harml.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/arshanrahman/anaconda/lib/python3.6/site-packages/sklearn/cross_validation.py:41: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. Also note that the interface of the new CV iterators are different from that of this module. This module will be removed in 0.20.\n",
      "  \"This module will be removed in 0.20.\", DeprecationWarning)\n",
      "/Users/arshanrahman/anaconda/lib/python3.6/site-packages/sklearn/grid_search.py:42: DeprecationWarning: This module was deprecated in version 0.18 in favor of the model_selection module into which all the refactored classes and functions are moved. This module will be removed in 0.20.\n",
      "  DeprecationWarning)\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.cross_validation import train_test_split\n",
    "from sklearn.cross_validation import cross_val_score\n",
    "from sklearn.grid_search import RandomizedSearchCV\n",
    "from sklearn.grid_search import GridSearchCV\n",
    "from sklearn import metrics\n",
    "from sklearn.preprocessing import normalize\n",
    "from scipy.stats import randint as sp_randint\n",
    "import matplotlib.pyplot as plt\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nCOLUMNS = [\\'gx\\',\\'gy\\',\\'gz\\', \\'tax\\', \\'tay\\', \\'taz\\']\\n\\nX_train = pd.DataFrame(COLUMNS)\\nfor i in range(0,6):\\n    X_train[i] = pd.read_csv(\\'data/train/\\'+INPUT_DATA[3+i]+\\'train.txt\\', sep=\" \", header=None)\\n    X_test[i] = pd.read_csv(\\'data/test/\\'+INPUT_DATA[3+i]+\\'test.txt\\', sep=\" \", header=None)\\n\\n\\n\\n#X_train = pd.read_csv(\\'data/X_train.txt\\', sep=\" \", header=None)\\nY_train = pd.read_csv(\\'data/Y_train.txt\\', sep=\" \", header=None)\\n\\n#X_test = pd.read_csv(\\'data/X_test.txt\\', sep=\" \", header=None)\\nY_test = pd.read_csv(\\'data/Y_test.txt\\', sep=\" \", header=None)\\n\\nprint(X_train.shape)\\nprint(Y_train.shape)\\nprint(X_test.shape)\\nprint(Y_test.shape)\\n\\nX_train.head()\\nY_train.head()\\n\\nX = X_train[0:1000]\\nY = Y_train[0:1000]\\nXt = X_test[0:300]\\nYt = Y_test[0:300]\\n\\nprint(X.shape)\\nprint(Y.shape)\\nprint(Xt.shape)\\nprint(Yt.shape)\\n'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "INPUT_DATA = [\n",
    "    \"body_acc_x_\",\n",
    "    \"body_acc_y_\",\n",
    "    \"body_acc_z_\",\n",
    "    \"body_gyro_x_\",\n",
    "    \"body_gyro_y_\",\n",
    "    \"body_gyro_z_\",\n",
    "    \"total_acc_x_\",\n",
    "    \"total_acc_y_\",\n",
    "    \"total_acc_z_\"\n",
    "]\n",
    "\n",
    "LABELS = [\n",
    "    \"WALKING\", \n",
    "    \"WALKING_UPSTAIRS\", \n",
    "    \"WALKING_DOWNSTAIRS\", \n",
    "    \"SITTING\", \n",
    "    \"STANDING\", \n",
    "    \"LAYING\"\n",
    "] \n",
    "\n",
    "TRAINING_SIZE = 1000\n",
    "TESTING_SIZE = 300\n",
    "\n",
    "'''\n",
    "COLUMNS = ['gx','gy','gz', 'tax', 'tay', 'taz']\n",
    "\n",
    "X_train = pd.DataFrame(COLUMNS)\n",
    "for i in range(0,6):\n",
    "    X_train[i] = pd.read_csv('data/train/'+INPUT_DATA[3+i]+'train.txt', sep=\" \", header=None)\n",
    "    X_test[i] = pd.read_csv('data/test/'+INPUT_DATA[3+i]+'test.txt', sep=\" \", header=None)\n",
    "\n",
    "\n",
    "\n",
    "#X_train = pd.read_csv('data/X_train.txt', sep=\" \", header=None)\n",
    "Y_train = pd.read_csv('data/Y_train.txt', sep=\" \", header=None)\n",
    "\n",
    "#X_test = pd.read_csv('data/X_test.txt', sep=\" \", header=None)\n",
    "Y_test = pd.read_csv('data/Y_test.txt', sep=\" \", header=None)\n",
    "\n",
    "print(X_train.shape)\n",
    "print(Y_train.shape)\n",
    "print(X_test.shape)\n",
    "print(Y_test.shape)\n",
    "\n",
    "X_train.head()\n",
    "Y_train.head()\n",
    "\n",
    "X = X_train[0:1000]\n",
    "Y = Y_train[0:1000]\n",
    "Xt = X_test[0:300]\n",
    "Yt = Y_test[0:300]\n",
    "\n",
    "print(X.shape)\n",
    "print(Y.shape)\n",
    "print(Xt.shape)\n",
    "print(Yt.shape)\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "\n",
    "def feature_extraction(x, y, z):\n",
    "    features = [np.mean(x), np.mean(y), np.mean(z), np.std(x), np.std(y), np.std(z)]\n",
    "    #Median Absolute Deviation\n",
    "    features.extend((np.mean(abs(x - features[0])), np.mean(abs(y - features[1])), np.mean(abs(z - features[2]))))\n",
    "    #Jerk Signals\n",
    "    features.extend((np.mean(np.diff(x)), np.mean(np.diff(y)), np.mean(np.diff(z)), np.std(np.diff(x)), np.std(np.diff(y)), np.std(np.diff(z))))\n",
    "    features.extend((np.mean(abs(np.diff(x) - features[9])), np.mean(abs(np.diff(y) - features[10])), np.mean(abs(np.diff(y) - features[11]))))\n",
    "    #features.extend((max(x), max(y), max(z), min(x), min(y), min(z)))\n",
    "    return features\n",
    "\n",
    "def feature_selection(X):\n",
    "    data = []\n",
    "    for i in range(X.shape[0]):\n",
    "        features = []\n",
    "        for j in range(0, X.shape[2], 3):\n",
    "            x = [X[i][u][j] for u in range(X.shape[1])]\n",
    "            y = [X[i][u][j+1] for u in range(X.shape[1])]\n",
    "            z = [X[i][u][j+2] for u in range(X.shape[1])]\n",
    "            features.append(feature_extraction(x, y, z))\n",
    "        data.append(normalize(features))\n",
    "    return data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(7352, 128, 9)\n",
      "(7352,)\n",
      "(2947, 128, 9)\n",
      "(2947,)\n",
      "[[  1.80851493e-04   1.07668098e-02   5.55606782e-02 ...,   1.01281703e+00\n",
      "   -1.23216704e-01   1.02934100e-01]\n",
      " [  1.01385601e-02   6.57948013e-03   5.51248305e-02 ...,   1.02283299e+00\n",
      "   -1.26875594e-01   1.05687201e-01]\n",
      " [  9.27557424e-03   8.92887823e-03   4.84047309e-02 ...,   1.02202797e+00\n",
      "   -1.24003701e-01   1.02102503e-01]\n",
      " ..., \n",
      " [ -1.14748406e-03   1.71443899e-04   2.64786393e-03 ...,   1.01844501e+00\n",
      "   -1.24069601e-01   1.00385197e-01]\n",
      " [ -2.22265502e-04   1.57418102e-03   2.38105701e-03 ...,   1.01937199e+00\n",
      "   -1.22745097e-01   9.98735502e-02]\n",
      " [  1.57550001e-03   3.07018892e-03  -2.26975698e-03 ...,   1.02117097e+00\n",
      "   -1.21325999e-01   9.49874073e-02]]\n",
      "5\n",
      "(1000, 128, 9)\n",
      "(1000,)\n",
      "(300, 128, 9)\n",
      "(300,)\n",
      "(1000, 3, 18)\n",
      "(300, 3, 18)\n",
      "[[  8.77741983e-02  -6.40452583e-02  -3.14372406e-01   1.13786540e-01\n",
      "    1.77998177e-01   7.15922028e-01   9.71394559e-02   1.40047414e-01\n",
      "    5.13833336e-01   4.24865816e-04  -2.34470271e-03  -1.76174996e-02\n",
      "    7.76501279e-02   8.70301587e-02   1.12325413e-01   5.55094007e-02\n",
      "    6.85380240e-02   6.92759651e-02]\n",
      " [  4.80530126e-01   4.19032305e-01   2.04479192e-01   3.11726081e-01\n",
      "    4.28223692e-01   1.28353260e-01   2.67393595e-01   3.55255563e-01\n",
      "    1.03360230e-01  -2.33550560e-04  -1.12321172e-02  -4.13303802e-03\n",
      "    8.63868156e-02   1.28836513e-01   8.03369635e-02   6.52248612e-02\n",
      "    7.56284188e-02   7.51383073e-02]\n",
      " [  9.87968572e-01  -1.20441740e-01   9.66657749e-02   1.73550856e-03\n",
      "    3.16946992e-03   3.84756373e-03   1.29878684e-03   2.57810150e-03\n",
      "    2.90832584e-03   6.37569038e-05   1.44297551e-05  -6.06487472e-05\n",
      "    1.94465931e-03   2.17041699e-03   2.58690567e-03   1.39224033e-03\n",
      "    1.70803729e-03   1.70744591e-03]]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "def load_X(X_signals_paths):\n",
    "    X_signals = []\n",
    "    \n",
    "    for signal_type_path in X_signals_paths:\n",
    "        file = open(signal_type_path, 'r')\n",
    "        # Read dataset from disk, dealing with text files' syntax\n",
    "        X_signals.append(\n",
    "            [np.array(serie, dtype=np.float32) for serie in \n",
    "             [row.replace('  ', ' ').strip().split(' ') for row in file]\n",
    "            ]\n",
    "        )\n",
    "        file.close()\n",
    "    \n",
    "    return np.transpose(np.array(X_signals), (1, 2, 0))\n",
    "\n",
    "\n",
    "def load_y(y_path):\n",
    "    file = open(y_path, 'r')\n",
    "    # Read dataset from disk, dealing with text file's syntax\n",
    "    y_ = np.array(\n",
    "        [elem for elem in [\n",
    "            row.replace('  ', ' ').strip().split(' ') for row in file\n",
    "        ]], \n",
    "        dtype=np.int32\n",
    "    )\n",
    "    file.close()\n",
    "    return y_\n",
    "\n",
    "\n",
    "X_train_input_paths = [\"data/train/\" + signal + \"train.txt\" for signal in INPUT_DATA]\n",
    "X_test_input_paths = [\"data/test/\" + signal + \"test.txt\" for signal in INPUT_DATA]\n",
    "X_train = load_X(X_train_input_paths)\n",
    "X_test = load_X(X_test_input_paths)\n",
    "\n",
    "y_train_path = \"data/y_train.txt\"\n",
    "y_test_path = \"data/y_test.txt\"\n",
