5s3b94MqVQKDw8P1f5JSUlQKpWQy+UwNDRUm4LTRHKyFt74wqd/tXJ8UqNQKFnnXDZvXnP06LEH5cub4ddfPdGiRWWEh8ewznmA7+e8wTrnHTFrLXrjY2dnl2YtT2JiIl68eJFm7c/nnjx5gufPn8PFxSXNYw0aNMCsWbPw3Xff5Xq+REWVm5st1q/3QLNmtihenNfmIaKCSfTGp2nTpli7di3Cw8NhYWEBIOVihImJiXBzy/i02MGDB6Nz585qsfXr1+Pp06dYsGABKlSooM20iQq1Q4ce4tixYKxe3VZtyrljRwcRsyIiyjnRG59evXrht99+w/DhwzF8+HCEhoZi4cKF6NChg9pU19SpU7F//37cu3cPQMpI0ZdTYfv27UNISAicnJzy9DUQFRbx8cmYOfMcNm78FwBQu3YpDBlST+SsiIhyj+iNj0wmw+bNmzF37lyMGjUKhoaG8PT0xIQJE9S2UyqVUCgUImVJVPg9eRKOQYMO486dD6rY7dvvv3qiARFRQSIRBEH4+mZFh0KhRFhY7p+hUnx3VUhj30BZrCxCu93P9eNTCl1dKSwsjLnoVkN7997H+PEnEBOTBAAwNNTBvHnN4eVVM92mh3XOG6xz3mCd8462am1paaz9s7qCg4Nx7do1hIeHo1u3brCyskJISAjMzMxgaGiY3cMSUR6Ki0vC9OlnsXXrbVWscmULBAR4okYNKxEzIyLSDo0bH4VCgZ9++gn79u1TDYE3bdoUVlZWmDlzJqpVq4bRo0drI1ciykWPHoVh0KDD+O+/j6pY9+7VsGhRC5iYZO9SEERE+Z3G14xeu3YtDh8+jB9//BGHDx/G5zNlTZo0wYULF3I1QSLSDj+/a6qmx8hIF76+bbB6dTs2PURUqGk84rNv3z4MHz4cAwYMSLPYuFy5cnj16lUGexJRfjJ3bjMEBb2GgYEOAgI84eBQXOyUiIi0TuPGJyQkBHXq1En3MQMDA8TE8NL1RPlRfHwyDA0//cibmhpg+/bOKF3aBMWK6YmYGRFR3tF4qqt48eJ4+fJluo89ffoUpUuXznFSRJR7BEHAtm234eT0K169Ur89TKVKFmx6iKhI0bjxcXNzw7p16xASEqKKSSQSREVFYevWrWjevHmuJkhE2RcdnYjhw49h7NgTePs2GkOGHEVSEq+HRURFl8ZTXT4+Pjh//jzat28PJycnSCQSLF++HI8ePYKuri6GDx+ujTyJSEN37nzA4MGHERwcropVq1YCCoUAPQ7yEFERpfGIT4kSJbB79254eHjg7t270NHRwf3799G0aVPs2LED5ubmWkiTiLJKEARs2vQv2rX7XdX0mJjoY/16Dyxd2lJtnQ8RUVGTrd+AJUqUwM8//5zbuRBRDkVFJWDcuBM4cOChKlarVkmsX++BSpUsRMyMiCh/4J9+RIXErVshGDToMJ49i1TFBg6sg1mzmsLAgD/qRERANhqfKVOmZPq4RCLB/Pnzs50QEWXPq1dRqqZHJjPAL7+0QocO9iJnRUSUv2jc+Pz1119pYhEREYiNjYVMJoOpqWmuJEZEmmnfvjK8vevi2rU3WL/eE7a2ZmKnRESU72jc+Jw+fTrd+JUrVzB79mysXLkyx0kR0dc9eRKOihXN1e6ePmNGUwCAvr6OWGkREeVrGp/VlREXFxd4eXlh3rx5uXVIIkqHIAhYu/YfuLpuxu+/31F7TF9fh00PEVEmcq3xAQA7Ozvcvn07Nw9JRJ8JD49D374HMHPmOSQnKzF16hk8fhwmdlpERAVGrp7qce3aNVhY8JRZIm24evUNhgw5gtevo1SxgQPrcC0PEZEGNG58/Pz80sSSkpLw4MEDnD9/HgMHDsyVxIgohVIpYPXqvzF//kUoFAIAoHhxI/j5tUWLFhVFzo6IqGDJlcZHX18f1tbW8PHxYeNDlIs+fozFqFHHcerUM1XM2dka/v7tUaYMz6AkItKUxo3P/fv3tZEHEX3h9u336N17H969iwEASCTA2LFOmDDBBbq6ubo8j4ioyNDot2d8fDzGjx+Pv//+W1v5ENH/s7Y2hVSacqp6iRLFsHNnV0ye3JhNDxFRDmj0G9TQ0BCnTp2CIAjayoeI/p+lpRHWrfNAs2a2OHPGC82a2YqdEhFRgafxn45Vq1bFw4cPv74hEWnk4sUXeP8+Ri3m7GyNnTu7oFQpE5GyIiIqXDRufCZMmIDAwEBcvXpVG/kQFTkKhRKLFl1G1667MWLEcSiV6iOqn1+ZmYiIciZLi5uvXbuG6tWrw9jYGLNnz0ZMTAz69esHmUyGkiVLqm0rkUhw8OBBrSRLVNi8exeNYcOO4tKlVwCAc+eeY8+e/9C9e3WRMyMiKpyy1Pj07dsXO3fuRK1atWBubg5zc3Mtp0VU+J0+/QwjRx7Dx49xAACpVILJkxuha9dqImdGRFR4Zanx+Xwx89atW7WWDFFRkJycMrW1cuWn6eIyZUzg798ezs7lRMyMiKjwy9VbVhBR5l6/jsKQIUdw9eobVaxly4pYtaotihc3EjEzIqKigY0PUR4JCYmGu/tWhIfHAwB0daWYOrUxhg+vr7peDxERaVeWG59+/fpl6ewSiUSCf/75J0dJERVGpUqZoG1bO2zffhflypnC398DDRqUFTstIqIiJcuNT8OGDWFpaanNXIgKvfnz3WFqqo/x451hYcGpLSKivJblxmfEiBGoVauWNnMhKlSOHXsMhUKAp2cVVczYWA9z5zYXMSsioqKNa3yIclliogI//3we69ffgImJPmrUsELFiuZip0VERMjGlZuJKGPPnkXA03MH1q+/AQCIjk7E9u13RM6KiIhSccSHKJccOvQQY8b8iaioRACAvr4OZs92ww8/1BY5MyIiSpWlxuf+/fvazoOowIqPT8bMmeewceO/qliFCmbYsMETtWqVEjEzIiL6Ekd8iHLgyZNwDBp0GHfufFDFOnVywLJlLWFqaiBiZkRElB42PkTZlJysRK9ee/HsWSQAwNBQB3PnNkefPjV5R3UionyKi5uJsklXV4qFC1sAACpXtsCxY9+jb99abHqIiPIxjvgQ5YC7ewUEBnqiefMKMDHRFzsdIiL6Co74EGXRH3/cw/DhxyAIglq8Qwd7Nj1ERAUER3yIviImJglTppzGjh13AQD16pXGoEF1Rc6KiIiyg40PUSbu3/+IwYOP4MGDUFXs8/8nIqKChY0PUToEQcD27XcxZcppxMUlAwCKFdPD4sUt0KNHdZGzIyKi7GLjQ/SF6OhETJx4Env2fLpwZ7VqJbBhgyeqVLEUMTMiIsopNj5En7lz5wMGDz6M4OBwVaxv31qYM8cNRkZ6ImZGRES5gY0P0WfWrPlb1fSYmOhj2bKW6Ny5qshZERFRbmHjQ/SZhQvdcfXqG5iZGSAgwAOVKlmInRIREeUiNj5UpMXFJalNYclkBti1qyvKlDGBoSF/PIiIChtewJCKJEEQEBh4A05Ov+LNmyi1xypWNGfTQ0RUSLHxoSInMjIeP/xwCFOmnMG7dzEYMuQokpOVYqdFRER5gH/WUpFy/fpbeHsfwYsXclWsTp1SUCqFTPYiIqLCgo0PFQmCIMDf/zrmzLmApKSU0R1zcwP4+rZF27Z2ImdHRER5hY0PFXrh4XHw8fkf/ve/J6pY/fpl4O/vgfLlZSJmRkREeS1fND5Pnz7F3Llz8c8//8DIyAgeHh6YMGECDA0NM9wnOjoaGzduxPnz5/H06VPo6uqiRo0aGDduHGrUqJGH2VN+du3aG3h7H8Hr158WMI8cWR9TpjSGnp6OiJkREZEYRF/cLJfL0a9fP8TExMDX1xeTJk3CoUOHMH369Ez3e/PmDXbu3IlGjRrhl19+wYIFC6BUKtGrVy/cvXs3j7Kn/O7du2hV02NpaYjff++EGTOasukhIiqiRB/x2bFjB+RyOfbv3w9Ly5T7IOno6GDChAkYNmwY7OzSX39Rrlw5nDhxAkZGRqpYo0aN0KJFC/z2229YsGBBnuRP+VuHDvb44YfauHv3I/z926NsWVOxUyIiIhGJPuJz/vx5uLi4qJoeAGjTpg309fVx7ty5DPcrVqyYWtMDAAYGBrCzs8P79++1li/lbw8fhkIQ1M/Qmj3bDfv2dWfTQ0RE4o/4BAcHo2vXrmoxfX192NjYIDg4WKNjxcbG4r///kPHjh1zlJOurhb6Qcmnf7Vy/CJOqRTwyy9/YcGCSwgM/BZdujioHtPV1Rcxs8JJR0eq9i9pB+ucN1jnvJMfai164yOXyyGTpT2zRiaTITIyUqNjrVixAnFxcfDy8sp2PlKpBBYWxtneP0OSlM5HKtHS8YuwkJBo9OmzDydOpJy1NXz4Ebi62qByZcuv7Ek5JZMZfX0jyjHWOW+wznlHzFqL3vhkRBAESCSSr2/4/w4dOoTNmzdjxowZsLW1zfbzKpUC5PLYbO+fETNBgBSAUhAQGR6T68cvqs6ff4EhQ44gJCSlphIJMGlSYxQvboBw1llrdHSkkMmMIJfHQaHgVa+1hXXOG6xz3tFWrWUyoyyPIone+MhkMsjl8jTxqKioDBc2f+nSpUuYMmUKBg4ciN69e+c4J63cvkD49C9vj5BzCoUSy5YFYdmyIKQu6SlZ0hgBAR749ttqCA+PYZ3zgEKhZJ3zAOucN1jnvCNmrUWf0LSzs0uzlicxMREvXrzIUuNz69YtjBw5Em3btsXEiRO1lSblIyEh0ejefQ+WLv3U9Li52eL0aS80aWIjbnJERJSvid74NG3aFEFBQQgPD1fFTpw4gcTERLi5uWW6b3BwMAYPHox69ephwYIFGk2NUcF0/fpbNG/+Gy5efAkgZU3WlCmNsXNnF5QsybVTRESUOdEbn169esHU1BTDhw/HhQsXsH//fsyZMwcdOnRQG/GZOnUqqlevrvo6NDQUAwcOhJ6eHgYNGoS7d+/i5s2buHnzJu7duyfGS6E8YGNjBl3dlAa3dGlj7NvXHWPHOkEqZdNLRERfly/W+GzevBlz587FqFGjYGhoCE9PT0yYMEFtO6VSCYVCofr68ePHePv2LQCgf//+attaW1vj9OnTWs+d8l6JEsXg7++B1av/xooVrVGiRDGxUyIiogJEInx5tbciTqFQIiws988GKr67KqSxb6AsVhah3e7n+vELq9Onn6FmzZKwsspag6OrK4WFhTEXN2sZ65w3WOe8wTrnHW3V2tLSOMtndYk+1UWUnqQkBWbPPo9evfZi5MhjUCrZnxMRUc6x8aF859UrOTp2/AOrV/8NADhz5jkOHXooclZERFQYiL7Gh+hzx48Hw8fnOCIiEgAAenpS/PRTE3z7rb3ImRERUWHAxofyhcREBebMuQB//+uqmI2NDOvXe6BevTIiZkZERIUJGx8S3fPnkfD2PowbN0JUMQ+PylixojXMzAxFzIyIiAobNj4kqlev5GjR4jfI5SlTW/r6Opg9uyl++KEOL0hJRES5joubSVTlysnQtm3KhSorVDDDkSO9MHBgXTY9RESkFRzxIdEtXOiO4sWNMGGCM0xNDcROh4iICjE2PpSn9u9/AD09KTw8qqhiJib6mD078/uyERER5QY2PpQn4uKS8NNP57Blyy2YmuqjRg0rVKhgLnZaRERUxHCND2nd48dhaNduO7ZsuQUAiIpKxN69vG0HERHlPY74kFbt2nUPEyeeQmxsEgDAyEgXCxe6o1evGiJnRkRERREbH9KK2NgkTJ16Gr//flcVc3AojoAAD1StWkLEzIiIqChj40O57v79jxg8+AgePAhVxb7/vgbmz3dHsWJ6ImZGRERFHRsfylVJSQp8//0+vHoVBQAoVkwPixe3QI8e1UXOjIiIiIubKZfp6elgyZKWAIBq1UrgxInebHqIiCjf4IgP5boWLSpi48YOcHevACMjTm0REVH+wREfyjZBELBlyy2MGHEMgiCoPebhUYVNDxER5Tsc8aFsiYpKwIQJJ7Fv3wMAwDfflMEPP9QRNykiIqKvYONDGrt9+z0GDTqMp08jVLFnzyLFS4iIiCiL2PhQlgmCgF9//RczZ55DYqICACCTGeCXX1qhQwd7kbMjIiL6OjY+lCWRkfEYO/YEDh9+pIrVrVsK/v4evOcWEREVGGx86Ktu3HiHwYOP4MWLT9NZQ4bUw08/NYG+vo6ImREREWmGjQ991dq1/6iaHjMzA/j6tkG7dpVFzoqIiEhzbHzoq5YsaYHr19/CysoY69d7oHx5mdgpERERZQsbH0ojJiYJxsafrsFjZmaIPXu6o2xZE+jpcWqLiIgKLl7AkFSUSgF+ftfg4vIr3r2LVnvM1taMTQ8RERV4bHwIABAaGgcvr/34+ecLePcuBkOHHkVyslLstIiIiHIVp7oIQUGvMGTIUbx9mzLKI5EADRuWFTkrIiKi3MfGpwhTKgX4+l7FokWXoVCk3GurRAkjrF7dDs2bVxA3OSIiIi1g41NEffgQixEjjuHs2eeqmKtreaxd2w6lSpmImBkREZH2sPEpgi5ffokhQ44iJCQGQMrU1vjxzhg/3hk6Olz2RUREhRcbnyLow4dYVdNTsqQx1q5thyZNbETOioiISPvY+BRBHTs64OLFl3j2LBKrV7dFyZLGYqdERESUJ9j4FAH3739E1aol1GLz5jWHrq4UUqlEpKyIiIjyHhd0FGLJyUosWHAJbm5b8Mcf99Qe09fXYdNDRERFDhufQurNmyh06bILv/zyFwQB+PHHU2p3VyciIiqKONVVCJ069RQjRhxDWFg8AEBHR4Lx451RrhxvLkpEREUbG59CJClJgQULLsHP729VzNraFP7+HrwSMxEREdj4FBqvXsnh7X0Ef//9VhVr06YSfH3bwMLCSMTMiIiI8g82PoVAUNBr9O27HxERCQAAPT0pfvqpCYYMqQeJhAuYiZRKJRSKZLHT0JhSKUF8vA4SExNUt5Wh3Mc6553s1FpHRxdSae4tSWbjUwhUqmQOfX1dAAmwsZFh/XoP1KtXRuy0iEQnCALk8jDExUWLnUq2ffwohVKpFDuNQo91zjvZqbWRkQlkMstc+WOejU8hULKkMdata4dNm25h2bKWMDMzFDslonwhtekxMbGAvr5BgRwB1dGRcBQiD7DOeUeTWguCgMTEBERHhwMAzMyK5/j52fgUQMePB6NBg7IoXvzT2h1XVxu4uvK2E0SplEqFqukxMSm4ZzTq6kqRnMyRCG1jnfOOprXW1zcAAERHh8PU1CLH0168jk8BkpCQjClTTqNv3wPw8TkOpZJ/nRBlRKFQAPj0S5OICq7Un+PcWKvHxqeAePIkHB4eOxAYeBMAcOLEU/zvf8HiJkVUABTE6S0iUpebP8ec6ioADhx4gLFjTyA6OhEAYGCgg7lzm6NtWzuRMyMiIipY2PjkY3FxSfjpp3PYsuWWKmZnZ4GAAE84OlqJmBkREVHBxMYnn3r8OAyDBh3GvXsfVbFu3aph8eIWMDHRFzEzIiKigouNTz707FkEWrbchtjYJACAkZEuFixwx3ff1eB6BaIiKDDQHxs3Bqi+NjMzg41NBfTtOwAuLq5pto+OjsZvv23C2bOnERLyFsWKFUPt2vXQt+8PqFq1Wprtk5OTceDAHhw/fhTPnj2FQpGMsmXLoXXrtujcuTtMTU21+vrE5ue3Am/fvsa8eUvETiVPvXjxHCtWLMWtWzdgaGiEli3bYNiwkTAwyPySKHFxcdi0aQPOnDmJ0NCPsLIqiVat2qJPnwHQ19dXHXvPnp34559rePfuLczMzFG/fkN4ew9HqVIlVcf63/+OYsuWX7Fly07o6Oho9fWmYuOTD9namqFtWzvs3Xsf9vaW2LDBE1WrlhA7LSISkYGBAVauXAcACA39gN9+24RJk8Zh9eoA1KxZW7VdeHgYRo70RlhYGPr0GYBq1aojLCwUu3fvxNChA/DzzwvRtGkz1faJiYmYOHEMbt26gU6duuGHH7xhYGCAx48fYu/e3Xj16iWmTp2Z1y83z3z48B579+6Cv3+g2KnkqaioKIwePQylS5fG3LmLER4eBj+/XyCXR2LGjDmZ7rt06QJcuHAO3t7DUKlSZdy7dwcbNqxDVJQcY8ZMBABcuxaEmzev49tvO6NyZXt8+PAev/66HkOH/oBt2/6Avn5Kc9WyZRts2LAOx44dhqdnR62/boCNT74kkUiwdGlLlC8vw5gxTjA21hM7JSISmVQqhaNjTdXXNWrUQufO7XDs2GG1xmfZsoV4/foVAgN/g51dZVXczc0dPj5DMW/eTDg67oWlZcqF4AID/XH9+jUsWbISzs6NVNvXq1cfnTt3x/Xrn256nJcSEuK/OvKQGw4c2AsbG1tUq1Y9x9fxyaucc8OBA3sQFSXHxo2/w9zcHEDKrSF+/nk6+vb9ARUqVEx3v+TkZJw5cwq9e/dFt269AKS8V969e4dTp06oGp8WLdqgS5cearMUdnZV0L//dzhz5hTatPH4/+fUQdu2Hti1a0eeNT48nT0f2LHjLo4fVz813cREH9OmubLpIaJ0lShRAubmFggJCVHF3r17h3PnzqB163ZqTQ8A6OrqYvDgYYiJicGhQ/sBAAkJCdi7dxeaNGmm1vSk0tPTg5OTS6Z5REVF4ZdfFqNz5/Zo3twF3bt/i3Xr/FSPd+vWAcuXL1Lb58yZk3B1rY+3b98AAN6+fQNX1/o4evQQFi2ai/btW2DQoL4IDPRH+/YtkJysfu2WJ08ew9W1Pq5cuaiKXb58EYMH94O7e2N4erbE0qULEBcXl2nuAHD8+BE0a+auFnv+/BlmzpyCLl080KJFY3h5dcf27b+p3WYho5yBlFE0f//V6NrVE82bu6B3727488/jas9x584tTJo0Fh07tkXLlq7o3/97HD9+5Kv55pagoMuoX7+hqukBgGbN3KGvr48rVy5luq9CkQxjY2O1mKmpKQTh07XlzM3N0yzNsLOrDB0dHXz8+EEt3rx5CwQHP8KjRw+y+Wo0wxEfEUVHJ2Ly5NP44497MDMzwKlTXrCxMRM7LaJCTf/ZPhj/Ow+SJHHu3yXomSCmznQk2nbK0XFiY2Mhl0fC2tpaFbt58x8IgqA2lfW5unW/gYmJKW7c+Af9+g3E/fv/IS4uFi4ujbOVQ2JiIkaPHoq3b99iwIDBsLOrjPfvQ3Dr1s1sHc/f3w+NGjXFrFnzoFAoYG1dDhs3BuCvv66gceMmqu1OnPgfzMzM0KCBM4CURmrmzKlo374DBg4cgtDQj1i3zg9RUXLMnr0gw+d7+fIF3r17i1q16qjFP3x4DxubCmjVqh2KFSuGx48fIjDQH/HxcRgwYHCmOQPAjBmTcevWvxgwYDAqVKiAK1cuYc6cn2Bqaqqq9bt3b1GzZm106tQV+voGuH37XyxcOAeCIKBdO89M6/RlI5geqVSa6RWOnz17Cg+Pb9Vi+vr6KFu2HJ4/f5rhfrq6umjf/lvs2fMHatasg4oVK+H+/Xs4dGgfunbtmWlOd+7cgkKhSDOaVLGiHUxMTHHt2l+oUsXhq68tp/JF4/P06VPMnTsX//zzD4yMjODh4YEJEybA0PDrQ4b79u2Dv78/Xr9+DVtbW4wYMQLt2rXLg6xz5t69Dxg8+AgePQoDAERGJuDQoUcYMaK+yJkRFW7F7q6EbuRD0XPITuOT+oGX+sFubGyC7t2/Uz3+4UPKX9KlSmV8k+LSpcvgw4f3AICPH1P+LVmylMa5ACmjJQ8fPsC6db/C0bGWKv61D+6M2NtXxaRJ09LETp78n1rjc+rUn2jWrAV0dXUhCAJWr14Jd/dWmDz5J9U2lpaW+PHHsejXbxAqVUr/mmf3798DAFSqpD46Vr9+Q9Sv3xBAyr2iatWqg/j4eOzZ80eaxufLnK9f/xsXL57H8uV+aNgwpTFr0MAZHz58wK+/+qsan5Yt26j2EQQBtWvXxfv3IThwYG+m9Xv79g26d/82w8dTDRgwGAMHDsnw8agoOUxM0i5aNzU1hVwuz/TYEyZMxtKlCzBkSH9VrFu3nmlq87nk5GT4+i6DjY2t2vcSSFneUblyFdy7dyfT580tojc+crkc/fr1Q9myZeHr64uwsDAsWLAAERERWLp0aab7Hj9+HJMnT4a3tzcaN26MkydPYuzYsTA1NYWra9ozHfIDQQC2br2FadPOID4+5a8DY2M9LFvWCl26VBU5O6LCL9ZxDIxvzhV1xCe2xmiN94uLi0OzZs6qr3V0dLBw4TKUL6/5PfpSpyBSpyaye7boP/9cRYUKFdWanpxwdk478tSyZRts3BigWj9z794dvHnzGq1atQUAvHz5HO/evYWPz3i1kZA6db6BRCLBgwf/Zdj4hIZ+hFQqhUymfi+3hIQE/PbbJvz55zGEhLxTO25sbCyKFSuWYc5XrwZBJjNDvXr11fb75psG+OWXxVAoFNDR0YFcLsevv/rjwoVz+Pjxg2q0yMws81H/EiWssGHDlky3Sd3ua9L/tgsZxD9Zt84Ply9fwI8/ToONjS0ePPgPgYHrYWoqy7DZWr58EZ48Ccbq1QHQ1dVNs57KzMwMoaGhX805N4je+OzYsQNyuRz79++HpaUlgJQf6AkTJmDYsGGws8v46sQrV65E27ZtMX78eACAs7Mznj59Cl9f33zZ+MjjDeC9wx07/z6pijk6WmHDBk9UqmQhYmZERUeibaccTzOJwcDAAKtXB0CpVOLVq5dYt84Pc+bMxJYtO1GiRMpZn1ZWKR92ISFvUaWKfbrHeffureqUdiurUv+//bts5RQZGYnixXPvYqoWFpZpYi1btsbatb64ePECWrRohZMn/4eSJUuhdu26AICIiAgAwNSpE9I9ZmavLTExETo6OmmmhNauXYVDh/ZhwIDBcHCoBlNTU1y4cA6bNwciMTFRrfH5MufIyAjI5ZFqTernQkM/omTJUpg/fxbu3LmF/v0HoWJFOxgbG2Pfvt04ffpEhvkCKeuuKldO/3v7ua/dyNPUVIaoqKg08aioaNjapr+wGUhZX7V9+1YsXLgMrq5uAIA6depBIpFi9eoV6NKle5qa/Prrehw5chDz5i1G1arV0z2uvr4BEhISvvaycoXojc/58+fh4uKianoAoE2bNpg6dSrOnTuXYePz8uVLPHnyBOPGjVOLe3p6YsqUKQgLC1M7pthuvCyOXgFd8fhjcVVswIDamD3bDYaGon8biCifk0qlqg+N6