CRISP-RCNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "##MCC-f1-installation\n",
    "#!pip install py-mcc-f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "evUCEDS8qSzO"
   },
   "outputs": [],
   "source": [
    "#change keras if original not working\n",
    "# TF should be TF 2.1\n",
    "# Keras==2.3.1\n",
    "# tensorflow-addons==0.9.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "nGoPPPB9qS6J",
    "outputId": "f5cd31a1-7a09-497a-c767-ed8bcaccc168"
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "from tqdm import tqdm\n",
    "from natsort import natsorted\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from tensorflow.keras.models import Sequential, Model\n",
    "from tensorflow.keras.layers import Layer,Input, Dense, Dropout, Activation, Concatenate, Flatten, MaxPooling2D, Convolution2D, Convolution1D, MaxPooling1D, GlobalMaxPooling1D, BatchNormalization, LSTM, GRU, Bidirectional, Lambda\n",
    "from tensorflow.keras.regularizers import l2,l1\n",
    "from tensorflow.keras.optimizers import SGD,Adam,RMSprop\n",
    "from tensorflow.compat.v1 import InteractiveSession\n",
    "import tensorflow.keras.backend as K\n",
    "from tensorflow.keras.preprocessing.image import array_to_img, img_to_array, load_img\n",
    "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint,ReduceLROnPlateau\n",
    "from tensorflow.keras.models import load_model\n",
    "import tensorflow.keras.metrics\n",
    "import tensorflow_addons as tfa\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "import tensorflow as tf\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from tensorflow.keras.utils import to_categorical, plot_model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "m5odlI1HrcWE"
   },
   "outputs": [],
   "source": [
    "le = LabelEncoder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "hc2Uokyyrflj",
    "outputId": "b7509972-7278-4c36-ae5e-6a06a87290b4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.3.0\n"
     ]
    }
   ],
   "source": [
    "print(tf.__version__) #should be TF 2.1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "k57gCOUCsiMV"
   },
   "outputs": [],
   "source": [
    "samples = 153924\n",
    "# samples #should be 153924"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "id": "V4bjy-a1as-E"
   },
   "outputs": [],
   "source": [
    "#load directly from npy files #this step skips the data loading from folders as images\n",
    "off_dataset = np.load('off_dataset.npy')\n",
    "target_dataset = np.load('target_dataset.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "1xUN2Iqnsoao",
    "outputId": "ac0343ec-5160-4e6c-eeec-99a0e02d37a8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((153924, 4, 28, 1), (153924, 4, 28, 1))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "off_dataset.shape, target_dataset.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "id": "iRmOStIxuV6l"
   },
   "outputs": [],
   "source": [
    "#get both types of output\n",
    "output = pd.read_csv('output2.csv')\n",
    "y = output['y']\n",
    "freq = output['freq']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "id": "8eYFGj3Muj5r"
   },
   "outputs": [],
   "source": [
    "Y = le.fit_transform(y).reshape(-1,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "d3l81yZMunIV",
    "outputId": "fba633d7-da8a-482c-e4b8-754ed7eabfe0"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[1],\n",
       "        [1],\n",
       "        [1],\n",
       "        ...,\n",
       "        [0],\n",
       "        [0],\n",
       "        [0]]), (153924, 1))"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y, Y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "#len(Y[Y==0]) #7190 is positive dataset, 146734 is negative dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "9P-fXUvcupG8",
    "outputId": "4864931b-16a6-4fca-b9e9-efe6652f3b25"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([[39.],\n",
       "        [39.],\n",
       "        [38.],\n",
       "        ...,\n",
       "        [ 0.],\n",
       "        [ 0.],\n",
       "        [ 0.]]), (153924, 1))"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "freq = freq.values.reshape(-1,1)\n",