    "Y_train = load_y(y_train_path).ravel()\n",
    "Y_test = load_y(y_test_path).ravel()\n",
    "\n",
    "print(X_train.shape)\n",
    "print(Y_train.shape)\n",
    "print(X_test.shape)\n",
    "print(Y_test.shape)\n",
    "\n",
    "print(X_train[0])\n",
    "print(Y_train[0])\n",
    "\n",
    "X = X_train[0:TRAINING_SIZE]\n",
    "Y = Y_train[0:TRAINING_SIZE]\n",
    "Xt = X_test[0:TESTING_SIZE]\n",
    "Yt = Y_test[0:TESTING_SIZE]\n",
    "\n",
    "print(X.shape)\n",
    "print(Y.shape)\n",
    "print(Xt.shape)\n",
    "print(Yt.shape)\n",
    "\n",
    "X = np.array(feature_selection(X))\n",
    "Xt = np.array(feature_selection(Xt))\n",
    "\n",
    "print(X.shape)\n",
    "print(Xt.shape)\n",
    "\n",
    "print(X[0])\n",
    "#print(Xt[0])\n",
    "\n",
    "np.savetxt(\"data/test_data_format.csv\", X[0], fmt='%5s', delimiter=\",\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'penalty': ['l1', 'l2'], 'C': array([ 0.95759337,  0.94301417,  0.88522036,  0.91193797,  0.93397789,\n",
      "        1.23673891,  1.32307459,  1.07341249,  0.99717614,  1.12592836])}\n",
      "The best performing penalty is: l1\n",
      "The best performing C value is:  1.32\n",
      "Best score\n",
      "0.841\n",
      "Best parameters set found on training set:\n",
      "{'penalty': 'l1', 'C': 1.3230745856698762}\n",
      "\n",
      "[mean: 0.84000, std: 0.06805, params: {'penalty': 'l1', 'C': 0.99717613800052618}, mean: 0.82400, std: 0.06151, params: {'penalty': 'l2', 'C': 1.0734124939596197}, mean: 0.82400, std: 0.05927, params: {'penalty': 'l2', 'C': 0.95759336690378638}, mean: 0.83900, std: 0.06782, params: {'penalty': 'l1', 'C': 0.95759336690378638}, mean: 0.82500, std: 0.05989, params: {'penalty': 'l2', 'C': 0.88522035823829193}, mean: 0.84100, std: 0.07399, params: {'penalty': 'l1', 'C': 1.3230745856698762}, mean: 0.82400, std: 0.06151, params: {'penalty': 'l2', 'C': 1.1259283594817016}, mean: 0.83900, std: 0.06782, params: {'penalty': 'l1', 'C': 0.94301416604124322}, mean: 0.84000, std: 0.07289, params: {'penalty': 'l1', 'C': 1.236738913982327}, mean: 0.82700, std: 0.06407, params: {'penalty': 'l2', 'C': 1.236738913982327}]\n",
      "\n",
      "Accuracy Rate:\n",
      "0.766666666667\n",
      "[[58  1  0  0  0  0]\n",
      " [ 8 38  0  0  0  0]\n",
      " [16  6 25  0  0  0]\n",
      " [ 0  0  0  8 38  0]\n",
      " [ 0  1  0  0 53  0]\n",
      " [ 0  0  0  0  0 48]]\n"
     ]
    }
   ],
   "source": [
    "#Logistic Regression\n",
    "\n",
    "penalty = [\"l1\", \"l2\"]\n",
    "np.random.seed(121)\n",
    "C_range = np.random.normal(1, 0.2, 10).astype(float)\n",
    "\n",
    "# Check that C>0 \n",
    "C_range[C_range < 0] = 0.0001\n",
    "\n",
    "hyperparameters = {'penalty': penalty, \n",
    "                    'C': C_range}\n",
    "\n",
    "print (hyperparameters)\n",
    "\n",
    "clf = RandomizedSearchCV(LogisticRegression(), param_distributions=hyperparameters, cv=3, scoring='accuracy')\n",
    "clf.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "best_penalty = clf.best_params_['penalty']\n",
    "best_C = clf.best_params_['C']\n",
    "\n",
    "print (\"The best performing penalty is: {}\".format(best_penalty))\n",
    "print (\"The best performing C value is: {:5.2f}\".format(best_C))\n",
    "\n",
    "# examine the best model\n",
    "print(\"Best score\")\n",
    "print(clf.best_score_)\n",
    "print(\"Best parameters set found on training set:\")\n",
    "print(clf.best_params_)\n",
    "print()\n",
    "print(clf.grid_scores_)\n",
    "print()\n",
    "\n",
    "Y_pred = clf.predict(Xt.reshape(Xt.shape[0], Xt.shape[1]*Xt.shape[2]))\n",
    "print(\"Accuracy Rate:\")\n",
    "print(metrics.accuracy_score(Yt, Y_pred))\n",
    "print(metrics.confusion_matrix(Yt, Y_pred))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: 84.0607522558742%\n",
      "Recall: 76.66666666666667%\n",
      "f1_score: 73.42779589322753%\n",
      "\n",
      "Confusion Matrix:\n",
      "[[58  1  0  0  0  0]\n",
      " [ 8 38  0  0  0  0]\n",
      " [16  6 25  0  0  0]\n",
      " [ 0  0  0  8 38  0]\n",
      " [ 0  1  0  0 53  0]\n",
      " [ 0  0  0  0  0 48]]\n",
      "\n",
      "Confusion matrix (normalised to % of total test data):\n",
      "[[ 19.33333206   0.33333334   0.           0.           0.           0.        ]\n",
      " [  2.66666675  12.66666603   0.           0.           0.           0.        ]\n",
      " [  5.33333349   2.           8.33333397   0.           0.           0.        ]\n",
      " [  0.           0.           0.           2.66666675  12.66666603   0.        ]\n",
      " [  0.           0.33333334   0.           0.          17.66666603   0.        ]\n",
      " [  0.           0.           0.           0.           0.          16.        ]]\n",
      "Note: training and testing data is not equally distributed amongst classes, \n",
      "so it is normal that more than a 6th of the data is correctly classifier in the last category.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAz8AAAM+CAYAAAA0Cb/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11a2fe518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "print(\"Precision: {}%\".format(100*metrics.precision_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"Recall: {}%\".format(100*metrics.recall_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"f1_score: {}%\".format(100*metrics.f1_score(Yt, Y_pred, average=\"weighted\")))\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion Matrix:\")\n",
    "confusion_matrix = metrics.confusion_matrix(Yt, Y_pred)\n",
    "print(confusion_matrix)\n",
    "normalised_confusion_matrix = np.array(confusion_matrix, dtype=np.float32)/np.sum(confusion_matrix)*100\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion matrix (normalised to % of total test data):\")\n",
    "print(normalised_confusion_matrix)\n",
    "print(\"Note: training and testing data is not equally distributed amongst classes, \")\n",
    "print(\"so it is normal that more than a 6th of the data is correctly classifier in the last category.\")\n",
    "\n",
    "# Plot Results: \n",
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.imshow(\n",
    "    normalised_confusion_matrix, \n",
    "    interpolation='nearest', \n",
    "    cmap=plt.cm.rainbow\n",
    ")\n",
    "plt.title(\"Confusion matrix \\n(normalised to % of total test data)\")\n",
    "plt.colorbar()\n",
    "tick_marks = np.arange(len(LABELS))\n",
    "plt.xticks(tick_marks, LABELS, rotation=90)\n",
    "plt.yticks(tick_marks, LABELS)\n",
    "plt.tight_layout()\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.914\n",
      "{'n_neighbors': 1, 'weights': 'uniform'}\n",
      "0.898\n",
      "{'weights': 'distance', 'n_neighbors': 10}\n"
     ]
    }
   ],
   "source": [
    "#KNN\n",
    "\n",
    "\n",
    "# define the parameter values that should be searched\n",
    "k_range = list(range(1, 31))\n",
    "weight_options = ['uniform', 'distance']\n",
    "param_dist = dict(n_neighbors=k_range, weights=weight_options)\n",
    "\n",
    "knn = KNeighborsClassifier()\n",
    "grid = GridSearchCV(knn, param_dist, cv=4, scoring='accuracy')\n",
    "grid.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "#print(grid.grid_scores_)\n",
    "#grid_mean_scores = [result.mean_validation_score for result in grid.grid_scores_]\n",
    "#print(grid_mean_scores)\n",
    "print(grid.best_score_)\n",
    "print(grid.best_params_)\n",
    "\n",
    "# plot the results\n",
    "#plt.plot(k_range, grid_mean_scores)\n",
    "#plt.xlabel('Value of K for KNN')\n",
    "#plt.ylabel('Cross-Validated Accuracy')\n",
    "\n",
    "\n",
    "rand = RandomizedSearchCV(knn, param_dist, cv=4, scoring='accuracy', n_iter=10, random_state=5)\n",
    "rand.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "#print(rand.grid_scores_)\n",
    "#rand_mean_scores = [result.mean_validation_score for result in rand.grid_scores_]\n",
    "#print(rand_mean_scores)\n",
    "print(rand.best_score_)\n",
    "print(rand.best_params_)\n",
    "\n",
    "#plt.plot(k_range, rand_mean_scores)\n",
    "#plt.xlabel('Value of K for KNN')\n",
    "#plt.ylabel('Cross-Validated Accuracy')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy Rate:\n",
      "0.816666666667\n",
      "[[56  2  1  0  0  0]\n",
      " [ 0 45  1  0  0  0]\n",
      " [ 7 23 17  0  0  0]\n",
      " [ 0  0  0 25 21  0]\n",
      " [ 0  0  0  0 54  0]\n",
      " [ 0  0  0  0  0 48]]\n",
      "Accuracy Rate:\n",
      "0.836666666667\n",
      "[[56  3  0  0  0  0]\n",
      " [ 2 44  0  0  0  0]\n",
      " [ 6 25 16  0  0  0]\n",
      " [ 0  0  0 33 13  0]\n",
      " [ 0  0  0  0 54  0]\n",
      " [ 0  0  0  0  0 48]]\n"
     ]
    }
   ],
   "source": [
    "Y_pred = grid.predict(Xt.reshape(Xt.shape[0], Xt.shape[1]*Xt.shape[2]))\n",
    "print(\"Accuracy Rate:\")\n",
    "print(metrics.accuracy_score(Yt, Y_pred))\n",