tUdYWNTAd7e/bBpUwAmTJgC4NMox8WL51UfSp/7998biI6OQt263wAAqlathmLFjBEUdAkdOnTSOCczMzMEBz/OdBt9fX0kJamvScloKiW9kYbUJufUqT/RvHkLnDlzCi1atFaNUslkKSMkY8f+iBo1HNPsn9nIh0wmQ1JSEhISEqCra6SKnzlzEh07doGXV39V7PLli+kcIW3OpqYymJtbYOnSlelub2FhiYSEBFy5cgkjRoxRnRkFQG1xcEZya6qrQoWKadbyJCYm4s2bV2nW/nzu6dOUfSpXVl+LU6WKPRQKBd69e6vW+Ozduwu//roeEydOTfc9mSoqSv7V0a7cIvonbnBwMLp27aoW09fXh42NDYKDM74J55MnTwAAlSpVUovb2dlBEAQ8efIk242Prm7unuyWkJCMb9e0xZvIlFXwpqb68PVtg44dtb+Iq6jR0ZGq/UvaURDqrFQW/It9pn6optcQVK1aDS1btsHRo4cwYMBgFC9eAqVLl4abW3McP34EPXp8p7Z2JTk5GQEBa2FsbKxqcgwMDNC5czds374V164FqRYLf77P9et/q9aqfKl+fSecOnUCd+7cVjvV/nNWVqXSfMBeu/ZXFiuQomXL1vD1XY7Lly/gw4f3qmkuALC1rYCSJUvhzZvX6Nq1h0bHtbGpAAB49+4N7OzsIJGkLEdIaYQ+nVGrUChw6tSfWTpmgwYN8fvvW6Crq4fKlauku010dDQUCgX09D49R2xsDC5ePP/V4+fWVJezcyNs3hyIyMgImJmZAwDOnz+DxMTETBe7ly6dsn7swYP/ULp0aVX8/v3//v/xsqrYyZP/w4oVSzBo0FB07NgFgPp7+vM+7+3bN6p1VZnR0ZHk+DNa9MZHLpenmV8FUjrxyMjIDPdLfezLfVM7xsz2zYxUKoGFhfHXN9TQun7/4FvfpqhfIRQ7T83m1JaWyWRGX9+Iciw/1zk+XgcfP0pz5Rel2FJHN758HQMHDsbJk//D7t07MGKEDwDgxx+n4unTJxg1aij69k25gGF4eBj++GM77t69g3nzFqFkyU8fikOGDMODB/cwadI4dOnSHU5OztDXN8CTJ4+xe/cfcHSsiUaN0p7qDgAeHp7Yt283Jk0ai4EDvWFnZ4f379/j5s3rmDIlZaFxixYtsXjxfGzalHKhxcuXL+D+/bsAUhpnXV2pWiOd3veqVavWWLFiKZYtWwgbG1vUqKE+XTJ69DjMmDENiYnxaNTIFUZGRnj37i0uXbqIYcNGwsbGNt38a9asCR0dXTx8eB92dnaqPJycnHH48H7Y2dnBwsICu3f/gaSkpP//HkgyzdnFpRFcXZtiwoRR8PLqh8qVqyAuLg5Pnz7By5cvMW3aDJiby1C9eg1s27YJxYtbQkdHB1u2bISJiQnCw8Myfb/q6hrA0THtyJamunbtjj17/sCUKRPwww+DEB4ehpUrl6NNm3aoXPnTTMu8ebNx9OhhXLp0DQDg6FgD1as7YunSBYiMDIetrS3u3buLX38NQMuWrWFllTKrcf36P5g7dyZq164LZ2dn3L//aeGyubkFypUrr/o6JiYaL148x+DBQzN87UqlBFKpFGZmxbJ04lNmRG98MiIIQpYW3H25TU4X6ymVAuTy2Gztm5kWAwZiv852NO3VE7DQR3h4TK4/B6X8EpLJjCCXx0GhyNnFyChjBaHOiYkJUCqVUCiEHF+YTiwSSUqtU3+vffk6rK1t0LJlG+zduwu9e/eHiYkJZDJz+PtvwtatG7Fv326sW/cOxYoVQ61adbFu3UZUrVpN7ThSqS6WLl2F/ft34/jxozhwYK/qVHI3N3f07Pl9hvWTSnWxYsUarF+/Bps3B0Iul8PKqiRatmyj2qd9+2/x8uVL7N27Gzt2bEOLFq0xaNAwzJ07EwqFEsnJStV7KPXrLxkby9CggROuXLmEAQMGp9nGza0Fli41xubNv+L48aMAUkYmnJwaQSazyDB/PT0DODu74PLlS2jXzgMKhRKCAIwePQFLlizAsmWLYWhoiHbtPNGkSTMsWjQXycnCV3OeM2cRfvttE3bv3oWQkLcwNjZBpUp2aN++g2rbGTPmYvHiefj55xmQyczQrVsvxMXFYseO3/Lk/WpkZIyVK9dixYolmDw55SzqlFtWjFJ7/uRkBRQKxWcxCRYtWo6AgLXYunUTQkNDUbJkKXTt2hN9+/6g2u7atatITk7GjRv/YNCg/mrP3b59B0ybNlM14nPx4iUYGhqiYUOXDF+7QiFAqVQiMjIWcXGKNI/LZEZZHoGWCFmZVNQiFxcXdO3aFRMmqC9M8/DwQJ06dTBv3rx09zt37hy8vb1x9OhRtXVAt27dQvfu3bFt2zbUr6/5qeEKhRJhYbnflOjqSmFhYYzw8JgC+0u4IGCd80ZBqHNSUiJCQ9+iePEy0NMruDf21dWV5tsaFwYXL57H7NnTcfToCejpGYidTpHw5Xt66tSJMDExyfTWKF/7eba0NM5y4yP6+K+dnV2atTyJiYl48eJFpmd0pa7tSV3rkyo4OBgSiSTN2h8iIqIvNW7cBOXL22D//j1ip1IkvX79CkFBl9Gv38A8e07RG5+mTZsiKCgI4eHhqtiJEyeQmJgIN7eMV4CXL18elSpVwtGjR9Xihw8fRq1atfLVGV1ERJQ/SSQSTJw4BUZG+Xe9WmH28eMHTJo0DdbW5fLsOUVf49OrVy/89ttvGD58OIYPH47Q0FAsXLgQHTp0UBvxmTp1Kvbv34979+6pYj4+Phg7dixsbGzQqFEjnDp1CpcuXcKGDRvEeClERFQAVatWAzVr1uSUoghq166ruiZTXhG98ZHJZNi8eTPmzp2LUaNGwdDQEJ6enmnW/KQsUlRf0NSuXTvEx8dj3bp1CAwMhK2tLX755Zd8efFCIiIiEp/oi5vzGy5uLthY57xREOqcuhjS0rI09PUL7qJVLm7OG6xz3slOrRMTExAW9q5wLG4mItIGHR0dACm/MImoYEv9OdbRyflElehTXURE2iCV6sDIyATR0SknTujrG2T7+l5iUiolUCg4MK9trHPe0aTWgiAgMTEB0dHhMDIy+eo9yLKCjQ8RFVoyWcrZnanNT0EklUqhVHIKRttY57yTnVobGZmofp5zio0PERVaEokEZmbFYWpqAYUi+es75DM6OhKYmRVDZGQsRyO0iHXOO9mptY6Obq6M9KRi40NEhZ5UKoVUWvCu3qyrK4WhoSHi4hRceKtFrHPeyQ+15uJmIiIiKjLY+BAREVGRwcaHiIiIigxewPALgiBAqdROSXR0pFAoOH+sbaxz3mCd8wbrnDdY57yjjVpLpZIsX66CjQ8REREVGZzqIiIioiKDjQ8REREVGWx8iIiIqMhg40NERERFBhsfIiIiKjLY+BAREVGRwcaHiIiIigw2PkRERFRksPEhIiKiIoONDxERERUZbHyIiIioyGDjQ0REREUGGx8iIiIqMtj45NDTp08xcOBA1KlTBy4uLpg7dy7i4+OztO++ffvQtm1b1KxZE56enjh27JiWsy3YslPr6OhorFq1Ct27d0f9+vXh7OyMgQMH4u7du3mUdcGTk/d0qhMnTsDBwQGenp5ayrLgy0mdIyIiMGvWLLi6uqJmzZpo06YNduzYoeWMC6bs1jk2NhZLly5Fy5YtUbt2bbRu3RqrVq1CYmJiHmRd8Dx//hwzZsxAx44dUb16dY1+9vP6s1BXq0cv5ORyOfr164eyZcvC19cXYWFhWLBgASIiIrB06dJM9z1+/DgmT54Mb29vNG7cGCdPnsTYsWNhamoKV1fXPHoFBUd2a/3mzRvs3LkTXbt2hY+PD5KTk7Flyxb06tULO3bsQI0aNfLwVeR/OXlPp4qPj8eCBQtQokQJLWdbcOWkzjExMejTpw8MDAwwdepUFC9eHM+fP0dSUlIeZV9w5KTOs2bNUv1erlKlCm7dugVfX19ERkZi+vTpefQKCo5Hjx7h3LlzqF27NpRKJQRByNJ+onwWCpRt/v7+Qu3atYXQ0FBV7ODBg4K9vb3w+PHjTPdt27at4OPjoxb74YcfhO7du2sl14Iuu7WOiYkRYmNj1WLx8fFC48aNhcmTJ2st34IqJ+/pVCtWrBB69+4tTJo0SfDw8NBWqgVaTuq8bNkyoWXLlkJcXJy20yzwslvnpKQkoWbNmsLKlSvV4jNnzhRcXFy0lm9BplAoVP+vyc++GJ+FnOrKgfPnz8PFxQWWlpaqWJs2baCvr49z585luN/Lly/x5MmTNEOBnp6euHXrFsLCwrSWc0GV3VoXK1YMRkZGajEDAwPY2dnh/fv3Wsu3oMpunVO9ePECGzdu5F/EX5GTOu/ZswfdunWDoaGhttMs8LJbZ0EQoFAoYGpqqhaXyWRZHskoaqRSzdsJsT4L2fjkQHBwMOzs7NRi+vr6sLGxQXBwcIb7PXnyBABQqVIltbidnR0EQVA9Tp9kt9bpiY2NxX///Zem/pTzOs+bNw8dO3ZE1apVtZVioZDdOr98+RIfP36ETCbDkCFD4OjoCCcnJ8yePVvjdVhFQXbrrKenhy5dumDr1q34999/ERMTg6CgIPzxxx/o3bu3ttMuMsT6LOQanxyQy+WQyWRp4jKZDJGRkRnul/rYl/uamZmpPU6fZLfW6VmxYgXi4uLg5eWVW+kVGjmp8+nTp3Hjxg0cP35cW+kVGtmt88ePHwEAixcvRtu2bREQEIDHjx9j+fLlSEpKwty5c7WWc0GUk/fzrFmzMHPmTPTo0UMV69OnD0aOHJnreRZVYn0WsvHRAkEQIJFIvrrdl9ukDqFmZV9KkdVapzp06BA2b96MGTNmwNbWVouZFS5fq3NCQgLmz5+PUaNGqU0rkGa+VmelUgkg5S/iBQsWAABcXFyQnJyMxYsXY/To0bCyssqTXAuyrPzeWLp0Kc6ePYs5c+agYsWKuHv3Lnx9fSGTyeDj45NHmRYNef1ZyKmuHJDJZJDL5WniUVFR6f6VkSqjbjb1WJntW1Rlt9afu3TpEqZMmYKBAwdyuDoD2a3z5s2bIZVK4eHhAblcDrlcjqSkJCiVSsjlcp4C/IXs1tnc3BwA4OzsrBZ3dnaGUqnUeNq3sMtunR8+fIhff/0Vs2fPRo8ePdCgQQP0798fo0ePhr+/P0JDQ7WZdpEh1mchG58csLOzS/OLJjExES9evEgzr/y51PnML+cvg4ODIZFIuPYkHdmtdapbt25h5MiRaNu2LSZOnKitNAu87Nb5yZMneP78OVxcXNCgQQM0aNAAhw8fRnBwMBo0aIA9e/ZoO/UCJbt1Ll++PPT09NLEU/9Czs4C08Isu3V+/PgxAKBatWpq8WrVqiE5ORmvX7/O/WSLILE+C/lTkgNNmzZFUFAQwsPDVbETJ04gMTERbm5uGe5Xvnx5VKpUCUePHlWLHz58GLVq1eJUQTqyW2sg5Ydo8ODBqFevHhYsWMCpxExkt86DBw/Gli1b1P5zdXWFtbU1tmzZAnd397xIv8DIbp319fXRuHFjXLlyRS1+5coV6OrqonLlylrLuSDKbp2tra0BIM2FTu/cuQMAKFeunBayLXpE+yzU2onyRUBkZKTQpEkToVevXsL58+eFffv2CU5OTsL48ePVtpsyZYpQrVo1tdjRo0cFBwcHYfny5UJQUJAwb948wcHBQbhw4UJevoQCI7u1/vjxo+Dm5iY0btxYuHz5snDjxg3Vf3fv3s3rl5Hv5eQ9/SVexydjOanzv//+K9SoUUOYOHGicOHCBWHjxo1C7dq1hXnz5uXlSygQslvn5ORkoVu3boKLi4vw+++/C1euXBHWr18v1KlTRxgzZkxev4wCITY2Vjh27Jhw7NgxwcvLS3Bzc1N9nXodpfzyWcjFzTkgk8mwefNmzJ07F6NGjYKhoSE8PT0xYcIEte2USiUUCoVarF27doiPj8e6desQGBgIW1tb/PLLL7xqcwayW+vHjx/j7du3AID+/furbWttbY3Tp09rPfeCJCfvacq6nNS5Vq1a8Pf3x7JlyzB06FCYm5vDy8sLo0ePzsuXUCBkt846OjpYt24dVq5ciYCAAHz8+BFlypSBl5cXhg4dmtcvo0AIDQ1N8x5M/XrLli1wcnLKN5+FEkHg1ZiIiIioaOAaHyIiIioy2PgQERFRkcHGh4iIiIoMNj5ERERUZLDxISIioiKDjQ8REREVGWx8iIiIqMhg40NERERFBhsfogJq7969cHBwSPe/RYsWZfk4r169goODA/bu3avFbNN/ztT/qlatCicnJwwePBg3btzQynP26dMHffr0UX0dFxeHVatW4a+//kqzbWptX716pZVcMvLXX3+p1aVatWpwdnbG0KFDcfv27Wwfd9u2bXn6/SXKz3jLCqICbsGCBWnuYlyyZEmRstFMnz594OnpCYVCgcePH8PPzw99+/bFzp07Ub169Vx9rpkzZ6p9HRcXBz8/P4wcORJOTk5qjzVr1gw7d+4UrY7jxo2Dk5MTkpOTce/ePaxevRp9+vTB/v37UaFCBY2Pt337dlhYWKBLly65nyxRAcPGh6iAq1KlCmrWrCl2GtlSpkwZ1KlTBwDwzTffwMbGBv3798fvv/+OuXPn5upzaXLncktLS+3dGToLbG1tVXWpX78+ZDIZJk2ahIMHD8LHx0e0vIgKA051ERVSz58/x5QpU9C6dWvUrl0bTZo0wdChQ/HgwYOv7hsWFoaffvoJbm5ucHR0hLOzM3r16oXLly+rbXf58mX069cP9erVQ+3atdGrVy9cuXIl2zmnfti/efNGFdu9eze+/fZb1KxZEw0bNsSIESMQHBystt/Lly8xduxYuLq6wtHREY0aNUK/fv3w33//qbb5fKrr1atXcHFxAQD4+fmpppYmT54MIO1U17x581CnTh1ER0enyXnMmDFo1KgRkpKSVLGjR4+iZ8+eqFOnDurWrYuBAwfi3r172a6Lo6MjAODjx49qcT8/P3Tv3h0NGzZEvXr10LlzZ+zatQuf34LR3d0djx49wtWrV1Wv093dXfV4dHQ0Fi1aBHd3dzg6OqJJkyaYN28eYmNjs50vUX7GER+iAk6pVCI5OVktpquri/fv38Pc3Bzjx4+HpaUlIiMjsW/fPvTo0QP79u1LMz32uYkTJ+LevXsYO3YsKlSoALlcjnv37iEiIkK1zYEDBzBp0iS0aNECixYtgq6uLnbu3ImBAwciMDBQ1Vho4vnz5wAACwsLAIC/vz+WL18OT09PjB8/HuHh4fDz80PPnj2xe/du1bTP4MGDoVQqMXHiRJQtWxbh4eG4ceMG5HJ5us9TsmRJbNiwAYMGDUK3bt3QvXt3AMhwlKdr167YsmULjh07ptoWAORyOU6dOoXevXtDT08PALBu3TqsWLECXbp0wbBhw5CUlITAwED07t0bu3bt0mjkKVVqA1axYkW1+OvXr9GzZ0+ULVsWAHDz5k3MnTsXISEhGDlyJICU5sjHxwempqaq6T59fX0AKdN9Xl5eePfuHYYOHQoHBwc8evQIvr6+ePjwITZt2gSJRKJxvkT5mkBEBdKePXsEe3v7dP9LSkpKs31ycrKQmJgotG7dWpg/f74q/vLlS8He3l7Ys2ePKlanTh1h3rx5GT53bGys0LBhQ2HIkCFqcYVCIXz77bdCt27dMs099TnXr18vJCUlCQkJCcKdO3eErl27Cvb29sLZs2eFyMhIoVatWsLgwYPV9n3z5o3g6OgojBs3ThAEQQgLCxPs7e2FTZs2ZfqcXl5egpeXl+rr0NBQwd7eXvD19U2zbWptX758qYp17txZ6Nmzp9p227ZtE+zt7YUHDx6ocqtevbowZ84cte2io6OFxo0bC6NHj840x6CgIMHe3l44cuSIkJSUJMTFxQn//POP0KZNG6F9+/ZCZGRkhvsqFAohKSlJ8PPzExo2bCgolUrVYx4eHmqvPZW/v79QtWpV4datW2rx48ePq74PRIUNR3yICrhFixbBzs5OLaarq4vk5GRs2LABBw8exIsXL9SmYr6cKvpSrVq1sG/fPpibm6NRo0aoUaOGakQDAG7cuIGIiAh07tw5zWhTkyZNsGHDBsTGxqJYsWKZPs/SpUuxdOlS1dclSpTAzz//DDc3N5w7dw7x8fHo3Lmz2j5lypSBs7MzgoKCAADm5uawsbFBYGAglEolnJycULVqVUiluTuT36VLF8yZMwdPnjxRjZbt3bsXNWvWhL29PQDg4sWLSE5ORseOHdXqYmBggAYNGqR7Bll6xo4dq/a1lZUVduzYAZlMpha/cuUK/P39cfv27TTTcKGhoShRokSmz3PmzBlUqVIF1apVU8vX1dUVEokEV69ehZubW5ZyJioo2PgQFXB2dnbpLm5euHAhtm3bhsGDB6NBgwYwMzODRCLB9OnTkZCQkOkxf/nlF6xduxa7d+/GypUrUaxYMbRq1QoTJ06ElZWVaq1JZgttIyMjv9r49O3bF99++y2kUilkMhnKlSunmlpJnVazsrJKs1/JkiVV640kEgk2bdqE1atXY8OGDVi4cCHMzc3RoUMHjBkzBiYmJpnmkFUdOnTAokWLsG/fPowfPx6PHz/G7du31c4WS61Lt27d0j1GVpuxCRMmwNnZGfHx8bh48SLWr1+PESNGYNeuXappqlu3bmHgwIFo2LAh5syZg9KlS0NPTw8nT57EunXrEB8f/9XnCQ0NxfPnz1GjRo10Hw8PD89SvkQFCRsfokLq4MGD6NSpE8aNG6cWDw8PTzNy8CVLS0tMmzYN06ZNw5s3b3D69GksW7YMoaGhCAwMVK3B+emnn1C7du10j1G8ePGv5li6dOkMz0gzNzcHAHz48CHNY+/fv1flAADW1taYP38+AODp06c4duwY/Pz8kJiYiJ9//vmreWSFmZkZWrRogf3792PMmDHYs2cPDAwM4OnpqdomNSdfX1/VupvsKF++vKouDRo0gKGhIVasWIGtW7di4MCBAIAjR45AV1cX/v7+MDAwUO178uTJLD+PhYUFDAwMVLVL73GiwoaND1EhJZFI1KanAODs2bMICQmBra1tlo9TtmxZeHl54cqVK7h+/ToAoF69epDJZHj8+DG8vLxyNe9UdevWhaGhIQ4ePIh27dqp4u/evUNQUBDatGmT7n4VK1bE8OHD8eeff2Z6JlXqyElWRkZSdenSBceOHcO5c+dw6NAhtGrVSq2JdHV1ha6uLl68eJFhftkxaNAg7Nu3D+vXr0fPnj1hYmICiUQCHR0dtVGk+Ph4HDx4MM3++vr66b7OZs2awd/fH+bm5ihfvnyu5UuUn7HxISqkmjVrpjp7y8HBAXfv3kVgYCBKly6d6X5RUVHo27cvPD09UalSJRgbG+P27du4cOECWrVqBQAwNjbG9OnTMXnyZERGRqJNmzYoXrw4wsLCcP/+fYSFhWH27Nk5yl8mk2H48OFYvnw5fvzxR3h4eCAiIgKrV6+GgYGB6qyl+/fvY86cOWjbti1sbW2hp6eHoKAgPHjwAN7e3hke38TEBNbW1jh16hRcXFxgZmYGCwsLlCtXLsN9XF1dUbp0acyePRsfPnxIc0HAcuXKwcfHBytWrMDLly/RtGlTyGQyfPz4Ebdv34aRkVG2rsOjp6eHsWPHYsyYMdiyZQuGDx8ONzc3bNy4EePHj0fPnj0RERGBwMBAVUP3OXt7exw5cgRHjx5FuXLlYGBgAAcHB/Tr1w9//vknvLy80L9/fzg4OECpVOLt27e4ePEifvjhhwxH9IgKKjY+RIXUtGnToKuri/Xr1yM2NhbVq1fHqlWrsHLlykz3MzAwQK1atXDgwAG8fv0aycnJKFOmDAYPHoxBgwaptuvYsSPKli2LDRs2YObMmYiJiYGlpSWqVauWZkFydg0ZMgSWlpbYunUrjh49CkNDQzRs2BDjxo1TncpuZWUFGxsb/P7773j37h2AlKmiSZMmqd2iIj3z5s3D4sWLMWzYMCQmJqJz585YuHBhhttLpVJ06tQJ69atQ5kyZdI9ZX/IkCGws7PDli1bcOTIESQmJsLKygqOjo747rvvsl2Ldu3aYePGjdi0aRP69OkDFxcXzJ8/HwEBARg6dChKlSqFHj16qKYpPzdq1Ch8+PAB06dPR0xMDKytrXH69GkUK1YM27Ztw/r167Fz5068evUKhoaGKFOmDBo1agRra+ts50uUX0kE4bMrXREREREVYrxyMxERERUZbHyIiIioyGDjQ0REREUGGx8iIiIqMtj4EBERUZHBxoeIiIiKDDY+REREVGSw8SEiIqIig40PERERFRlsfIiIiKjIYONDRERERcb/AfUdfHeXpwt1AAAAAElFTkSuQmCC",
"text/plain": [
""
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"plt.figure()\n",
"lw = 2\n",
"plt.plot(fpr, tpr, color='darkorange',\n",
" lw=lw, label='ROC curve (area = %0.2f)' % roc_auc)\n",
"plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')\n",
"plt.xlim([-0.05, 1.05])\n",
"plt.ylim([-0.05, 1.05])\n",
"plt.xlabel('False Positive Rate')\n",
"plt.ylabel('True Positive Rate')\n",
"plt.title('Receiver operating characteristic example')\n",
"plt.legend(loc=\"lower right\")\n",
"plt.show()"
]
}
],
"metadata": {
"colab": {
"collapsed_sections": [],
"include_colab_link": true,
"name": "notebook2.ipynb",
"provenance": []
},
"interpreter": {
"hash": "625a8f875bfb3f569e4f618df17e1f8389970b6b26ee2c84acb92c5fbebf95c3"
},
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.10.13"
}
},
"nbformat": 4,
"nbformat_minor": 4
}