    "freq, freq.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "6HHDQGaQurfH",
    "outputId": "924dde4b-a5c9-4de2-fa96-9dd5fa18be25"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((153924, 4, 28, 1), (153924, 4, 28, 1), (153924, 1), (153924, 1))"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#data summary\n",
    "off_dataset.shape, target_dataset.shape, Y.shape, freq.shape\n",
    "#should be ((153924, 4, 28, 1), (153924, 4, 28, 1), (153924, 1), (153924, 1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "id": "d4WlRSMDuujH"
   },
   "outputs": [],
   "source": [
    "#now model\n",
    "def clear_sess():\n",
    "  try:\n",
    "    del model, history\n",
    "  except:\n",
    "    pass\n",
    "  from tensorflow.keras import backend as K\n",
    "  K.clear_session()\n",
    "  return None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "ZkplbAmPuyIo"
   },
   "outputs": [],
   "source": [
    "clear_sess()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "y6C8ulHEe6in"
   },
   "outputs": [],
   "source": [
    "# # CNN-LSTM tester code works good; ############################################## DO NOT EDIT\n",
    "\n",
    "# #input_1\n",
    "\n",
    "# input_1 = Input(shape = (4,28,1), name='i_1')\n",
    "# conv1_1 = Convolution2D(128, 3, activation = 'relu',padding='valid')(input_1)\n",
    "# bn1_1 = BatchNormalization()(conv1_1)\n",
    "# conv1_2 = Convolution2D(128, 2, activation = 'relu',padding='valid')(bn1_1)\n",
    "# bn1_2 = BatchNormalization()(conv1_2)\n",
    "# #pool1_1 = MaxPooling2D(pool_size=2)(conv1_2)\n",
    "# #testing reshape layer\n",
    "# reshape_1 = Lambda(lambda x: K.squeeze(x, 1))(bn1_2)\n",
    "# bilstm_1 = Bidirectional(LSTM(128, return_sequences=True))(reshape_1)\n",
    "# #bilstm_11 = Bidirectional(LSTM(256, return_sequences=True))(bilstm_1)\n",
    "# flat_1 = Flatten()(bilstm_1)\n",
    " \n",
    "# # kernel_regularizer=l2(0.01), bias_regularizer=l2(0.01)\n",
    "# #\n",
    "# #\n",
    "# #input_2\n",
    "\n",
    "# input_2 = Input(shape = (4,28,1), name='i_2')\n",
    "# conv2_1 = Convolution2D(128, 3, activation = 'relu',padding='valid')(input_2)\n",
    "# bn2_1 = BatchNormalization()(conv2_1)\n",
    "# conv2_2 = Convolution2D(128, 2, activation = 'relu',padding='valid')(bn2_1)\n",
    "# bn2_2 = BatchNormalization()(conv2_2)\n",
    "# #pool2_1 = MaxPooling2D(pool_size=2)(conv2_2)\n",
    "# # testing reshape layer\n",
    "# reshape_2 = Lambda(lambda x: K.squeeze(x, 1))(bn2_2)\n",
    "# bilstm_2 = Bidirectional(LSTM(128, return_sequences=True))(reshape_2)\n",
    "# #bilstm_21 = Bidirectional(LSTM(256, return_sequences=True))(bilstm_2)\n",
    "# flat_2 = Flatten()(bilstm_2)\n",
    " \n",
    "# # concatenate\n",
    "# concat   = Concatenate()([flat_1, flat_2])\n",
    "# dense1   = Dense(256, activation = 'relu', kernel_regularizer=l2(0.0005))(concat)\n",
    "# #dropout1 = Dropout(0.5)(dense1)\n",
    "# dense2   = Dense(256, activation = 'relu')(dense1)\n",
    "\n",
    "# #dense3   = Dense(512, activation = 'relu')(dense2) #goes to regression\n",
    "\n",
    "\n",
    "# #dense4  = Dense(1024, activation = 'relu')(dense1)\n",
    "# dense4  = Dense(256, activation = 'relu')(concat)\n",
    "# #dropout2 = Dropout(0.2)(dense4) #goes to classification\n",
    "# #dense5  = Dense(128, activation = 'relu')(dropout2) \n",
    "\n",
    "\n",
    "# #regression output\n",
    "# output_1  = Dense(1, activation = 'linear', name='r')(dense2)\n",
    "\n",
    "\n",
    "# #classification output\n",
    "# output_2  = Dense(1, activation = 'sigmoid', name='c')(dense4)\n",
    " \n",
    "# # create model with two inputs\n",
    "# model = Model(inputs=[input_1,input_2], outputs=[output_1,output_2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# import tensorflow.compat.v1 as tf\n",
    "# tf.disable_v2_behavior()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# ###################### TESTING ONLY CNN ###########\n",
    "\n",
    "# #input_1\n",
    "\n",
    "# input_1 = Input(shape = (4,28,1), name='i_1')                       ###### testing dilated\n",