    "print(metrics.confusion_matrix(Yt, Y_pred))\n",
    "\n",
    "Y_pred1 = rand.predict(Xt.reshape(Xt.shape[0], Xt.shape[1]*Xt.shape[2]))\n",
    "print(\"Accuracy Rate:\")\n",
    "print(metrics.accuracy_score(Yt, Y_pred1))\n",
    "print(metrics.confusion_matrix(Yt, Y_pred1))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: 85.6495015316068%\n",
      "Recall: 81.66666666666667%\n",
      "f1_score: 79.88979311117632%\n",
      "\n",
      "Confusion Matrix:\n",
      "[[56  2  1  0  0  0]\n",
      " [ 0 45  1  0  0  0]\n",
      " [ 7 23 17  0  0  0]\n",
      " [ 0  0  0 25 21  0]\n",
      " [ 0  0  0  0 54  0]\n",
      " [ 0  0  0  0  0 48]]\n",
      "\n",
      "Confusion matrix (normalised to % of total test data):\n",
      "[[ 18.66666603   0.66666669   0.33333334   0.           0.           0.        ]\n",
      " [  0.          15.00000095   0.33333334   0.           0.           0.        ]\n",
      " [  2.33333325   7.66666698   5.66666698   0.           0.           0.        ]\n",
      " [  0.           0.           0.           8.33333397   7.           0.        ]\n",
      " [  0.           0.           0.           0.          18.           0.        ]\n",
      " [  0.           0.           0.           0.           0.          16.        ]]\n",
      "Note: training and testing data is not equally distributed amongst classes, \n",
      "so it is normal that more than a 6th of the data is correctly classifier in the last category.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAz8AAAM+CAYAAAA0Cb/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11c12a5f8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Precision: {}%\".format(100*metrics.precision_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"Recall: {}%\".format(100*metrics.recall_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"f1_score: {}%\".format(100*metrics.f1_score(Yt, Y_pred, average=\"weighted\")))\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion Matrix:\")\n",
    "confusion_matrix = metrics.confusion_matrix(Yt, Y_pred)\n",
    "print(confusion_matrix)\n",
    "normalised_confusion_matrix = np.array(confusion_matrix, dtype=np.float32)/np.sum(confusion_matrix)*100\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion matrix (normalised to % of total test data):\")\n",
    "print(normalised_confusion_matrix)\n",
    "print(\"Note: training and testing data is not equally distributed amongst classes, \")\n",
    "print(\"so it is normal that more than a 6th of the data is correctly classifier in the last category.\")\n",
    "\n",
    "# Plot Results: \n",
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.imshow(\n",
    "    normalised_confusion_matrix, \n",
    "    interpolation='nearest', \n",
    "    cmap=plt.cm.rainbow\n",
    ")\n",
    "plt.title(\"Confusion matrix \\n(normalised to % of total test data)\")\n",
    "plt.colorbar()\n",
    "tick_marks = np.arange(len(LABELS))\n",
    "plt.xticks(tick_marks, LABELS, rotation=90)\n",
    "plt.yticks(tick_marks, LABELS)\n",
    "plt.tight_layout()\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.349 (+/-0.177) for {'kernel': 'rbf', 'gamma': 100.0, 'C': 3.1622776601683795}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 0.0001, 'C': 0.0031622776601683794}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.0031622776601683794}\n",
      "0.349 (+/-0.177) for {'kernel': 'rbf', 'gamma': 100.0, 'C': 100.0}\n",
      "0.549 (+/-0.029) for {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 0.10000000000000001}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 0.0001, 'C': 3.1622776601683795}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 0.0031622776601683794}\n",
      "0.948 (+/-0.125) for {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 100.0}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.0001}\n",
      "0.209 (+/-0.001) for {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.10000000000000001}\n",
      "The best performing gamma value is:  0.10\n",
      "The best performing C value is: 100.00\n",
      "Best score\n",
      "0.948\n",
      "Best parameters set found on training set:\n",
      "{'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 100.0}\n",
      "\n",
      "[mean: 0.34900, std: 0.08863, params: {'kernel': 'rbf', 'gamma': 100.0, 'C': 3.1622776601683795}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 0.0001, 'C': 0.0031622776601683794}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.0031622776601683794}, mean: 0.34900, std: 0.08863, params: {'kernel': 'rbf', 'gamma': 100.0, 'C': 100.0}, mean: 0.54900, std: 0.01445, params: {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 0.10000000000000001}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 0.0001, 'C': 3.1622776601683795}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 0.0031622776601683794}, mean: 0.94800, std: 0.06271, params: {'kernel': 'rbf', 'gamma': 0.10000000000000001, 'C': 100.0}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.0001}, mean: 0.20900, std: 0.00060, params: {'kernel': 'rbf', 'gamma': 100.0, 'C': 0.10000000000000001}]\n",
      "\n",
      "Accuracy Rate:\n",
      "0.743333333333\n",
      "[[59  0  0  0  0  0]\n",
      " [ 0 33 13  0  0  0]\n",
      " [12 13 22  0  0  0]\n",
      " [ 0  0  0  7 39  0]\n",
      " [ 0  0  0  0 54  0]\n",
      " [ 0  0  0  0  0 48]]\n"
     ]
    }
   ],
   "source": [
    "#SVM\n",
    "\n",
    "param_dist = {\n",
    "    'kernel': ['rbf'],\n",
    "    'gamma': np.logspace(-4, 2, 3),\n",
    "    'C': np.logspace(-4, 2, 5),\n",
    "}\n",
    "svm = SVC(probability=True)\n",
    "clf = RandomizedSearchCV(\n",
    "    estimator=svm,\n",
    "    param_distributions=param_dist,\n",
    "    cv=5,\n",
    ")\n",
    "clf.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "for params, mean_score, scores in clf.grid_scores_:\n",
    "    print(\"%0.3f (+/-%0.03f) for %r\" % (mean_score, scores.std() * 2, params))\n",
    "\n",
    "best_gamma  = clf.best_params_['gamma']\n",
    "best_C      = clf.best_params_['C']\n",
    "\n",
    "print(\"The best performing gamma value is: {:5.2f}\".format(best_gamma))\n",
    "print(\"The best performing C value is: {:5.2f}\".format(best_C))\n",
    "\n",
    "print(\"Best score\")\n",
    "print(clf.best_score_)\n",
    "print(\"Best parameters set found on training set:\")\n",
    "print(clf.best_params_)\n",
    "print()\n",
    "print(clf.grid_scores_)\n",
    "print()\n",
    "    \n",
    "svm = SVC(kernel='rbf', C=clf.best_params_['C'], gamma=clf.best_params_['gamma'])\n",
    "svm.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "Y_pred = svm.predict(Xt.reshape(Xt.shape[0], Xt.shape[1]*Xt.shape[2]))\n",
    "print(\"Accuracy Rate:\")\n",
    "print(metrics.accuracy_score(Yt, Y_pred))\n",
    "print(metrics.confusion_matrix(Yt, Y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: 78.97528828887303%\n",
      "Recall: 74.33333333333333%\n",
      "f1_score: 70.53259037764987%\n",
      "\n",
      "Confusion Matrix:\n",
      "[[59  0  0  0  0  0]\n",
      " [ 0 33 13  0  0  0]\n",
      " [12 13 22  0  0  0]\n",
      " [ 0  0  0  7 39  0]\n",
      " [ 0  0  0  0 54  0]\n",
      " [ 0  0  0  0  0 48]]\n",
      "\n",
      "Confusion matrix (normalised to % of total test data):\n",
      "[[ 19.66666794   0.           0.           0.           0.           0.        ]\n",
      " [  0.          11.           4.33333349   0.           0.           0.        ]\n",
      " [  4.           4.33333349   7.33333302   0.           0.           0.        ]\n",
      " [  0.           0.           0.           2.33333325  13.           0.        ]\n",
      " [  0.           0.           0.           0.          18.           0.        ]\n",
      " [  0.           0.           0.           0.           0.          16.        ]]\n",
      "Note: training and testing data is not equally distributed amongst classes, \n",
      "so it is normal that more than a 6th of the data is correctly classifier in the last category.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAz8AAAM+CAYAAAA0Cb/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nKQAAAAhJREFUSU34f1zGfUpnZIlPAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11a2fd240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Precision: {}%\".format(100*metrics.precision_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"Recall: {}%\".format(100*metrics.recall_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"f1_score: {}%\".format(100*metrics.f1_score(Yt, Y_pred, average=\"weighted\")))\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion Matrix:\")\n",
    "confusion_matrix = metrics.confusion_matrix(Yt, Y_pred)\n",
    "print(confusion_matrix)\n",
    "normalised_confusion_matrix = np.array(confusion_matrix, dtype=np.float32)/np.sum(confusion_matrix)*100\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion matrix (normalised to % of total test data):\")\n",
    "print(normalised_confusion_matrix)\n",
    "print(\"Note: training and testing data is not equally distributed amongst classes, \")\n",
    "print(\"so it is normal that more than a 6th of the data is correctly classifier in the last category.\")\n",
    "\n",
    "# Plot Results: \n",
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.imshow(\n",
    "    normalised_confusion_matrix, \n",
    "    interpolation='nearest', \n",
    "    cmap=plt.cm.rainbow\n",
    ")\n",
    "plt.title(\"Confusion matrix \\n(normalised to % of total test data)\")\n",
    "plt.colorbar()\n",
    "tick_marks = np.arange(len(LABELS))\n",
    "plt.xticks(tick_marks, LABELS, rotation=90)\n",
    "plt.yticks(tick_marks, LABELS)\n",
    "plt.tight_layout()\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.883 (+/-0.144) for {'bootstrap': False, 'max_depth': 35, 'max_features': 1, 'min_samples_leaf': 13, 'min_samples_split': 16, 'n_estimators': 300}\n",
      "0.899 (+/-0.182) for {'bootstrap': True, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 13, 'min_samples_split': 14, 'n_estimators': 100}\n",
      "0.884 (+/-0.196) for {'bootstrap': False, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 5, 'min_samples_split': 6, 'n_estimators': 50}\n",
      "0.884 (+/-0.201) for {'bootstrap': True, 'max_depth': 25, 'max_features': 2, 'min_samples_leaf': 15, 'min_samples_split': 9, 'n_estimators': 300}\n",
      "0.885 (+/-0.168) for {'bootstrap': False, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 9, 'min_samples_split': 18, 'n_estimators': 300}\n",
      "0.892 (+/-0.204) for {'bootstrap': True, 'max_depth': 25, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 6, 'n_estimators': 200}\n",
      "0.869 (+/-0.158) for {'bootstrap': True, 'max_depth': 30, 'max_features': 1, 'min_samples_leaf': 14, 'min_samples_split': 15, 'n_estimators': 200}\n",
      "0.880 (+/-0.140) for {'bootstrap': True, 'max_depth': 35, 'max_features': 1, 'min_samples_leaf': 15, 'min_samples_split': 19, 'n_estimators': 300}\n",
      "0.878 (+/-0.200) for {'bootstrap': True, 'max_depth': 20, 'max_features': 2, 'min_samples_leaf': 18, 'min_samples_split': 12, 'n_estimators': 200}\n",
      "0.892 (+/-0.199) for {'bootstrap': True, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 11, 'min_samples_split': 13, 'n_estimators': 100}\n",
      "0.901 (+/-0.189) for {'bootstrap': False, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 19, 'n_estimators': 50}\n",
      "0.874 (+/-0.177) for {'bootstrap': True, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 13, 'min_samples_split': 14, 'n_estimators': 100}\n",
      "0.899 (+/-0.192) for {'bootstrap': False, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 18, 'min_samples_split': 18, 'n_estimators': 100}\n",
      "0.898 (+/-0.206) for {'bootstrap': False, 'max_depth': 20, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 19, 'n_estimators': 300}\n",
      "0.866 (+/-0.175) for {'bootstrap': False, 'max_depth': 20, 'max_features': 1, 'min_samples_leaf': 16, 'min_samples_split': 7, 'n_estimators': 50}\n",
      "Best score\n",
      "0.901\n",
      "Best parameters set found on training set:\n",
      "{'bootstrap': False, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 19, 'n_estimators': 50}\n",
      "\n",
      "[mean: 0.88300, std: 0.07189, params: {'bootstrap': False, 'max_depth': 35, 'max_features': 1, 'min_samples_leaf': 13, 'min_samples_split': 16, 'n_estimators': 300}, mean: 0.89900, std: 0.09110, params: {'bootstrap': True, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 13, 'min_samples_split': 14, 'n_estimators': 100}, mean: 0.88400, std: 0.09807, params: {'bootstrap': False, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 5, 'min_samples_split': 6, 'n_estimators': 50}, mean: 0.88400, std: 0.10061, params: {'bootstrap': True, 'max_depth': 25, 'max_features': 2, 'min_samples_leaf': 15, 'min_samples_split': 9, 'n_estimators': 300}, mean: 0.88500, std: 0.08422, params: {'bootstrap': False, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 9, 'min_samples_split': 18, 'n_estimators': 300}, mean: 0.89200, std: 0.10193, params: {'bootstrap': True, 'max_depth': 25, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 6, 'n_estimators': 200}, mean: 0.86900, std: 0.07898, params: {'bootstrap': True, 'max_depth': 30, 'max_features': 1, 'min_samples_leaf': 14, 'min_samples_split': 15, 'n_estimators': 200}, mean: 0.88000, std: 0.07021, params: {'bootstrap': True, 'max_depth': 35, 'max_features': 1, 'min_samples_leaf': 15, 'min_samples_split': 19, 'n_estimators': 300}, mean: 0.87800, std: 0.09987, params: {'bootstrap': True, 'max_depth': 20, 'max_features': 2, 'min_samples_leaf': 18, 'min_samples_split': 12, 'n_estimators': 200}, mean: 0.89200, std: 0.09972, params: {'bootstrap': True, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 11, 'min_samples_split': 13, 'n_estimators': 100}, mean: 0.90100, std: 0.09448, params: {'bootstrap': False, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 19, 'n_estimators': 50}, mean: 0.87400, std: 0.08862, params: {'bootstrap': True, 'max_depth': 25, 'max_features': 1, 'min_samples_leaf': 13, 'min_samples_split': 14, 'n_estimators': 100}, mean: 0.89900, std: 0.09591, params: {'bootstrap': False, 'max_depth': 35, 'max_features': 2, 'min_samples_leaf': 18, 'min_samples_split': 18, 'n_estimators': 100}, mean: 0.89800, std: 0.10312, params: {'bootstrap': False, 'max_depth': 20, 'max_features': 2, 'min_samples_leaf': 12, 'min_samples_split': 19, 'n_estimators': 300}, mean: 0.86600, std: 0.08760, params: {'bootstrap': False, 'max_depth': 20, 'max_features': 1, 'min_samples_leaf': 16, 'min_samples_split': 7, 'n_estimators': 50}]\n",
      "\n",
      "Accuracy Rate:\n",
      "0.836666666667\n",
      "[[59  0  0  0  0  0]\n",
      " [ 3 43  0  0  0  0]\n",
      " [15  4 28  0  0  0]\n",
      " [ 0  0  0 24 20  2]\n",
      " [ 0  0  0  5 49  0]\n",
      " [ 0  0  0  0  0 48]]\n"
     ]
    }
   ],
   "source": [
    "#Random Forest\n",
    "\n",
    "\n",
    "\n",
    "param_dist = {\n",
    "    \"max_depth\": [20, 25, 30, 35],\n",
    "    \"max_features\": sp_randint(1, min(300, X.shape[1])),\n",
    "    \"min_samples_split\": sp_randint(2, 20),\n",
    "    \"min_samples_leaf\": sp_randint(2, 20),\n",
    "    'n_estimators': [50, 100, 200, 250, 300],\n",
    "    'bootstrap': [True, False],\n",
    "}\n",
    "\n",
    "r_forest = RandomForestClassifier(random_state=1)\n",
    "n_iter_search = 15\n",
    "clf = RandomizedSearchCV(\n",
    "    estimator=r_forest,\n",
    "    param_distributions=param_dist,\n",
    "    n_iter=n_iter_search,\n",
    "    cv=5,\n",
    "    random_state=1,\n",
    ")\n",
    "\n",
    "clf.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "\n",
    "for params, mean_score, scores in clf.grid_scores_:\n",
    "    print(\"%0.3f (+/-%0.03f) for %r\" % (mean_score, scores.std() * 2, params))\n",
    "    \n",
    "print(\"Best score\")\n",
    "print(clf.best_score_)\n",
    "print(\"Best parameters set found on training set:\")\n",
    "print(clf.best_params_)\n",
    "print()\n",
    "print(clf.grid_scores_)\n",
    "print()\n",
    "\n",
    "r_forest = RandomForestClassifier(bootstrap=clf.best_params_['bootstrap'], max_depth=clf.best_params_['max_depth'], max_features=clf.best_params_['max_features'], \n",
    "                            min_samples_leaf= clf.best_params_['min_samples_leaf'], min_samples_split= clf.best_params_['min_samples_split'],\n",
    "                           n_estimators = clf.best_params_['n_estimators'])\n",
    "    \n",
    "#r_forest = RandomForestClassifier(bootstrap=False, max_depth=25, max_features=23, min_samples_leaf= 3, min_samples_split= 2, n_estimators = 250)\n",
    "r_forest.fit(X.reshape(X.shape[0], X.shape[1]*X.shape[2]), Y)\n",
    "Y_pred = r_forest.predict(Xt.reshape(Xt.shape[0], Xt.shape[1]*Xt.shape[2]))\n",
    "\n",
    "print(\"Accuracy Rate:\")\n",
    "print(metrics.accuracy_score(Yt, Y_pred))\n",
    "print(metrics.confusion_matrix(Yt, Y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision: 85.59656339832294%\n",
      "Recall: 83.66666666666667%\n",
      "f1_score: 82.76898087323048%\n",
      "\n",
      "Confusion Matrix:\n",
      "[[59  0  0  0  0  0]\n",
      " [ 3 43  0  0  0  0]\n",
      " [15  4 28  0  0  0]\n",
      " [ 0  0  0 24 20  2]\n",
      " [ 0  0  0  5 49  0]\n",
      " [ 0  0  0  0  0 48]]\n",
      "\n",
      "Confusion matrix (normalised to % of total test data):\n",
      "[[ 19.66666794   0.           0.           0.           0.           0.        ]\n",
      " [  1.          14.33333302   0.           0.           0.           0.        ]\n",
      " [  5.           1.33333337   9.33333302   0.           0.           0.        ]\n",
      " [  0.           0.           0.           8.           6.66666698\n",