    "# conv1_1 = Convolution2D(128, 3, activation = 'relu',padding='valid',dilation_rate=(1, 4))(input_1)\n",
    "# bn1_1 = BatchNormalization()(conv1_1)\n",
    "# conv1_2 = Convolution2D(128, 2, activation = 'relu',padding='valid',dilation_rate=(1, 4))(bn1_1)\n",
    "# bn1_2 = BatchNormalization()(conv1_2)\n",
    "\n",
    "\n",
    "# flat_1 = Flatten()(bn1_2)\n",
    " \n",
    "# # kernel_regularizer=l2(0.01), bias_regularizer=l2(0.01)\n",
    "# #\n",
    "# #\n",
    "# #input_2\n",
    "\n",
    "# input_2 = Input(shape = (4,28,1), name='i_2')                  \n",
    "# conv2_1 = Convolution2D(128, 3, activation = 'relu',padding='valid',dilation_rate=(1, 4))(input_2)\n",
    "# bn2_1 = BatchNormalization()(conv2_1)\n",
    "# conv2_2 = Convolution2D(128, 2, activation = 'relu',padding='valid',dilation_rate=(1, 4))(bn2_1)\n",
    "# bn2_2 = BatchNormalization()(conv2_2)\n",
    "\n",
    "# flat_2 = Flatten()(bn2_1)\n",
    " \n",
    "# # concatenate\n",
    "# concat   = Concatenate()([flat_1, flat_2])\n",
    "# dense1   = Dense(256, activation = 'relu', kernel_regularizer=l2(0.0005))(concat)\n",
    "# #dropout1 = Dropout(0.5)(dense1)\n",
    "# dense2   = Dense(256, activation = 'relu')(dense1)\n",
    "\n",
    "# #dense3   = Dense(512, activation = 'relu')(dense2) #goes to regression\n",
    "\n",
    "\n",
    "# #dense4  = Dense(1024, activation = 'relu')(dense1)\n",
    "# dense4  = Dense(256, activation = 'relu')(concat)\n",
    "# #dropout2 = Dropout(0.2)(dense4) #goes to classification\n",
    "# #dense5  = Dense(128, activation = 'relu')(dropout2) \n",
    "\n",
    "\n",
    "# #regression output\n",
    "# output_1  = Dense(1, activation = 'linear', name='r')(dense2)\n",
    "\n",
    "\n",
    "# #classification output\n",
    "# output_2  = Dense(1, activation = 'sigmoid', name='c')(dense4)\n",
    " \n",
    "# # create model with two inputs\n",
    "# model = Model(inputs=[input_1,input_2], outputs=[output_1,output_2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "id": "tHMCMCFWfo0x"
   },
   "outputs": [],
   "source": [
    "# model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "n7eB3kFevLK5"
   },
   "outputs": [],
   "source": [
    "# model_name  = 'only_cnn_bn_cnn_bn_l2p0005_256r_256c' #l2=0.001, 0.0005"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "o2TaBMZjwkPf"
   },
   "outputs": [],
   "source": [
    "# plot_model(model,  to_file=model_name+'.png', show_shapes=True, show_layer_names=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "id": "c0-hmvzGu-7g"
   },
   "outputs": [],
   "source": [
    "# #metrics for classification\n",
    "# metrics_c = [\n",
    "#     #keras.metrics.FalseNegatives(name=\"fn\"),\n",
    "#     #keras.metrics.FalsePositives(name=\"fp\"),\n",
    "#     #keras.metrics.TrueNegatives(name=\"tn\"),\n",
    "#     #keras.metrics.TruePositives(name=\"tp\"),\n",
    "#     #keras.metrics.BinaryAccuracy(name='accuracy'),\n",
    "#     tensorflow.keras.metrics.Precision(name=\"precision\"),\n",
    "#     tensorflow.keras.metrics.Recall(name=\"recall\"),\n",
    "#     tensorflow.keras.metrics.AUC(name=\"auc_pr\",curve=\"PR\")\n",
    "# ]\n",
    "\n",
    "# #metric_r = tfa.metrics.r_square.RSquare()\n",
    "# metric_r = tfa.metrics.r_square.RSquare(dtype=tf.float32, y_shape=(1,))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "id": "ivi889-qvC7i"
   },
   "outputs": [],
   "source": [
    "# #model compile is one of the most important step\n",
    "# model.compile(optimizer=tensorflow.keras.optimizers.Adam(lr=0.0001),\n",
    "#               loss={'r': 'mse','c': 'binary_crossentropy'},\n",
    "#               loss_weights={'r': 0.001,'c': 1.0},\n",
    "#               metrics={'r':[metric_r],'c':metrics_c})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "7maNGgGzzBhE"
   },
   "outputs": [],
   "source": [
    "# #model compile is one of the most important step\n",
    "# model.compile(optimizer=keras.optimizers.Adam(lr=0.0001),\n",
    "#               loss={'r': 'mse','c': 'binary_crossentropy'},\n",
    "#               loss_weights={'r': 0.0002,'c': 1.0},\n",