      "    0.66666669]\n",
      " [  0.           0.           0.           1.66666675  16.33333206   0.        ]\n",
      " [  0.           0.           0.           0.           0.          16.        ]]\n",
      "Note: training and testing data is not equally distributed amongst classes, \n",
      "so it is normal that more than a 6th of the data is correctly classifier in the last category.\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAz8AAAM+CAYAAAA0Cb/mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3Xm4JVV19/HvrxsnBEEBCYNMBhEQQeiAYkwwYgIEh5gY\nQTRiNJjBJAQNDhCDRiVBCQZFfRERnAAn8iqiiPElKIPSILMoIBBmaEAGB6T7rvePqhsOlztUN/f2\nPYf6fp7nPH2qalftXbtbueuuVbtSVUiSJEnSo92C+R6AJEmSJK0MBj+SJEmSesHgR5IkSVIvGPxI\nkiRJ6gWDH0mSJEm9YPAjSZIkqRcMfiRJkiT1gsGPJEmSpF4w+JEkSZLUCwY/kiRJknphlfkegCRJ\nkqQVt1tSS+Z7EFM4H06rqt3mexzjDH4kSZKkEbYEWDzfg5hCYO35HsMgy94kSZIk9YKZH0mSJGnU\nLRzSnMaysfkewUMM6SxJkiRJ0uwy+JEkSZLUC5a9SZIkSaMswMLM9ygmt2y+B/BQZn4kSZIk9YLB\njyRJkqResOxNkiRJGmkZ3tXehqzubVhnSZIkSZJmlcGPJEmSpF6w7E2SJEkaZQFWGdLV3oaMmR9J\nkiRJvWDwI0mSJKkXLHuTJEmSRlkY4tXehouzJEmSJKkXDH4kSZIk9YJlb5IkSdKoW+hqb12Y+ZEk\nSZLUCwY/kiRJknrBsjdJkiRplCWu9taRsyRJkiSpFwx+JEmSJPWCZW+SJEnSKPMlp505S5IkSZJ6\nweBHkiRJUi9Y9iZJkiSNOl9y2omZH0mSJEm9YPAjSZIkqRcse5MkSZJGmS857cxZkiRJktQLBj+S\nJEmSesGyN0mSJGnUudpbJ2Z+JEmSJPWCwY8kSZKkXrDsTZIkSRplAVYxp9GFsyRJkiSpFwx+JEmS\nJPWCZW+SJEnSKEtc7a0jMz+SJEmSesHgR5IkSVIvWPYmSZIkjbqF5jS6cJYkSZIk9YLBjyRJkqRe\nsOxNkiRJGmXBsreOnCVJkiRJvWDwI0mSJKkXLHuTJEmSRpkvOe3MzI8kSZKkXjD4kSRJktQLlr1J\nkiRJo87V3jpxliRJkiT1gsGPJEmSpF6w7E2SJEkaZcHV3joy8yNJkiSpFwx+JEmSJPWCZW+SJEnS\nSIurvXXkLEmSJEnqBYMfSZIkSb1g2ZskSZI0ylztrTMzP5IkSZJ6weBHkiRJUi9Y9iZJkiSNsuBq\nbx05S5IkSZJ6weBHkiRJUi9Y9iZJkiSNOld768TMjyRJkqReMPiRJM2aJE9I8rUkdyf54iO4zj5J\nvjWbY5svSV6Q5MfzPQ5JkmVvktRLSV4NHAA8E7gXuBB4X1V97xFe+k+AdYG1qmrpil6kqj4HfO4R\njmXOJSlg86q6aqo2VfVdYIuVNypJvZO42ltHzpIk9UySA4APAe+nCVQ2Ao4CXjoLl98Y+MkjCXwe\nTZL4S0ZJGiIGP5LUI0nWAN4D/E1VfaWqfl5VD1TVKVV1YNvmcUk+lOSm9vOhJI9rj+2S5IYkb0ly\nW5Kbk7y+PfZu4F3Aq5Lcl+QNSQ5J8tmB/jdJUuNBQZJ9k/w0yb1Jrkmyz8D+7w2ct3OS89pyuvOS\n7Dxw7Iwk/5LkrPY630qy9hT3Pz7+AwfG//IkeyT5SZI7k7xzoP2OSc5J8rO27UeSPLY9dmbb7KL2\nfl81cP23JbkF+NT4vvacp7d9bN9ur5/k9iS7PKK/WElSJwY/ktQvzwMeD5w8TZuDgOcC2wHbAjsC\nBw8c/w1gDWAD4A3AUUmeXFX/TJNNOqmqVquqT043kCRPBI4Edq+q1YGdacrvJrZ7CvD1tu1awL8D\nX0+y1kCzVwOvB54KPBZ46zRd/wbNHGxAE6x9AngNsAPwAuCfkmzatl0G/AOwNs3cvQj4a4Cq+p22\nzbbt/Z40cP2n0GTB9hvsuKquBt4GfDbJqsCngOOr6oxpxitJM1u4YDg/Q2b4RiRJmktrAUtmKEvb\nB3hPVd1WVbcD7wZeO3D8gfb4A1V1KnAfK/5MyxjwrCRPqKqbq+qySdr8IXBlVX2mqpZW1QnAFcBL\nBtp8qqp+UlW/BL5AE7hN5QGa55seAE6kCWz+o6rubfu/nCboo6rOr6pz236vBf4P8Lsd7umfq+r+\ndjwPUVWfAK4Cvg+sRxNsSpJWAoMfSeqXO4C1Z3gWZX3guoHt69p9/3uNCcHTL4DVlncgVfVz4FXA\nXwI3J/l6kmd2GM/4mDYY2L5lOcZzR1Uta7+PBye3Dhz/5fj5SZ6R5JQktyS5hyazNWlJ3YDbq+pX\nM7T5BPAs4MNVdf8MbSVJs8TgR5L65RzgfuDl07S5iaZka9xG7b4V8XNg1YHt3xg8WFWnVdWLaTIg\nV9AEBTONZ3xMN67gmJbHx2jGtXlVPQl4JzDTmwRruoNJVqNZcOKTwCFtWZ8krbjQvOR0GD9DxuBH\nknqkqu6mec7lqPZB/1WTPCbJ7kkOa5udABycZJ124YB3AZ+d6pozuBD4nSQbtYstvGP8QJJ1k7ys\nffbnfpryubFJrnEq8Iwkr06ySpJXAVsBp6zgmJbH6sA9wH1tVuqvJhy/FdhsOa/5H8DiqnojzbNM\nH3/Eo5QkdWLwI0k9U1WH07zj52DgduB64M3Af7ZN3gssBi4GLgEuaPetSF+nAye11zqfhwYsC9px\n3ATcSfMszcTggqq6A9gTeAtN2d6BwJ5VtWRFxrSc3kqzmMK9NFmpkyYcPwQ4vl0N7k9nuliSlwG7\n8eB9HgBsP77KnSRpbqVq2uy8JEmSpCG26MlPqMW7bDpzw3mQ//zR+VW1aL7HMc7MjyRJkqReMPiR\nJEmS1AvTLXUqSZIkaRQM4cpqXSQ5lua5ztuq6lntvpN48P1xawI/q6qHvb8tybU0z2QuA5Z2Ka8z\n+JEkSZI0X44DPgJ8enxHVb1q/HuSw4G7pzn/hcuzAI5lb5I0gySHJtl/vscxkySbJKnxF5gm+UaS\n181yH/sm+d5sXnNlaJfVPjPJve1/SFd2/0M7b0nOSPLGObr24UketoKfJI2rqjNpVvx8mCQB/pTm\nFQyzwuBHkqaRZB3gz4D/M99jWV5VtXtVHb+y+psYfK3A+U9Lcm6SOycGKG0g90hWC9oPWAI8qare\nMknfxyXpvJz38raf4VqPaN7malxTXP/aJLsuxykfBN6Z5LFzNSZJtC85XTCcH1g7yeKBz37LcWcv\nAG6tqiunOF7At5Oc3/W6lr1J0vT2BU6tql/O9oWTrFJVS2f7uiPsHcDxwOeBC5KcUFWL25eaXlNV\nix/BtTcGLi/f77BSVdXNSa4AXgp8ab7HI2leLHkES13vzfRZn9+uqhuTPBU4PckVbSZpSmZ+JGl6\nuwP/Pb6RZJckNyR5S5Lbktyc5PUDx9dI8ukktye5LsnBSRa0x/ZNclaSI5LcARwyYd/Pkvw0yc7t\n/uvbPl43cP0/TPLDJPe0xw+ZauCD5UxJfjPJfye5O8mS9mHS8XbPTHJ6m3H58eDLOpOsleSrbX8/\nAJ4+zVyN/wfnZ0nuS/K8JAvaObiuvZdPJ1ljivM3Bb5TVXcD5wGbJXkS8HbgndP0Oz7WnZOc197j\neUl2bvcfB7wOOLAd164TztsP2Gfg+Nfa/Vu2c/izJJcleekM7d+e5Oq2tO7yJH8005inmrf2en+e\n5EdJ7kpyWpKN2/1p/73c1v69XJLkWVONa5J5enGSK9p5+gjN74zHjz09yXeS3NH+O/lckjXbY58B\nNgK+1l7/wHb/F5Pc0l7vzCRbT+jyDOAPO86FJAHNLwiBV/Dwl0v/r6q6sf3zNuBkYMeZrmvwI0nT\n2wb48YR9vwGsAWwAvAE4KsmT22Mfbo9tBvwuTcnc6wfO3Qn4KbAu8L6BfRcDa9FkPU4Efgv4TeA1\nwEeSrNa2/Xl7zTVpfqD8qyQv73Af/wJ8C3gysGE7TpI8ETi97fepwF7AR5Ns1Z53FPArYD3gz9vP\nVH6n/XPNqlqtqs6hyZztC7ywnZPVaB5sncylwIvbH7Z3AC5rx/2hqvrZdDeX5CnA14Ejaebx34Gv\nJ1mrqvYFPgcc1o7r24PnVtXRE46/JMljgK/RzNlTgb8FPpdki8nat5e6mqZEYw3g3cBnk6w33bhb\nD5u3JC+jCfheAawDfJcHf/v5++05z2j7+lPgjmnGNThPawNfAQ4G1m7H/PzBJsChwPrAlsDTgEPa\neXot8D/AS9rrH9ae8w1g83aeLmjHMOhHwLYd5kHSikqa1d6G8bPidgWuqKobJr/lPDHJ6uPfaf6/\n8dKZLmrwI0nTW5NmGc1BDwDvqaoHqupU4D5giyQLaYKHd1TVvVV1LXA48NqBc2+qqg9X1dKBUrpr\nqupTVbWM5jdcT2uvf39VfQv4NU0gRFWdUVWXVNVYVV1M8wPx73a4jwdoSr/Wr6pfVdX4w/d7Ate2\n/S+tqh8CXwZe2d7PHwPvqqqfV9WlNGVpy2Mf4N+r6qdVdR9Nadtemfz5lkNpgof/Bj4KPBZ4Nk2m\n4fNtVuHNU/Tzh8CVVfWZ9j5OAK4AHhYAdPRcmkDtX6vq11X1HeAUmhKMSVXVF6vqpvbv5iTgSjr8\nFnIKfwkcWlU/aksj3w9s12Z/HgBWB54JpG1zc8fr7gFcVlVfqqoHgA8Btwzcw1VVdXr7b+92miBy\n2n9fVXVs++/9fppAadsJ2b17af53JEkPk+QE4Bya/47ekOQN7aG9mFDylmT9JKe2m+sC30tyEfAD\n4OtV9c2Z+vOZH0ma3l00P2gOumPCszq/oPlBeW3gMcB1A8euo8kQjbt+kj5uHfj+S4CqmrhvNYAk\nOwH/CjyLJjh4HPDFDvdxIE0W5QdJ7gIOr6pjaQKinZIMZlZWAT5Dk3FYZcKYB++ti/V5+HysQvMf\nrRsHG1bVncCrANKUCp5JEwS8nea3efvSPAv0X1X1oxn6Ge9rA1bM+sD1VTXW9XpJ/gw4ANik3TX+\nb2JFbAz8Rx668EOADarqO2252lHAxkm+Ary1qu7pcN31Gfj7rKpK8r/bSdYF/oMmCF2d5pekd011\nsTZAfh/wSpp/L+PztTYPLk27OjBt5k5Sf1XVpL9UarP2E/fdRPNLHKrqp6xAVtnMjyRN72Ka8qIu\nlvBghmXcRjz0h/xH+sD954GvAk+rqjWAjzPwzMZUquqWqvqLqlofeBNNadtv0vwg/N9VtebAZ7Wq\n+ivgdmApTSZq8H6m7GaSfTfx8PlYykMDvsnsB5zbZpu2ARZX1a+BS9rtmfoZ7+vGSdpOZuLYbwKe\n1gZhk13vIe3bjMwngDcDa1XVmjQBW5eaj8nm7XrgTRP+Xp5QVWcDVNWRVbUDsBXNv89/nOZag25m\n4O8zSXjo3+/722tsU1VPoim7HLyHidd/NfAymvKUNXgw8Bs8Z0vgohnGJemRmu9V3aZe7W2oDN+I\nJGm4nEq3sjLasrUvAO9Lsnr7A/EBwGdncTyrA3dW1a+S7Ejzw+eMkrwyyYbt5l00P8SO0ZRyPSPJ\na5M8pv38VpIt2/v5Cs3CDKu2zwFN996g29trbjaw7wTgH5Js2j639H7gpOlWuUuzas/f0D5rAlwD\nvLA9fxHNM1MTndrex6uTrJJmhbit2vvr4tYJ4/4+TUbvwHZOdqEpoTtxivZPpJnT29t7eD1Ndq6L\nyebt48A7xhcPSLOQxivb77+VZKf2uaSf0zyTNZ5xmTiuib4ObJ3kFW3p4d/RPMM2bnWaMs67k2zA\ng0HVuInXXx24H7gDWJXm73ei36V5LkiS5p3BjyRN79PAHkme0LH939L8QPpT4Hs0mZpjZ3E8fw28\nJ8m9wLtogq0ufgv4fpL7aDJHf98+h3MvzUOie9FkO24B/o2mnA6aTMZq7f7jgE9N1UFV/YKmBOqs\nNCukPZfm3j9DU8J2Dc0P6n87w1g/SPPM033t9qHA79FkQ7422ZLXVXUHzfNLb6H5QfxAYM/q/tbv\nTwJbteP+zzbL9BKa1f6W0DyD9GdVdcUU7S+neb7rHJoAYRvgrC4dTzZvVXUyzd/DiUnuocki7d6e\n8iSaLNNdNKV4dwAfmGxck/S1hKZE7V/b8zafMM53A9vTlKx9nSb4HXQocHB7/bfS/O/jOpqM2OXA\nuYON2wUftgIeNhZJmg8pX3kgSdNK8n7gtqr60HyPRRol7TNLV1fVR+d7LNKj2aK1V63Fe24x38OY\nVI6/8Pxa8ff8zDoXPJCkGVTVjO+YkfRwVfWW+R6DJA2y7E2SJElSL5j5kSRJkkZahnJltWHkLEmS\nJEnqBYMfSZIkSb1g2Ztm1dpJbTLfg+ipm9hhvocgSVIv/Ixr+UUt6fIS45UjwMLhGc4wM/jRrNoE\neNgLOLRSHOLMS5K0UhzN0KzcrOVk2ZskSZKkXjDzI0mSJI2y4GpvHTlLkiRJknrB4EeSJElSL1j2\nJkmSJI00X3LalbMkSZIkqRcMfiRJkiT1gmVvkiRJ0igLsMCXnHZh5keSJElSLxj8SJIkSeoFy94k\nSZKkUedqb504S5IkSZJ6weBHkiRJUi9Y9iZJkiSNsgALXe2tCzM/kiRJknrB4EeSJElSL1j2JkmS\nJI20uNpbR86SJEmSpF4w+JEkSZLUC5a9SZIkSaPM1d46M/MjSZIkqRcMfiRJkiT1gmVvkiRJ0qhb\nYE6jC2dJkiRJUi8Y/EiSJEnqBcveJEmSpFGWuNpbR2Z+JEmSJPWCwY8kSZKkXrDsTZIkSRplARaa\n0+jCWZIkSZLUCwY/kiRJknrBsjdJkiRp1LnaWydmfiRJkiT1gsGPJEmSpF6w7E2SJEkaZQksMKfR\nhbMkSZIkqRcMfiRJkiT1gmVvkiRJ0qhztbdOzPxIkiRJ6gWDH0mSJEm9YNmbJEmSNMoCLDSn0YWz\nNI+SHJFk/4Ht05IcM7B9eJID2u/7J/lVkjUGju+S5JRJrntGkkXt902TXJnkDwbbJ9k3yViSZw+c\nd2mSTdrvqyX5WJKrk1yQ5PwkfzH7syBJkiStHAY/8+ssYGeAJAuAtYGtB47vDJzdft8bOA94RdeL\nJ9kQ+Cbwlqo6bZImNwAHTXH6McBdwOZVtT2wG/CUrn1LkiRJw8bgZ36dDTyv/b41cClwb5InJ3kc\nsCVwQZKnA6sBB9MEQV2sB3wLOKiqvjpFm1OArZNsMbiz7W9H4OCqGgOoqtur6t+635okSZJWmgUZ\nzs+QMfiZR1V1E7A0yUY0WZ5zgO/TBESLgEuq6tfAXsCJwHeBLZKs2+HyxwMfqaovTdNmDDgMeOeE\n/VsDF40HPpIkSdKjgcHP/DubJvAZD37OGdg+q22zN3BiG4x8GXhlh+t+G3hNklVnaPd54LlJNp2q\nQZKDklyY5KYpju+XZHGSxbd3GJgkSZI0Hwx+5t/4cz/b0JS9nUuT+dkZODvJNsDmwOlJrqXJAnUp\nfTuM5hmhLyaZclW/qloKHA68bWD35cC27XNIVNX7qmo74ElTXOPoqlpUVYvW6TAwSZIkzaKkWe1t\nGD9DZvhG1D9nA3sCd1bVsqq6E1iTJgA6mybQOaSqNmk/6wPrJ9m4w7X3B+4BPplkuqLL44BdgXUA\nquoqYDHw3iQLAZI8nmYhRUmSJGkkGfzMv0toVnk7d8K+u6tqCU2m5+QJ55zc7gd4UZIbBj7jCyhQ\nVQW8jmbxg8OmGkD7XNGRwFMHdr8RWAu4Ksli4HTgwBW4P0mSJGko+JLTeVZVy5hQTlZV+w5832yS\ncw4Y2HzCJJfdZaDtr4HfHzh2Rrv/OJqMz3i7I2kCoPHte4A3dbgFSZIkzbchXFltGJn5kSRJktQL\nBj+SJEmSesHgR5IkSVIv+MyPJEmSNMrCUC4rPYycJUmSJEm9YPAjSZIkqRcse5MkSZJGWlzquiMz\nP5IkSZJ6weBHkiRJUi9Y9iZJkiSNMld768xZkiRJktQLBj+SJEmSesGyN0mSJGnUudpbJ2Z+JEmS\nJPWCwY8kSZKkXrDsTZIkSRpliau9deQsSZIkSeoFgx9JkiRJvWDZmyRJkjTqXO2tEzM/kiRJknrB\n4EeSJElSL1j2JkmSJI2y4GpvHTlLkiRJknrB4EeSJElSL1j2JkmSJI20uNpbR2Z+JEmSJPWCwY8k\nSZKkXjD4kSRJkkZZgAULhvMz09CTY5PcluTSgX2HJLkxyYXtZ48pzt0tyY+TXJXk7V2myuBHkiRJ\n0nw5Dthtkv1HVNV27efUiQeTLASOAnYHtgL2TrLVTJ0Z/EiSJEmaF1V1JnDnCpy6I3BVVf20qn4N\nnAi8bKaTXO1NkiRJGnULH3Wrvf1tkj8DFgNvqaq7JhzfALh+YPsGYKeZLmrmR5IkSdJcWTvJ4oHP\nfh3O+RiwGbAdcDNw+GwNxsyPJEmSpLmypKoWLc8JVXXr+PcknwBOmaTZjcDTBrY3bPdNy+BHkiRJ\nGmVJp5XVRkWS9arq5nbzj4BLJ2l2HrB5kk1pgp69gFfPdG2DH0mSJEnzIskJwC405XE3AP8M7JJk\nO6CAa4E3tW3XB46pqj2qammSNwOnAQuBY6vqspn6M/iRJEmSNC+qau9Jdn9yirY3AXsMbJ8KPGwZ\n7OkY/EiSJEmjbsGjbrW3OfHoKQ6UJEmSpGkY/EiSJEnqBcveJEmSpFEWHo0vOZ0TZn4kSZIk9YLB\njyRJkqResOxNs+q6x+/AfpucN9/D6KUjPvbX8z2EXvvACz8630OQJPXZo+glp3PJWZIkSZLUCwY/\nkiRJknrBsjdJkiRplCWM+ZLTTsz8SJIkSeoFgx9JkiRJvWDZmyRJkjTCChhztbdOnCVJkiRJvWDw\nI0mSJKkXLHuTJEmSRpyrvXVj5keSJElSLxj8SJIkSeoFy94kSZKkEVYJyxaa0+jCWZIkSZLUCwY/\nkiRJknrBsjdJkiRpxLnaWzdmfiRJkiT1gsGPJEmSpF6w7E2SJEkaZYFaYE6jC2dJkiRJUi8Y/EiS\nJEnqBcveJEmSpBFWuNpbV2Z+JEmSJPWCwY8kSZKkXrDsTZIkSRpliWVvHZn5kSRJktQLBj+SJEmS\nesGyN0mSJGmENau9mdPowlmSJEmS1AsGP5IkSZJ6wbI3SZIkacS52ls3Zn4kSZIk9YLBjyRJkqRe\nsOxNkiRJGmGVsCzmNLpwliRJkiT1gsGPJEmSpF6w7E2SJEkaca721o2ZH0mSJEm9YPAjSZIkqRcs\ne5MkSZJGnGVv3Zj5kSRJktQLBj+SJEmSesGyN0mSJGmEVaAWmNPowlmSJEmS1AsGP5IkSZJ6wbI3\nSZIkaaTF1d46GonMT5Ijkuw/sH1akmMGtg9PckD7ff8kv0qyxsDxXZKcMsl1z0iyqP2+aZIrk/zB\nYPsk+yYZS/LsgfMuTbJJ+321JB9LcnWSC5Kcn+QvprmXh40lyXFJ/mRgTD9OclGSs5Js0e7fM8kP\n2/2XJ3lTkoOSXNh+lg18/7uBa1+Y5MSO/Z2XZLuBdn+e5JIkF7f3/LKp7kuSJEkadiMR/ABnATsD\nJFkArA1sPXB8Z+Ds9vvewHnAK7pePMmGwDeBt1TVaZM0uQE4aIrTjwHuAjavqu2B3YCndO17CvtU\n1bbA8cAHkjwGOBp4Sbv/OcAZVfW+qtquqrYDfjn+vaqObO9rS2Ah8IIkT+zQ30eBD7Tnbtje829X\n1bOB5wIXP8L7kiRJkubNqAQ/ZwPPa79vDVwK3JvkyUkeB2wJXJDk6cBqwME0QVAX6wHfAg6qqq9O\n0eYUYOvxLMy4tr8dgYOragygqm6vqn/rfmvTOhP4TWB1mhLFO9o+7q+qH3c4f2/gMzT31yVrcw6w\nQfv9qcC9wH1tn/dV1TXLNXpJkiTNvcDYggVD+Rk2wzeiSVTVTcDSJBvRZHnOAb5PExAtAi6pql8D\newEnAt8FtkiybofLHw98pKq+NE2bMeAw4J0T9m8NXDQe+MyBl9Dc253AV4HrkpyQZJ82AzaTV9HM\nxwl0CwZ3A/6z/X4RcCtwTZJPJXnJVCcl2S/J4iSLf7X09g7dSJIkSSvfSAQ/rbNpAp/x4Oecge2z\n2jZ7Aye2wciXgVd2uO63gdckWXWGdp8Hnptk06kaDDyDc9M016kO+z+X5ELg+cBbAarqjcCLgB+0\n+46dbrDts0xLqup/gP8CnpNkqnK8zyW5hqbM7ai2v2U0wdCfAD8BjkhyyKQDrzq6qhZV1aLHr7LO\ndMOSJEmS5s0oBT/jz/1sQ1P2di5N5mdn4Owk2wCbA6cnuZYmC9Ql23EYzTNCX0wy5ep3VbUUOBx4\n28Duy4Ftx7Mw48/gAE+apr87gCdP2PcUYMnA9j7tszsvr6rrB8ZwSVUdAbwY+OMZ7mtv4JntXFzd\njmmqc/YBNqPJgn14oL+qqh9U1aE08zlTn5IkSVrJChhLhvIzbEYp+Dkb2BO4s6qWtaVga9IEQGfT\n/LB/SFVt0n7WB9ZPsnGHa+8P3AN8Mpn2b+k4YFdgHYCqugpYDLw3yUKAJI8HprvGle24tmzbbwxs\nC1w41QntinK7DOzaDrhumvYLgD8FthmfD5pnfqYMBquqgH+iyW49M8n6Sbbv2qckSZI07EYp+LmE\nZpW3cyfsu7uqltBkJk6ecM7J7X6AFyW5YeAzvoDC+A/+r6NZ/OCwqQbQPld0JM1iAOPeCKwFXJVk\nMXA6cOA017gfeA3wqba07UvAG6vq7invvAmmDmyXpL4QeDew7zTtXwDc2D4rNe5MYKsk600ztl/S\nZLf+EXgM8MEkV7R9vgr4+2n6lCRJkoZamp/7pdmxzhMW1R9tct58D6OXjvjY38z3EHrtAy/86HwP\nQZK0khzNIm6qxUNT07XNs9ark7/8hvkexqQ2f+b7zq+qRfM9jnGjlPmRJEmSpBU25QP+emTaBRg+\nM2H3/VW103yMR5IkSeo7g585UlWX0CwSIEmSJM2ZSobyhaLDyFmSJEmS1AsGP5IkSZJ6wbI3SZIk\nacQtG8IXig4jMz+SJEmSesHgR5IkSVIvWPYmSZIkjbACV3vryFmSJEmS1AsGP5IkSZJ6wbI3SZIk\naaSFcrW3Tsz8SJIkSeoFgx9JkiRJvWDZmyRJkjTKAmMLLHvrwsyPJEmSpF4w+JEkSZLUC5a9SZIk\nSSOsgLFqIYcoAAAgAElEQVSY0+jCWZIkSZLUCwY/kiRJknrBsjdJkiRpxLnaWzdmfiRJkiT1gsGP\nJEmSpF6w7E2SJEkaZQljseytCzM/kiRJknrB4EeSJElSL1j2JkmSJI2wApYtMKfRhbMkSZIkqRcM\nfiRJkiTNiyTHJrktyaUD+z6Q5IokFyc5OcmaU5x7bZJLklyYZHGX/gx+JEmSpBE31q74NmyfDo4D\ndpuw73TgWVX1bOAnwDumOf+FVbVdVS3q0pnBjyRJkqR5UVVnAndO2Petqlrabp4LbDhb/Rn8SJIk\nSRpWfw58Y4pjBXw7yflJ9utyMVd7kyRJkkZYwTC/5HTtCc/jHF1VR3c5MclBwFLgc1M0+e2qujHJ\nU4HTk1zRZpKmZPAjSZIkaa4s6fo8zqAk+wJ7Ai+qqpqsTVXd2P55W5KTgR2BaYMfy94kSZIkDY0k\nuwEHAi+tql9M0eaJSVYf/w78PnDpZG0HmfmRJEmSRllCjehLTpOcAOxCUx53A/DPNKu7PY6mlA3g\n3Kr6yyTrA8dU1R7AusDJ7fFVgM9X1Tdn6s/gR5IkSdK8qKq9J9n9ySna3gTs0X7/KbDt8vY3miGi\nJEmSJC0nMz+SJEnSiBvi1d6GipkfSZIkSb1g8CNJkiSpFyx706xa8rSlfPKIJfM9jF761YEfme8h\n9No6Y9+e7yH01uoLdp3vIUjSvBryl5wOFTM/kiRJknrB4EeSJElSL1j2JkmSJI04y966MfMjSZIk\nqRcMfiRJkiT1gmVvkiRJ0girhLGY0+jCWZIkSZLUCwY/kiRJknrBsjdJkiRpxLnaWzdmfiRJkiT1\ngsGPJEmSpF6w7E2SJEkaYQUsW2DZWxdmfiRJkiT1gsGPJEmSpF6w7E2SJEkaZb7ktDNnSZIkSVIv\nGPxIkiRJ6gXL3iRJkqQRV77ktBMzP5IkSZJ6weBHkiRJUi9Y9iZJkiSNsALGsOytCzM/kiRJknrB\n4EeSJElSL1j2JkmSJI24MVd768TMjyRJkqReMPiRJEmS1AuWvUmSJEkjLYzFnEYXzpIkSZKkXjD4\nkSRJktQLlr1JkiRJI6xwtbeuzPxIkiRJ6gWDH0mSJEm9YNmbJEmSNMoCyyx768TMjyRJkqReMPiR\nJEmS1AuWvUmSJEkjzNXeujPzI0mSJKkXDH4kSZIk9YJlb5IkSdJIC2PmNDpxliRJkiT1gsGPJEmS\npF6w7E2SJEkaceVqb52Y+ZEkSZLUCwY/kiRJknphzoKfJEck2X9g+7QkxwxsH57kgPb7/kl+lWSN\ngeO7JDllkuuekWRR+33TJFcm+YPB9kn2TTKW5NkD512aZJP2+2pJPpbk6iQXJDk/yV9Mcy+bJPll\nkh8m+VGSHyTZd0Kblye5uD1+SZKXt/u3TXLhQLu922s9pt3eJsnFA/e2eKDtoiRntN9XTfK59tqX\nJvleko2TXNh+bkly48D2YwfGVUmeOeF+Lh2Y57vbc65I8sGBdusmOSXJRUkuT3LqVHMkSZKk+TH+\nktNh/Aybucz8nAXsDJBkAbA2sPXA8Z2Bs9vvewPnAa/oevEkGwLfBN5SVadN0uQG4KApTj8GuAvY\nvKq2B3YDnjJDl1dX1XOqaktgL2D/JK9vx7It8EHgZe3xlwIfbIOvS4CNkqzeXmdn4EfAcwa2zx7o\n56lJdp+k/78Hbq2qbarqWcAbgFuqaruq2g74OHDE+HZV/bo9b2/ge+2fU/lue43nAHsmeX67/z3A\n6VW1bVVtBbx9hjmSJEmShtZcBj9nA89rv28NXArcm+TJSR4HbAlckOTpwGrAwUz/A/qg9YBvAQdV\n1VenaHMKsHWSLQZ3tv3tCBxcVWMAVXV7Vf1b1xurqp8CBwB/1+56K/D+qrqmPX4NcCjwj20fi4Gd\n2rY7AEfRBobtn2cNXP4DTB60rQfcODCGH1fV/dONM8lqwG/TBEp7dbivXwIXAhsM9HnDwPGLZ7qG\nJEmSNKzmLPipqpuApUk2ovkB/xzg+zQB0SLgkjY7sRdwIvBdYIsk63a4/PHAR6rqS9O0GQMOA945\nYf/WwEXjgc8jcAEwXkq2NXD+hOOLeTDTdRawc5IntuM6g4cGP4OZn3OAXyd54YTrHQu8Lck5Sd6b\nZPMOY3wZ8M2q+glwR5Idpmuc5MnA5sCZ7a6jgE8m+X9JDkqy/hTn7ZdkcZLF3H1Hh2FJkiRpNo2R\nofwMm7le8OBsmh/ux4Ofcwa2x7MdewMntsHIl4FXdrjut4HXJFl1hnafB56bZNOpGrQ/1F+Y5KYO\n/T7k1OVoOz4POwLnVdXVwG8mWQdYrd0e9F6aTNj/qqoLgc1oMkNPAc5LsuUM/e5NE1jS/jlVZu0F\nSS6iySydVlW3tH2e1vb5CZpA74ftmB+iqo6uqkVVtYg11pphSJIkSdL8mOvgZ/y5n21oyt7Opcn8\n7AycnWQbmkzD6UmupckCdSl9O4zmGaEvJpnyXUVVtRQ4HHjbwO7LgW3b55Coqve1z7s8aflujefQ\nPLszfs2JWZUdgMva7+cCvwU8nyYAhKacbK+B7cFxfwd4AvDcCfvvq6qvVNVfA58F9phqcEmeAvwe\ncEw7t/8I/Gky6ZNn362qbWkyVW9Ist1An3dW1eer6rU0c/47U/UpSZIkDbOVkfnZE7izqpZV1Z3A\nmjQB0Nk0gc4hVbVJ+1kfWD/Jxh2uvT9wD01Z1nRZmOOAXYF1AKrqKpqStPcmWQiQ5PEsRyanXTXu\ng8CH210fBN4xsJrcJjTldoe3fd4LXA+8ngeDnXPaexh83mfQe4EDB/p8fluWRruS21bAddMM80+A\nz1TVxu3cPg24BnjBVCe0zyr9K22wmOT3xrNr7YINTwf+Z5o+JUmStJIVYSwLhvIzbOZ6RJfQrPJ2\n7oR9d1fVEprMx8kTzjmZBx/Of1GSGwY+4wsoUFUFvI7mofzDphpA+1zRkcBTB3a/EVgLuKpdWvp0\nBgKNKTx9fKlr4AvAkVX1qbaPC2kChq8luQL4GnBgu3/cWcDjqur6dvscmpKywed9Bsd9KnD7YP/A\nfye5BPghTQD35WnGuzcPn9svM3Nm7ePA77QB3A7A4nYp7nOAY6rqvBnOlyRJkoZSmhhCmh15xna1\n4MjT53sYvbTPu3zeaj6t8/3vzPcQemv1BbvO9xAk9czRLOKmWjw0T/NvssNm9a7v/8t8D2NSb3jM\na86vqkXzPY5xUz4vI0mSJGk0DOPKasPI4GdAuwDDZybsvr+qdpqsvSRJkqTRYfAzoKouAbabsaEk\nSZKkkWPwI0mSJI2wCoxNu/ixxg3f+nOSJEmSNAcMfiRJkiT1gmVvkiRJ0ohb5mpvnZj5kSRJktQL\nBj+SJEmSesGyN0mSJGmEFXG1t47M/EiSJEnqBYMfSZIkSb1g2ZskSZI04srV3jox8yNJkiSpFwx+\nJEmSJPWCZW+SJEnSiBuLOY0unCVJkiRJvWDwI0mSJKkXpix7S/Kk6U6sqntmfziSJEmSlkcBY672\n1sl0z/xcRjOXgzM5vl3ARnM4LkmSJEmaVVMGP1X1tJU5EEmSJEmaS51We0uyF7BZVb0/yYbAulV1\n/twOTZIkSdLMYtlbRzMueJDkI8ALgde2u34BfHwuByVJkiRJs61L5mfnqto+yQ8BqurOJI+d43FJ\nkiRJ0qzqEvw8kGQBzSIHJFkLGJvTUUmSJEnqzLK3brq85+co4MvAOkneDXwP+Lc5HZUkSZIkzbIZ\nMz9V9ekk5wO7trteWVWXzu2wJEmSJGl2dVrtDVgIPEBT+tYlWyRJkiRpJShgWSx766LLam8HAScA\n6wMbAp9P8o65HpgkSZIkzaYumZ8/A55TVb8ASPI+4IfAoXM5MEmSJEmaTV2Cn5sntFul3SdJkiRp\nCLjaWzdTlr0lOSLJvwN3ApclOSbJJ4BLgCUra4CSJEmSHp2SHJvktiSXDux7SpLTk1zZ/vnkKc7d\nLcmPk1yV5O1d+pvumZ9LgcuArwOHAOcA5wLvAb7R8X4kSZIkaSrHAbtN2Pd24L+qanPgv9rth0iy\nkOaVPLsDWwF7J9lqps6mLHurqk92H7MkSZKk+VCEsRFdkLmqzkyyyYTdLwN2ab8fD5wBvG1Cmx2B\nq6rqpwBJTmzPu3y6/mZ85ifJ04H30URUjx8Y6DNmOleSJEmSltO6VTW+xsAtwLqTtNkAuH5g+wZg\np5ku3CVEPA74FBCatNIXgJM6nCdJkiSp39ZOsnjgs9/ynFxVRfMqo1nRZbW3VavqtCQfrKqrgYOT\nLAb+abYGIUmSJGnF1fCu9rakqhYt5zm3Jlmvqm5Osh5w2yRtbgSeNrC9YbtvWl0yP/cnWQBcneQv\nk7wEWL3LqCVJkiRpOX0VeF37/XXA/52kzXnA5kk2TfJYYK/2vGl1CX7+AXgi8HfA84G/AP68w3mS\nJEmSNKUkJ9CsKr1FkhuSvAH4V+DFSa4Edm23SbJ+klMBqmop8GbgNOBHwBeq6rKZ+pux7K2qvt9+\nvRd47fLfkiRJkqS5NKovOa2qvac49KJJ2t4E7DGwfSpw6vL0N2Xwk+Rkpnm4qKpesTwdSZIkSdJ8\nmi7z85GVNgo9aqx35Srst/s68z0MaeVbsOt8j6C3Lrrv/PkeQq9dddua8z2E3nrR8zeb7yH01rIl\n8z0CrajpXnL6XytzIJIkSZKWXzG6ZW8r22i+ClaSJEmSlpPBjyRJkqRe6PKSUwCSPK6q7p/LwUiS\nJElafpa9dTNj5ifJjkkuAa5st7dN8uE5H5kkSZIkzaIuZW9HAnsCdwBU1UXAC+dyUJIkSZI027qU\nvS2oquuSh6TSls3ReCRJkiQthyIss+ytky7Bz/VJdgQqyULgb4GfzO2wJEmSJGl2dSl7+yvgAGAj\n4Fbgue0+SZIkSRoZM2Z+quo2YK+VMBZJkiRJK6Ase+tkxuAnySdoXhz7EFW135yMSJIkSZLmQJdn\nfr498P3xwB8B18/NcCRJkiRpbnQpeztpcDvJZ4DvzdmIJEmSJC0XX3LaTZcFDybaFFh3tgciSZIk\nSXOpyzM/d/HgMz8LgDuBt8/loCRJkiRptk0b/KR5s+m2wI3trrGqetjiB5IkSZLmRwHLyrK3LqYt\ne2sDnVOraln7MfCRJEmSNJK6PPNzYZLnzPlIJEmSJGkOTVn2lmSVqloKPAc4L8nVwM+B0CSFtl9J\nY5QkSZI0DVd762a6Z35+AGwPvHQljUWSJEmS5sx0wU8AqurqlTQWSZIkSZoz0wU/6yQ5YKqDVfXv\nczAeSZIkScuhCGXZWyfTBT8LgdXAmZQkSZI0+qYLfm6uqvestJFIkiRJ0hya8ZkfSZIkScNtrNMb\nbDTdLL1opY1CkiRJkubYlMFPVd25MgciSZIkSXNpurI3SZIkSSNgrHxipQuLAyVJkiT1gsGPJEmS\npF6w7E2SJEkaYQUsc6HmTsz8SJIkSeoFgx9JkiRJvWDZmyRJkjTSQrnaWydmfiRJkiT1gsGPJEmS\npF6w7E2SJEkaYQWMudpbJ2Z+JEmSJPWCwY8kSZKkXrDsTZIkSRplBctc7a0TMz+SJEmSesHgR5Ik\nSVIvWPYmSZIkjTBXe+vOzI8kSZKkXjD4GWJJDkpyWZKLk1yYZKckZyRZlOT77b7/SXJ7+/3CJLdO\nsX+TJNcmWbu9diU5fKCvtyY5ZGD7NW2/lyW5KMkxSdach2mQJEmSZoVlb0MqyfOAPYHtq+r+Nmh5\n7PjxqtqpbbcvsKiq3jzh/IftTx6SDr0feEWSQ6tqyYRzdwP+Adi9qm5MshB4HbAu8LNZu0lJkiTN\ninK1t07M/Ayv9YAlVXU/QFUtqaqbZvH6S4GjaYKciQ4C3lpVN7Z9L6uqY6vqx7PYvyRJkrRSGfwM\nr28BT0vykyQfTfK7c9DHUcA+SdaYsH9r4IKuF0myX5LFSRb/gttndYCSJEnSbDH4GVJVdR+wA7Af\ncDtwUlvKNpt93AN8Gvi7qdok2aZ9ZujqJK+a4jpHV9Wiqlq0KuvM5hAlSZI0ozA2pJ9hY/AzxNpy\nszOq6p+BNwN/PAfdfAh4A/DEgX2XAdu3Y7ikqrYDvgE8YQ76lyRJklYKg58hlWSLJJsP7NoOuG62\n+6mqO4Ev0ARA4w4FPphkw4F9Bj6SJEkaaa72NrxWAz7cLi+9FLiKpgTuS3PQ1+E0mSUAqurUJOsA\n32hXevsZcClw2hz0LUmSpEeggDFXe+vE4GdIVdX5wM6THNplQrvjgOMmOf9h+6tqk4Hvqw18vxVY\ndULb44Hjl2/UkiRJ0vCy7E2SJElSL5j5kSRJkkbcMsveOjHzI0mSJKkXDH4kSZIk9YJlb5IkSdKI\nqyF8oegwMvMjSZIkqRcMfiRJkiT1gmVvkiRJ0gjzJafdmfmRJEmS1AsGP5IkSZJ6wbI3SZIkaZRV\nfMlpR2Z+JEmSJPWCwY8kSZKkXrDsTZIkSRphzWpv8z2K0WDmR5IkSVIvGPxIkiRJ6gXL3iRJkqQR\nV6721omZH0mSJEm9YPAjSZIkqRcse5MkSZJGWLPam2VvXZj5kSRJktQLBj+SJEmSesGyN0mSJGnE\njWHZWxdmfiRJkiT1gsGPJEmSpF6w7E2SJEkaYQUsc7W3Tsz8SJIkSeoFgx9JkiRJvWDZmyRJkjTK\nKpRlb52Y+ZEkSZLUCwY/kiRJknrBsjdJkiRpxI2NWfbWhZkfSZIkSb1g8CNJkiSpFyx7kyRJkkbY\nKL/kNMkWwEkDuzYD3lVVHxposwvwf4Fr2l1fqar3rEh/Bj+SJEmS5kVV/RjYDiDJQuBG4ORJmn63\nqvZ8pP1Z9iZJkiRpGLwIuLqqrpurDsz8SJIkSaOsYGxEy94m2As4YYpjOye5mCYz9NaqumxFOjD4\nkSRJkjRX1k6yeGD76Ko6emKjJI8FXgq8Y5JrXABsVFX3JdkD+E9g8xUZjMGPJEmSpLmypKoWdWi3\nO3BBVd068UBV3TPw/dQkH02ydlUtWd7BGPxIkiRJI65Gv+xtb6YoeUvyG8CtVVVJdqRZt+COFenE\n4EeSJEnSvEnyRODFwJv+f3t3Hm5JVd/7//3pVgIqSBREZBBRHAAHBgFJ4oRxCmqCeqGNUQzqxRsn\njDhEHH+IA+DVKIZfO1zQ3OvAVQiKCqiAA6A2yKhGcEBBDODEICp0f+8fVS3bw+nuQ9N719lnvV/P\nU0/vWlV71+dUN4fzPWvVWiNtBwJU1dHAM4AXJbkZuBHYr6pqba5l8SNJmmp3fMrDho7QtAsf9fih\nIzTrTVd+aegIzVo8dIAFpqpuAO4+o+3okdfvA963Lq7lVNeSJEmSmmDPjyRJkjTFiiyUqa7Hzp4f\nSZIkSU2w+JEkSZLUBIe9SZIkSVNuxVrNfdYee34kSZIkNcHiR5IkSVITHPYmSZIkTbEqWL7C2d7m\nwp4fSZIkSU2w+JEkSZLUBIe9SZIkSVOuXOR0Tuz5kSRJktQEix9JkiRJTXDYmyRJkjTlVjjsbU7s\n+ZEkSZLUBIsfSZIkSU1w2JskSZI0xQoXOZ0re34kSZIkNcHiR5IkSVITHPYmSZIkTbOKs73NkT0/\nkiRJkppg8SNJkiSpCQ57kyRJkqZYAbVi6BTTwZ4fSZIkSU2w+JEkSZLUBIe9SZIkSVPO2d7mxp4f\nSZIkSU2w+JEkSZLUBIe9SZIkSdOsYMUKh73NhT0/kiRJkppg8SNJkiSpCQ57kyRJkqZYAcud7W1O\n7PmRJEmS1ASLH0mSJElNcNibJEmSNOXK2d7mxJ4fSZIkSU2w+JEkSZLUBIe9SZIkSVOsgBU1dIrp\nYM+PJEmSpCZY/EiSJElqgsPeJEmSpGlWYbmzvc2JPT+SJEmSmmDxI0mSJKkJFj8TkuR1SS5OckGS\n85Kc1v95aZLf9K/PS7Jnf/4mSW5KcuCMz/lxkk+N7D8jyTH96/2TXJ3k20kuSXLyys/rjx+T5Bn9\n69OTLBs5tmuS00f2d+vPuSTJuUlOSvLgcd0fSZIkrZ0CVqzIvNzmG5/5mYAkjwD2Bnauqt8n2QRY\nr6p+luTRwCurau8Zb3smcDawBDh6xrFdkmxfVd+Z5XKfqKoX99d9DPDpJI+pqu/Ocu49kjypqj4/\nI+9mwCeBZ1XVmX3bXwL3BS68DV+6JEmSNG/Y8zMZmwPXVNXvAarqmqr62RreswT4Z2CLJFvOOHYk\n8Lo1XbSqTgOWAi9cxSmHr+JzXgwcu7Lw6T/ra1V1wpquKUmSJM1XFj+TcQqwVZLvJ3l/kket7uQk\nWwGbV9U36Xpg9p1xyieBnZPcbw7XPhd44CqOnQX8oe8hGrVD/z5JkiRNgarMy22+sfiZgKq6HtiF\nrgfmauATSfZfzVv2pStwAD5O1ws0ajldr81r53D5Nf2rOxQ4ZLUfkHwjyXeTvGcVx1+YZFmSZb/l\n6jlEkiRJkibP4mdCqmp5VZ1eVW+kG1b29NWcvgTYP8mPgROBhyTZbsY5HwUeCWy1hkvvBMz2vM/K\nXF8GNgD2GGm+GNh55JzdgdcDd13FZyytql2ratc7seka4kiSJEnDsPiZgCQPmFG8PAy4bBXn3h+4\nS1VtUVXbVNU2wNuY0ftTVTcB/xM4aDXXfRRdb9MH1hDxUOBVI/tH0RVfe4603WkNnyFJkqQhFKxY\nMT+3+cbZ3ibjLsB7k2wM3AxcyqonIVgCHD+j7VPAJ4C3zGj/ELcesrZvPzPbnYAfAU9fxUxvf1RV\nn0ty9cj+z5PsC7wjyRbAVcA1s1xfkiRJmhoWPxNQVecAe67i2OnA6SP7b57lnAuAB/Wvtxlp/z1w\nr5H9Y4BjVpNj/5HXj55xbJcZ+2cDq52YQZIkSZomFj+SJEnSFFu5yKnWzGd+JEmSJDXB4keSJElS\nExz2JkmSJE2zguUOe5sTe34kSZIkNcHiR5IkSVITHPYmSZIkTbEizvY2R/b8SJIkSWqCxY8kSZKk\nJjjsTZIkSZpytWLoBNPBnh9JkiRJTbD4kSRJktQEh71JkiRJ06xgeTnb21zY8yNJkiSpCRY/kiRJ\nkprgsDdJkiRpihW4yOkc2fMjSZIkqQkWP5IkSZKa4LA3SZIkacqtcJHTObHnR5IkSVITLH4kSZIk\nNcFhb5IkSdI0Kyhne5sTe34kSZIkNcHiR5IkSVITHPYmSZIkTTEXOZ07e34kSZIkNcHiR5IkSVIT\nHPYmSZIkTbOC5S5yOif2/EiSJElqgsWPJEmSpCY47E2SJEmaYkWc7W2O7PmRJEmS1ASLH0mSJElN\ncNibJEmSNM0KarnD3ubCnh9JkiRJTbD4kSRJktQEh71JkiRJU6xwkdO5sudHkiRJ0mCS/DjJhUnO\nS7JsluNJ8q9JLk1yQZKd1/Za9vxIkiRJGtpjquqaVRx7ErBdv+0O/Fv/521m8SNJkiRNuQW+yOnT\ngI9UVQFnJ9k4yeZVdeVt/SCHvUmSJEkaUgFfTHJOkhfOcnwL4Kcj+5f3bbeZPT+SJEmSxmWTGc/x\nLK2qpTPO+cuquiLJPYBTk3yvqr4yjjAWP5Kkqbb9aYuHjtC0N532paEjNOtNhz9p6AjN+ux7Lhk6\nwp8qWDF/Z3u7pqp2Xd0JVXVF/+dVSY4HdgNGi58rgK1G9rfs224zh71JkiRJGkSSOyfZcOVr4PHA\nRTNOOxF4Tj/r2x7Ab9bmeR+w50eSJEnScDYDjk8CXW3yf6rqC0kOBKiqo4HPAU8GLgV+CzxvbS9m\n8SNJkiRNuczT2d5qTcerfgg8dJb2o0deF/BP6yKPw94kSZIkNcHiR5IkSVITHPYmSZIkTbOCxcvn\n57C3m4cOMIM9P5IkSZKaYPEjSZIkqQkOe5MkSZKmWIBF83eR03nFnh9JkiRJTbD4kSRJktQEh71J\nkiRJ06zConm6yOl8Y8+PJEmSpCZY/EiSJElqgsPeJEmSpCmX5UMnmA72/EiSJElqgsWPJEmSpCY4\n7E2SJEmaYilY7Gxvc2LPjyRJkqQmWPxIkiRJaoLD3iRJkqQpt2jF0Ammgz0/kiRJkppg8SNJkiSp\nCQ57kyRJkqZYChYtd7a3ubDnR5IkSVITLH4kSZIkNcFhb5IkSdKUi4uczok9P5IkSZKaYPEjSZIk\nqQkOe5MkSZKmWAoWLx86xXSw50eSJElSEyx+JEmSJDXBYW+SJEnSVAuLnO1tTuz5kSRJktQEix9J\nkiRJTXDYmyRJkjTNChY529uc2PMjSZIkqQkWP5IkSZKa4LA3SZIkaYoFiLO9zYk9P5IkSZKaYPEj\nSZIkqQkOe5MkSZKmWcFiZ3ubE3t+JEmSJDXB4mdKJLl+NcfOS/Lxkf0XJPnEyP5GSX6QZNskxyR5\nRt9+epJlI+ftmuT0kf3d+nMuSXJukpOSPHidf3GSJEnSBFj8TLkkDwIWA3+V5M598weBrZI8rt9/\nC/DhqvrhLB9xjyRPmuVzNwM+CfxLVW1XVTsDbwPuu86/CEmSJK21AItWzM9tvrH4mX5LgI8CpwBP\nA6iqAg4E3p1kV2Av4PBVvP9w4HWztL8YOLaqzlzZUFVfq6oT1mF2SZIkaWIsfqbfvsDHgY/RFUIA\nVNUFwMnAl4CXVNUfVvH+s4A/JHnMjPYdgHPXfVxJkiRpGBY/U6zv1bmmqn5CV+TslORuI6ccBVxR\nVaev4aMOBQ5Zw7W+keS7Sd4zy7EXJlmWZNlvufq2fRGSJEm6fQoWLc+83OYbi5/ptgR4YJIfAz8A\nNgKePnJ8Rb+tVlV9GdgA2GOk+WJg55FzdgdeD9x1lvcvrapdq2rXO7HpWnwZkiRJ0vhZ/EypJIuA\n/wY8uKq2qapt6J75WbLaN67aocCrRvaPAvZPsudI253W8rMlSZKkwbnI6fS4U5LLR/Y/QDek7Wcj\nbV8Btk+yeVVdeVs+vKo+l+Tqkf2fJ9kXeEeSLYCrgGvoZo6TJEnSPJJ5OLPafGTxMyWqarZeujfP\nOGc5cM+R/R8DO844Z/+R14+ecWyXGftnA49ay8iSJEnSvOKwN0mSJElNsOdHkiRJmmIpWDwPZ1ab\nj5lHXXgAACAASURBVOz5kSRJktQEix9JkiRJTXDYmyRJkjTlFi0fOsF0sOdHkiRJUhMsfiRJkiQ1\nwWFvkiRJ0hRLwaIVzvY2F/b8SJIkSWqCxY8kSZKkJjjsTZIkSZpycba3ObHnR5IkSVITLH4kSZIk\nNcFhb5IkSdI0q7B4ubO9zYU9P5IkSZKaYPEjSZIkqQkOe5MkSZKmWAoWOdvbnNjzI0mSJKkJFj+S\nJEmSmuCwN0mSJGnKLVoxdILpYM+PJEmSpCZY/EiSJElqgsPeJEmSpGlWEBc5nRN7fiRJkiQ1weJH\nkiRJUhMc9iZJkiRNsQCLXeR0Tuz5kSRJktQEix9JkiRJTXDYmyRJkjTNChY57G1O7PmRJEmS1ASL\nH0mSJElNcNibJEmSNMUCLHKR0zmx50eSJElSEyx+JEmSJDXBYW+SJEnSNCvIiqFDTAd7fiRJkiQ1\nweJHkiRJUhMc9iZJkiRNsQCLXeR0Tuz5kSRJktQEix9JkiRJTXDYmyRJkjTNanoXOU2yFfARYDOg\ngKVV9Z4Z5zwa+A/gR33Tp6vqLWtzPYsfSZIkSUO5Gfjnqjo3yYbAOUlOrarvzDjvq1W19+29mMPe\nJEmSJA2iqq6sqnP719cB3wW2GNf17PnROnUl51zzZnLZ0Dluh02Aa4YO0Sjv/bC8/8Px3g9nqu/9\nmw8eOsHtNs33/95DB/gTBYsWwGxvSbYBdgK+McvhPZNcAFwBvLKqLl6ba1j8aJ2qqk2HznB7JFlW\nVbsOnaNF3vthef+H470fjvd+WN7/ZmySZNnI/tKqWjrzpCR3AT4FvLyqrp1x+Fxg66q6PsmTgROA\n7dYmjMWPJEmSpHG5Zk1FbpI70hU+/7uqPj3z+GgxVFWfS/L+JJtU1W3uObT4kSRJkqZYmN5hb0kC\nfAj4blW9axXn3BP4r6qqJLvRzVvwi7W5nsWP9Kdu1Q2rifHeD8v7Pxzv/XC898Py/gvgL4B/AC5M\ncl7f9i/A1gBVdTTwDOBFSW4GbgT2q6pam4tlLd8nSZIkaR7YaJNd6uF7nz10jFl9+dj1zplPz3bZ\n8yNJkiRNsyle5HTSXOdHkiRJUhMsfiRJkiQ1wWFvkiYuyVOAC6rqsn7/DcDTgcuAl1XVj4bM15Ik\ndwceCfykqs4ZOs9Cl2QjYLOquqTffyawQX/45Kr6r8HCSZpa0zzb26TZ86MmJTkgycEj+1ckuTbJ\ndUkOHDJbI94KXA2QZG/g2cA/AicCRw+Ya8FL8tkkO/avNwcuorv3H03y8kHDteEIupmNVnob8HC6\nAvTNgyRqRJIdkjx1ZP9/Jvlwv+08ZLYWeP81X1j8qFUHAh8e2b+qqjYCNgWWDBOpKVVVv+1f7wN8\nqKrOqaoP0v0daHzuU1UX9a+fB5xaVU8BdqcrgjReDweOHdm/rqpeUlXPB3YcKFMr3g6MLoj4BOAk\n4DTgDYMkaov3X/OCw97UqlTV6OJYxwFU1e+SbLCK92jdSZK7AL8F9gLeP3Js/WEiNeOmkdd7AR8A\nqKrrkqwYJlJT7jBjbYp/GHm98aTDNGbzqjpzZP/aqvoUQJL/PlCmlnj/x6kc9jZXFj9q1Z/8kFFV\nhwEkWQRsMkiitrwbOA+4lm5F52UASXYCrhwyWAN+muQlwOXAzsAXAPqi/45DBmvEiiT3rKqfA6zs\nhUuyBWDxOV4bju5U1R4ju/eYcJYWef81LzjsTa06Jcmhs7S/BThl0mFaU1UfBh4FHAA8eeTQz4H9\nh8jUkAOAHeju875V9eu+fQ/gfw0VqiGHA59J8sgkG/bbo4AT+mMan58l2X1mY5I9gJ8NkKc13n/N\nC/b8qFUHAx9Mcilwft/2UGAZ8PzBUjWkqq4ArpjRvCHd380LJp+oDVV1Fd0zbzPbT0vygwEiNaWq\n/j3JNcChdEUodJNOvKGqPj9csia8GvhEkmOAc/u2XYDnAvsOFaoh3v9xctjbnFn8qElVdQOwJMm2\n3PIDyHeqyh/+JiDJQ+hmvboX3W+8jwLeR/fQ/ZEDRmtCkkcAWwBfqaqr+r+P1wB/BWw1aLgGVNUX\n6IcbanKq6pt9L8M/cUsP88XAHk4xPn7ef80XFj9qUpKt+5c3c0vPzx/bq+onQ+RqyAeAfwPOAp5I\n9/zPscDfV9Xvhgy20CU5HNib7p6/OsnJdL2db8PZ3sauX9NqVaqq/r+JhWlQ/0O2M4sNxPuv+cDi\nR606CSi6dcFWKrpplu8BLB4iVEP+rKqO6V//Z5KXVdWrhgzUkL8BdupnNvxz4KfAjlX142FjNeOG\nWdruTPcs1t0Bi58xSXIa3ff52VRV7TXJPK3x/o9XCIuWZ80nyuJHbaqqB4/uJ9mGbjzy44DDBojU\nmvX7md1Wfqf+/eh+VZ27ynfq9vrdyt61qvpVkkssfCanqv44rDPJhsDL6NZb+jgO+Ry3V87Stgfw\nKuCqCWdpkfdf84LFj5qWZDvgddzyrMlLq+qm1b9L68DPgXetYr+Ax048UTu2TXLiyP59Rver6qmz\nvEfrUJK7Aa8A/p5uuOfOVfWrYVMtfFV1zsrX/Qx7r6dbV+xAJ5sYP++/5guLHzUpyY50Rc8OwDuB\nA6rKeVImpKoePXSGhj1txr69DRPUP3O1D7AUeHBVXT9wpKYkeQJwCPB74K1VddrAkZri/R8jZ3ub\ns/zpQtNSG5Isp3vW4STgVt8uquqlEw/VkCT7rO54VX16UlmkSUqygu4Hv5v50+cfQvfcw0aDBGtA\nkm/RPdd5ON1kK3/C4bbj5f0fr4033rUe9chvDB1jVid+5g7nVNWuQ+dYyZ4fteoAVv3gpcbvKas5\nVoDFz5gkuZDZ/+2v/OH7IROO1JSqcnHx4dwAXA88o99GOdx2/Lz/mhcsftSkkZnGNICqet6qjiXZ\nbJJZGrT30AFa1j/vs0pV9ctJZWmNw22H5f0frzjsbc4sftSkJJ9hNT0/PvQ9WUk2Bp4OPAt4EN3i\npxqDqrpstvYkfwksoVuAUONzDreeZn+lAradbJx2ONx2WN5/zRcWP2rVEUMHaF2SDegevn8WsBOw\nIfC3wFeGzNWSfnrxZwHPBH6Eww0n4dGrKkA1dg63HZb3X/OCxY9atV5VnTrbgSTvAM6YcJ6mJPk/\nwF8BpwDvBb4MXFpVpw+ZqwVJ7k/Xw7MEuAb4BN3kN48ZNFg7jgd2HjpEi1Y33FYT8WbXFBsvh73N\njQ9eqlVHJfmb0YYki5IcAzx0mEhN2R74FfBd4Lv9NONOQDEZ36N7sHjvqvrLqnovs8x4qLFxCfYB\nJXlAkiOTnNRvR/S/END4fTHJa5L4i3cNyn+AatUTgM8nWa+qju+HYB0HXMvqu+a1DlTVw5I8kK73\n4YtJrgE2TLJZVf3XwPEWun2A/YDTknwB+Dj+QD5JWyT511UddJr98UnyCLqhVUv7LXRDbk9Psk9V\nnT1kvgbsBLwFOCfJi6vqq0MHUpssftSkqvpRkscBJ/eziz0b+FZVHTRwtGZU1feANwJvTLILXSH0\nrSSXV9Wew6ZbuKrqBOCEJHeme+bq5cA9kvwbcHxVnTJowIXvRrpJDzR5bwCWzBhee0KSL9N9L3rS\nIKkaUVXXAQf13++/lORyYAVOs79OdLO9+XusuXCRUzUpycox9/cCjgVOBd658riLrY1X/1u/983S\nHuCvqspJD8YkyR2q6uYZbX9ON+nBvlW11zDJ2pDk3KrymZ8BJPl+Vc06xC3Jf1bVAyadqTVJHgu8\nBzgZOIqu+AFWPROl5uZuG+1ae+3+zaFjzOr/fnGxi5xK88CRI68vADYbaXOxtfH7R+BWxU91v42x\n8BmvbzLjgfuq+hW3DAXSeG0+dICGXbeaYzdMLEWjknwc2BJ4VlVdOHQetcviR01a3cxWSfaYZBZp\nwhwXMayfDx2gYVut4nmrAFtMOkyDvlhVH5ztgM97rhvO9jY3Fj/SrX0S2HroEAvcQ5JcO0v7yrHf\nG006UEM2TfKKVR2sqndNMkyDHGs+nINXc2zZxFI0ambh4+LWGorFj3Rr/mZ8/C6sqp2GDtGoxcBd\n8N/5ULZ0trdhVNWxQ2donYtbaz6w+JFuzd/MaiG7sqreMnSIhjnb20CS/C9W/f29quqASeZpjYtb\nj1k57G2uLH7UpCSfYfb/CQa4+4TjtOi4oQM0zB6fYf3CHojBfHaWtq2Ag+h6RDVet1rcOom/bNTE\nWfyoVUes5TGtG1cn2a6qLumnt/4w3djvHwP7O9X4WD0tyR2r6iboVrwHngxcVlWfHjZaE/4wdIBW\nVdWnVr5Osi3wL8AjgbcDHxoqVytc3FrzhcWPmlRVZ8zWnmQrYD9g1uNaZ14GHNO/XgI8BLgP3Rjw\n99ANjdB4/DtwAHBJkvsBZwH/G9g7ycOr6rWDplv4/mlknbFbsfAfr/6H70PovtccDhw4c90rjY+L\nW49PHPY2ZxY/al6STekWeFxCN9vM8cMmasLNK3segL2Bj1TVL+h+G/jO1bxPt9+fV9Ul/evnAh+r\nqpckWY/uWRSLn/E6gm7I7crhhzOH/bjG2JgkOQ7YhW5Nt4OA5cBGXeczVNUvh0vXnqo6Bzgnyavo\nClJpIix+1KQkGwL70M04c3/g08B9qmrLQYO1Y0WSzenGf+8FvHXk2AbDRGrG6A/bj6X77TdV9Yck\nK2Z/i9ahVwM/raorAZI8l1uGfL5puFhNeDjdv/9XAv/ct40WodsOEap1VbUiyfMBJ2LRRFj8qFVX\n0a10fwjwtaqqJH83cKaWvIFuXY3FwIlVdTFAkkcBPxwyWAMuSHIEcAVwP7qZl1auuaHxOxp4HECS\nRwJvA14CPAxYCjxjuGgLW1VtM3QGrZITsawDDnubm0VDB5AG8lrgz4D3A69Nct+B8zSlqj4L3Bt4\nUFW9YOTQMmDfYVI14wXANcA2wOOr6rd9+/Y42cckLB4ZXrUvsLSqPlVVr6crRjVBSe6b5PVJLh46\nS+Oc9U0TY/GjJlXVu6tqD7rF1gBOAO6V5NVJ7j9gtCYk2Q74v8BXk3wsyRYAVXVDVV0/bLqFrapu\nrKq3V9XLqur8kfYzq+qjQ2ZrxOIkK0dd7EW31slKjsaYgCT3SnJQkm8BF9P9LLTfwLEWvCTXJbl2\nlu06uudtpYnwG62alOTlwNeBb1fVYcBhSXakm/Tgc/gb2HH7MPARulW9n0q34N0+gyZqRJLTWP1C\nj3tNMk+DPgac0U/zeyPwVYB+5r3fDBlsoUvyQrrv8VsAn6Sb9fA/qurNgwZrRFVtOHSGhayb7c3R\ng3Nh8aNWbUk3pfIDk1xIVwidCRxZVa8bNFkbNqyqD/SvD0/i9L6T88pZ2vYAXkX3LJzGqKremuRL\nwObAKVW1shBdRPfsj8bnfXRTuz+rqpYBuMim1B6LHzWpql4J0E/vuyuwJ/A8YGmSX1fV9kPma8D6\nSXbilodcNxjdd62T8emnlwX+OMHE64H16dY7+fxgwRpSVWfP0vb9IbI0ZnO6ZQ2OTHJPut6fOw4b\nSdKkWfyodRsAGwF37befARcOmqgNPwfetYr9wrVOxirJE+hmOvw98NaqOm3gSNLY9WuJHQ0cnWRL\nugkn/ivJd4Hjq+pfBg0o3U7O9jY3Fj9qUpKlwA7AdcA36Ia8vauqfjVosEZU1aOHztCq/iHvTenW\n9zmrb9t55XF73bRQJdljZa9bVV1Ot9jpkf0kN054IDXC4ket2ppuqutL6NY7uRz49aCJGpJk5uQG\nRTf98nlVdd0AkVpyA3A93XoyM9eUsddNC9n7gZ1nNvZDDl1gU2qExY+aVFVPTBK63p896Vb73jHJ\nL4GzquqNgwZc+J4yS9vdgIckOaCqvjzLca0D9rpJ0gJUDnubK4sfNaufZemiJL+mm2L2N8DewG6A\nxc8YVdXzZmtPcm+6h5B3n2yidiQ5n252w68DZ1bVjwaOJE3KtklOXNXBqnrqJMNIGobFj5qU5KV0\nPT57AjfRPfNzJt36M054MJCquiyJsy+N19/T/bv/a+CNSe5M9+zPymLoG0OGk8boarrnfCQ1zOJH\nrdoGOA44qKquHDiLekkeQDcDmcakqi4CLgKWAiTZhO5h75cDRwCLh0snjdX1VXXG0CGkcYjD3ubM\n4kdNqqpXDJ2hZUk+Q/dw/ai70a3D8ezJJ2pHksXATnS9P38B3Jdu0o8P0s/+Ji1Qv0pyz6r6OUCS\n5wBPBy4D3lRVvxw0naSJsPiRNIQjZuwX8Avgkqr6wwB5WnId8B3gKOA1PvOjhmwM/AEgySOBtwMv\nAR5G1xM6c/ZDSQuQxY+kiZvr0JMkZ1XVI8adpzEHAI8Ang88r1/35yy6WQ6vGDSZNF6LRnp39gWW\nVtWngE8lOW/AXNI64bC3ubH4kTSfrT90gIWmqj4GfAwgyZ3oZjfcE3hbkvWq6t5D5pPG6A5J7lBV\nNwN7AS8cPTZQJkkT5n/skuazmc8FaR3oZ3jbnVue+3k48FO6Gd+khepjwBlJrgFuBL4KkOR+dEsd\nSGqAxY8kNSTJt4GtgHPoip0jgbOr6vpBg0ljVlVvTfIluolVTunXegNYRPfsjzS1nO1t7ix+JM1n\nGTrAAvRc4MKRH/ykZlTV2bO0fX+ILJKGsWjoAJK0Gv8wdICFpqouAHZIcmySZf12bJKHDJ1NkqRx\ns/iRNHFJDkhy8Mj+FUmuTXJdkgNXtvcLcmodSvI04HjgDOAf++0MuhmvnjZkNknSWipYdPP83OYb\nh71JGsKBwBNH9q+qqi2SrA+cDBw9TKwmvAX466r68UjbBUm+DPxHv0mStCDZ8yNpCKmqX4zsHwdQ\nVb8DNhgmUjPuMKPwAaBvu+PE00iSNEH2/EgawsajO1V1GECSRcAmgyRqx81Jtq6qn4w2Jrk3MA8H\nKEiS5mLRcucImgt7fiQN4ZQkh87S/hbglEmHacwbgS8m2T/Jg/vteXT3/Q0DZ5Mkaazs+ZE0hIOB\nDya5FDi/b3sosAx4/mCpGlBVJyT5EfDP3LK2yXeA/1ZV56/6nZIkTT+LH0kTV1U3AEuSbAvs0Dd/\np6p+MGCsZvRFznOGziFJWjdc5HTuHPYmaeKSbJ1ka7pnTM7vt5tG2jVGSZ6b5JwkN/TbsiQWQ5Kk\nBc+eH0lDOAkoYPTpzAI2Be4BLB4iVAuSPBd4OfAK4Fy6v4OdgcOTVFV9dMh8kiSNk8WPpImrqgeP\n7ifZBng18DjgsAEiteRFwN/NmO76y0meDnwcsPiRpCnksLe5cdibpMEk2S7JMcDngXOA7avqvcOm\nWvA2Ws06PxtNPI0kSRNkz4+kiUuyI/A6uskO3gkcUFX+zmoyblzLY5IkTT2LH0lDOB/4Kd2zP7sB\nuyW3PP5TVS8dKFcLHpTkglnaA2w76TCSpNvP2d7mzuJH0hAOoJvgQJP3oKEDSJI0FIsfSRNXVccM\nnaFVVXXZXM5LclZVPWLceSRJmiSLH0kTl+QzrKbnp6qeOsE4mt36QweQJM2Rw97mzOJH0hCOGDqA\n1shhiZKkBcfiR9IQ1quqU2c7kOQdwBkTziNJkhpg8SNpCEclOaiqTlrZkGQR8GHgnsPF0ois+RRJ\n0nzhsLe5sfiRNIQnAJ9Psl5VHZ9kA+A44FrgKcNGU+8fhg4gSdK6tmjoAJLaU1U/Ah4HHJrkQOBU\n4JKqelZV3TRsuoUtyQFJDh7ZvyLJtUmu6/8uAKiqi4ZJKEnS+NjzI2nikuzcv3w1cCxd8fPRle1V\nde5Q2RpwIPDEkf2rqmqLJOsDJwNHDxNLkrS2XOR07ix+JA3hyJHXFwCbjbQV8NiJJ2pHquoXI/vH\nAVTV7/rhh5IkLVgWP5Imrqoes6pjSfaYZJYGbTy6U1WHwR8nnNhkkESSJE2IxY+k+eaTwNZDh1jA\nTklyaFUdMqP9LcApQwSSJN1OBYtuHjrEdLD4kTTfOMXyeB0MfDDJpcD5fdtDgWXA8wdLJUnSBFj8\nSJpvaugAC1lV3QAsSbItsEPf/J2q+sGAsSRJmgiLH0kTl+QzzF7kBLj7hOM0JcnKIYU3c0vPzx/b\nq+onQ+SSJK294Gxvc2XxI2kIR6zlMd1+J9EVnqPDCwvYFLgHsHiIUJIkTYLFj6SJq6ozZmtPshWw\nHzDrcd1+VfXg0f0k29Ctt/Q44LABIkmSNDEWP5IGlWRT4JnAEuBewPHDJmpDku2A1wG7062x9NKq\numnYVJKkteIip3Nm8SNp4pJsCOwDPAu4P/Bp4D5VteWgwRqQZEe6omcH4J3AAVXl/zIlSU2w+JE0\nhKuAbwKHAF+rqkrydwNnasX5wE/pnv3ZDdgtueXxn6p66UC5JEkNSvJE4D10z5x+sKrePuN4+uNP\nBn4L7F9V567t9Sx+JA3htXTP9rwf+FiSTwycpyUH4HTikrTgTOOwtySLgaOAvwYuB76V5MSq+s7I\naU8Ctuu33YF/6/9cKxY/kiauqt4NvLtfa2Y/4ATgXkleDRxfVd8fNOACVlXHDJ1BkqTebsClVfVD\ngCQfB54GjBY/TwM+UlUFnJ1k4ySbV9WVa3NBix9JE5fk5cDXgW9X1WHAYf2zKEuAzwH3GzLfQraa\nNZYAqKqnTjCOJKltW9ANxV7pcm7dqzPbOVsAFj+SpsaWdON3H5jkQrpC6EzgyKp63aDJFj7XUZKk\nBeZKzjn5TWSToXOswvpJlo3sL62qpUOFsfiRNHFV9UqAJOsBuwJ7As8Dlib5dVVtP2S+BW69qjp1\ntgNJ3oFrLEnS1KmqJw6dYS1dAWw1sr9l33Zbz5mzRWv7RklaBzYANgLu2m8/A74xaKKF76gkfzPa\nkGRRkmOAhw4TSZLUqG8B2yW5T/8L0f2AE2eccyLwnHT2AH6zts/7gD0/kgaQZCndOjPX0RU7ZwLv\nqqpfDRqsDU8APp9kvao6PskGwHHAtcBTho0mSWpJVd2c5MXAyXRTXX+4qi5OcmB//Gi6Z4GfDFxK\nN9X1827PNdNNnCBJk5PkC8AmwEV0hc9ZwEXlN6SJSLIl3f9o3gs8G/hWVR00bCpJksbP4kfSIPpF\ny3age95nT2BH4JfAWVX1xiGzLWRJdu5f3gs4FjgVeOfK47dn4ThJkuY7ix9Jg+p7If6CrgDaG7h7\nVW08bKqFK8lpqzlcVfXYiYWRJGnCLH4kTVySl3JLj89NdEPfVm4XVtWKAeM1K8keVXX20DkkSRoX\nix9JE5fkXfRr+9yeGVu0biX5SVVtPXQOSZLGxeJHkgRAkp9W1VZrPlOSpOnkOj+SpJX8bZgkaUFz\nnR9JakiSzzB7kRPg7hOOI0nSRDnsTZIakuRRqzteVWdMKoskSZNm8SNJIslWwH5VdfjQWSRJGhef\n+ZGkRiXZNMn/SPJV4HRgs4EjSZI0Vj7zI0kNSbIhsA/wLOD+wKeB+1TVloMGkyRpAhz2JkkNSXIj\n8E3gEOBrVVVJflhV2w4cTZKksXPYmyS15bXAnwHvB16b5L4D55EkaWLs+ZGkBiXZFtgPWAJsB7wR\nOL6qvj9oMEmSxsjiR5IakuTlwNeBb1fVzX3bjnRF0L5Vdb8h80mSNE4WP5LUkCRHAHsCDwQupCuE\nzgTOrKpfDplNkqRxs/iRpAYlWQ/Yla4QekS//bqqth80mCRJY+RU15LUpg2AjYC79tvP6HqCJEla\nsOz5kaSGJFkK7ABcB3wDOBs4u6p+NWgwSZImwKmuJaktW9NNdf1z4ArgcuDXgyaSJGlC7PmRpMYk\nCV3vz579tiPwS+CsqnrjkNkkSRonix9JalSSLYG/oCuA9gbuXlUbD5tKkqTxsfiRpIYkeSm39Pjc\nRD/Ndb9dWFUrBownSdJYOdubJLVlG+A44KCqunLgLJIkTZQ9P5IkSZKa4GxvkiRJkppg8SNJkiSp\nCRY/kqTbJMnyJOcluSjJcUnudDs+69FJPtu/fmqS16zm3I2T/I+1uMabkrxyru0zzjkmyTNuw7W2\nSXLRbc0oSZoMix9J0m11Y1U9rKp2BP4AHDh6MJ3b/P+Xqjqxqt6+mlM2Bm5z8SNJ0koWP5Kk2+Or\nwP36Ho//TPIR4CJgqySPT3JWknP7HqK7ACR5YpLvJTkX2GflByXZP8n7+tebJTk+yfn9tifwduC+\nfa/T4f15Byf5VpILkrx55LNel+T7Sb4GPGBNX0SSF/Sfc36ST83ozXpckmX95+3dn784yeEj1/7v\nt/dGSpLGz+JHkrRWktwBeBJwYd+0HfD+qtoBuAE4BHhcVe0MLANekWR94APAU4BdgHuu4uP/FTij\nqh4K7AxcDLwG+EHf63Rwksf319wNeBiwS5JHJtkF2K9vezLw8Dl8OZ+uqof31/sucMDIsW36a/wN\ncHT/NRwA/KaqHt5//guS3GcO15EkDch1fiRJt9UGSc7rX38V+BBwL+Cyqjq7b98D2B74ehKA9YCz\ngAcCP6qqSwCS/Dvwwlmu8VjgOQBVtRz4TZI/n3HO4/vt2/3+XeiKoQ2B46vqt/01TpzD17RjkkPp\nhtbdBTh55Ngn+8VfL0nyw/5reDzwkJHnge7aX/v7c7iWJGkgFj+SpNvqxqp62GhDX+DcMNoEnFpV\nS2ac9yfvu50CvK2q/v8Z13j5WnzWMcDfVtX5SfYHHj1ybOaCeNVf+yVVNVokkWSbtbi2JGlCHPYm\nSRqHs4G/SHI/gCR3TnJ/4HvANknu25+3ZBXv/xLwov69i5PcFbiOrldnpZOBfxx5lmiLJPcAvgL8\nbZINkmxIN8RuTTYErkxyR+DvZxx7ZpJFfeZtgf/sr/2i/nyS3D/JnedwHUnSgOz5kSStc1V1dd+D\n8rEkf9Y3H1JV30/yQuCkJL+lGza34Swf8TJgaZIDgOXAi6rqrCRf76eS/nz/3M+DgLP6nqfrgWdX\n1blJPgGcD1wFfGsOkV8PfAO4uv9zNNNPgG8CGwEHVtXvknyQ7lmgc9Nd/Grgb+d2dyRJQ0nVzN58\nSZIkSVp4HPYmSZIkqQkWP5IkSZKaYPEjSZIkqQkWP5IkSZKaYPEjSZIkqQkWP5IkSZKaYPEj8pZM\n4wAAABNJREFUSZIkqQkWP5IkSZKa8P8AJxB+N7m8Jc4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11a434e80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "print(\"Precision: {}%\".format(100*metrics.precision_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"Recall: {}%\".format(100*metrics.recall_score(Yt, Y_pred, average=\"weighted\")))\n",
    "print(\"f1_score: {}%\".format(100*metrics.f1_score(Yt, Y_pred, average=\"weighted\")))\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion Matrix:\")\n",
    "confusion_matrix = metrics.confusion_matrix(Yt, Y_pred)\n",
    "print(confusion_matrix)\n",
    "normalised_confusion_matrix = np.array(confusion_matrix, dtype=np.float32)/np.sum(confusion_matrix)*100\n",
    "\n",
    "print(\"\")\n",
    "print(\"Confusion matrix (normalised to % of total test data):\")\n",
    "print(normalised_confusion_matrix)\n",
    "print(\"Note: training and testing data is not equally distributed amongst classes, \")\n",
    "print(\"so it is normal that more than a 6th of the data is correctly classifier in the last category.\")\n",
    "\n",
    "# Plot Results: \n",
    "width = 12\n",
    "height = 12\n",
    "plt.figure(figsize=(width, height))\n",
    "plt.imshow(\n",
    "    normalised_confusion_matrix, \n",
    "    interpolation='nearest', \n",
    "    cmap=plt.cm.rainbow\n",
    ")\n",
    "plt.title(\"Confusion matrix \\n(normalised to % of total test data)\")\n",
    "plt.colorbar()\n",
    "tick_marks = np.arange(len(LABELS))\n",
    "plt.xticks(tick_marks, LABELS, rotation=90)\n",
    "plt.yticks(tick_marks, LABELS)\n",
    "plt.tight_layout()\n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}