    "#               metrics={'r':'mse','c':metrics_c})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "Q_vOw8QgvOBk"
   },
   "outputs": [],
   "source": [
    "\n",
    "# use of indexes to refer data later\n",
    "img_indexes = list(range(0,samples))\n",
    "\n",
    "\n",
    "#for conv2D\n",
    "off_dataset_train, off_dataset_test, target_dataset_train, target_dataset_test, Y_train, Y_test, freq_train, freq_test, train_index, test_index = train_test_split(off_dataset, target_dataset,Y,freq,img_indexes, test_size=0.1, random_state=10) \n",
    "\n",
    "\n",
    "\n",
    "\n",
    "#off_dataset_train, off_dataset_valid, target_dataset_train, target_dataset_valid, Y_train, Y_valid, freq_train, freq_valid, train_index, valid_index = train_test_split(off_dataset_train, target_dataset_train,Y_train,freq_train,train_index, test_size=1/9, random_state=2)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zNK3O0-vbDJp"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "a10y6pWav5nc",
    "outputId": "142db00d-c2de-49ff-ab35-65f163597262"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((138531, 4, 28, 1), (138531, 4, 28, 1), (138531, 1), (138531, 1))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#train summary\n",
    "off_dataset_train.shape,target_dataset_train.shape,freq_train.shape,Y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "3JD0bMd16gda"
   },
   "outputs": [],
   "source": [
    "#validation summary\n",
    "#off_dataset_valid.shape,target_dataset_valid.shape,freq_valid.shape,Y_valid.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "gbS1yoIhwBw2",
    "outputId": "ca064afe-7603-4c73-ca85-cb4a2952784c"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((15393, 4, 28, 1), (15393, 4, 28, 1), (15393, 1), (15393, 1))"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#test summary\n",
    "off_dataset_test.shape, target_dataset_test.shape, freq_test.shape,Y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "BP6fmKqdwEx6"
   },
   "outputs": [],
   "source": [
    "# early_stopping = EarlyStopping(monitor='val_loss', patience=10,restore_best_weights=True, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "id": "998xIEczwIhs"
   },
   "outputs": [],
   "source": [
    "# reduce_lr = ReduceLROnPlateau(monitor='val_loss', factor=0.5,patience=5, min_lr=0.00000001, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "id": "oF1s9Uq1UrJ2"
   },
   "outputs": [],
   "source": [
    "# #checkpoint_filepath = './checkpoints/'\n",
    "# checkpoint_filepath = 'weights-improvement-{epoch:02d}-{val_c_auc_pr:.4f}-{val_r_r_square:.4f}.hdf5'\n",
    "# model_checkpoint_callback = ModelCheckpoint(filepath=checkpoint_filepath,save_weights_only=False,monitor='val_c_auc_pr',mode='max',save_best_only=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uNApd4jIwLIj",
    "outputId": "68662032-08fb-4162-d0cf-e50e4e04f4bc"
   },
   "outputs": [],
   "source": [
    "# try:\n",
    "#   history=model.fit([off_dataset_train,target_dataset_train],[freq_train,Y_train], \n",
    "#                   batch_size=64, epochs=70, \n",
    "#                   validation_data=([off_dataset_test,target_dataset_test],[freq_test,Y_test]),\n",
    "#                   #class_weight={0: 0.525, 1: 10.704},\n",
    "#                   #validation_split=0.1,\n",
    "#                   class_weight={'c': {0: 0.525, 1: 10.704}},\n",
    "#                   #class_weight={0: 0.525, 1: 10.704},\n",
    "#                   #callbacks=[reduce_lr,model_checkpoint_callback],\n",
    "#                     verbose=1)\n",
    "# except KeyboardInterrupt:\n",
    "#   old_history = history\n",
    "#   model.save(model_name+'.hdf5')\n",
    "#   print('Model saved to drive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "dFqjjjHqZo_N"
   },
   "outputs": [],
   "source": [
    " #save model\n",
    "#model.save(model_name+'_epoch80'+'.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/xinfy/anaconda3/lib/python3.7/site-packages/tensorflow/python/compat/v2_compat.py:96: disable_resource_variables (from tensorflow.python.ops.variable_scope) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "non-resource variables are not supported in the long term\n"
     ]
    }
   ],
   "source": [
    "# #### some worked around for TF installed and TF needed\n",
    "# import tensorflow.compat.v1 as tf\n",
    "# tf.disable_v2_behavior() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "id": "3irnjbZyucMR"
   },
   "outputs": [],
   "source": [
    "###model = tf.keras.models.load_model('seed_07/weights-improvement-60-0.7588-0.7289.hdf5',compile=False)\n",
    "\n",
    "####model = tf.keras.models.load_model('seed_07/weights-improvement-50-0.7708-0.6933.hdf5',compile=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "yCwq1Ukh1SAP"
   },
   "outputs": [],
   "source": [
    "# #model compile for the load model\n",
    "# model.compile(optimizer=keras.optimizers.Adam(lr=0.0000001),\n",
    "#               loss={'r': 'mse','c': 'binary_crossentropy'},\n",
    "#               loss_weights={'r': 0.001,'c': 1.0},\n",
    "#               metrics={'r':[metric_r],'c':metrics_c})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "uDNq6_8Vqqyt"
   },
   "outputs": [],
   "source": [
    "# #training plots\n",
    "# fig = plt.figure(figsize=(10,15))\n",
    "# plt.subplot(3,2,1)\n",
    "# plt.plot(history.history['c_loss'])\n",
    "# plt.plot(history.history['val_c_loss'])\n",
    "# plt.title(\"Classification C losses\")\n",
    "# plt.ylabel(\"Classification C Losses\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Classification Loss\",\"Valid Classification Loss\"])\n",
    "# plt.subplot(3,2,2)\n",
    "# plt.plot(history.history['r_loss'])\n",
    "# plt.plot(history.history['val_r_loss'])\n",
    "# plt.title(\"Regression Loss\")\n",
    "# plt.ylabel(\"Regression Loss\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Regression Loss\",\"Validation Regression Loss\"])\n",
    "# plt.subplot(3,2,3)\n",
    "# plt.plot(history.history['r_r_square'])\n",
    "# plt.plot(history.history['val_r_r_square'])\n",
    "# plt.title(\"Regression r2\")\n",
    "# plt.ylabel(\"Regression r2\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Regression r2\",\"Valid Regression r2\"])\n",
    "# plt.subplot(3,2,4)\n",
    "# plt.plot(history.history['c_precision'])\n",
    "# plt.plot(history.history['val_c_precision'])\n",
    "# plt.title(\"Classification precision\")\n",
    "# plt.ylabel(\"Classification precision\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Classification precision\",\"Valid Classification precision\"])\n",
    "# plt.subplot(3,2,5)\n",
    "# plt.plot(history.history['c_recall'])\n",
    "# plt.plot(history.history['val_c_recall'])\n",
    "# plt.title(\"Classification recall\")\n",
    "# plt.ylabel(\"Classification recall\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Classification recall\",\"Valid Classification recall\"])\n",
    "# plt.subplot(3,2,6)\n",
    "# plt.plot(history.history['c_auc_pr'])\n",
    "# plt.plot(history.history['val_c_auc_pr'])\n",
    "# plt.title(\"Classification auc\")\n",
    "# plt.ylabel(\"Classification auc\")\n",
    "# plt.xlabel(\"Epoch\")\n",
    "# plt.legend([\"Training Classification auc\",\"Valid Classification auc\"])\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "DLku2Ro2as-f"
   },
   "outputs": [],
   "source": [
    "# ### MOdel LOAD\n",
    "\n",
    "model_loaded = 'june_rerun_models/model10_epoch40-0.8465.hdf5'  #right click copy path and paste here\n",
    "\n",
    "model = tf.keras.models.load_model(model_loaded,compile=False) #put here the model path and run all the tabs below"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "J_C1XdzpqulM"
   },
   "outputs": [],
   "source": [
    "# predict on test set\n",
    "yreg_pred_test, ycls_pred_test = model.predict([off_dataset_test,target_dataset_test])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "whXygdilqxup"
   },
   "outputs": [],
   "source": [
    "# predict on training set\n",
    "yreg_pred_train, ycls_pred_train = model.predict([off_dataset_train,target_dataset_train])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.style.use('classic')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "id": "_t1XjOKZq0Ct"
   },
   "outputs": [],
   "source": [
    "#plots\n",
    "from sklearn.metrics import precision_recall_curve\n",
    "from sklearn.metrics import average_precision_score\n",
    "from sklearn.metrics import roc_curve, auc\n",
    "\n",
    "fpr = dict()\n",
    "tpr = dict()\n",
    "roc_auc = dict()\n",
    "\n",
    "\n",
    "for i in range(1):\n",
    "    fpr[i], tpr[i], _ = roc_curve(Y_test[:, i], ycls_pred_test[:, i])\n",
    "    roc_auc[i] = auc(fpr[i], tpr[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "id": "BVs_UDshq6YO"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x560 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot of a ROC curve for a specific class\n",
    "fig, ax = plt.subplots(figsize=(8,7))\n",
    "ax.plot(fpr[0], tpr[0], alpha=1,linewidth=4.0, color='r')\n",
    "ax.plot([0, 1], [0, 1], 'k--')\n",
    "ax.set_ylim([0.0, 1.001])\n",
    "ax.set_xlim([0.0, 1.001])\n",
    "ax.set_yticks(np.arange(0, 1.001, 0.1))\n",
    "ax.set_xticks(np.arange(0, 1.001, 0.1))\n",
    "ax.set_xlabel('False Positive Rate', fontsize=14)\n",
    "ax.set_ylabel('True Positive Rate', fontsize=14)\n",
    "ax.set_title('ROC Curve', fontsize=14)\n",
    "ax.tick_params(axis=\"x\", labelsize=12)\n",
    "ax.tick_params(axis=\"y\", labelsize=12) \n",
    "ax.grid(linestyle='-.', linewidth=0.9)\n",
    "\n",
    "#ax.legend(fontsize=12)\n",
    "#####plt.savefig('seed10_june21_runs_roc_manuscript_best_0.9869.png', dpi=1200, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "id": "JNyZJeZtq9CN"
   },
   "outputs": [],
   "source": [
    "precision = dict()\n",
    "recall = dict()\n",
    "average_precision = dict()\n",
    "for i in range(1):\n",
    "    precision[i], recall[i], _ = precision_recall_curve(Y_test[:, i], ycls_pred_test[:, i])\n",
    "    average_precision[i] = average_precision_score(Y_test[:, i], ycls_pred_test[:, i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 466
    },
    "id": "4w2tCldMq9t4",
    "outputId": "2675d421-61a4-430f-d2ee-65e61107479a"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 0.1, 'aucPR=0.886')"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x640 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# A \"micro-average\": quantifying score on all classes jointly\n",
    "precision[\"micro\"], recall[\"micro\"], _ = precision_recall_curve(Y_test.ravel(), ycls_pred_test.ravel())\n",
    "average_precision[\"micro\"] = average_precision_score(Y_test, ycls_pred_test, average=\"micro\")\n",
    "#print('Average precision score , micro-averaged over all classes: {0:0.2f}'\n",
    "#    .format(average_precision[\"micro\"]))\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "ax.step(recall['micro'], precision['micro'], where='mid',linewidth=4.0,color='r')\n",
    "\n",
    "ax.set_xlabel('Recall', fontsize=14)\n",
    "ax.set_ylabel('Precision', fontsize=14)\n",
    "ax.plot([0, 1], [1, 0], 'k--')\n",
    "ax.set_ylim([0.0, 1.0001])\n",
    "ax.set_xlim([0.0, 1.0001])\n",
    "ax.set_yticks(np.arange(0, 1.0001, 0.1))\n",
    "ax.set_xticks(np.arange(0, 1.0001, 0.1))\n",
    "ax.set_title('Precision-Recall Curve', fontsize=14)\n",
    "ax.tick_params(axis=\"x\", labelsize=12)\n",
    "ax.tick_params(axis=\"y\", labelsize=12) \n",
    "ax.grid(linestyle='-.', linewidth=0.9)\n",
    "ax.text(0.5, 0.1, 'aucPR={0:0.3f}'.format(average_precision[\"micro\"]), bbox=dict(facecolor='red', alpha=0.5),fontsize=14)\n",
    "######plt.savefig('seed10_june21_runs_aupr_manuscript_best_0.886.png', dpi=1200, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "id": "LhV1_qJgrAGP"
   },
   "outputs": [],
   "source": [
    "#confusion matrix\n",
    "from sklearn.metrics import plot_confusion_matrix\n",
    "from sklearn import metrics as sklearn_metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "id": "k4GlGIg4rCw-"
   },
   "outputs": [],
   "source": [
    "y_label_pred = np.where(ycls_pred_test > 0.5, 1, 0)\n",
    "y_label_pred_train = np.where(ycls_pred_train > 0.5, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "id": "ib4Cqh_zrEsx"
   },
   "outputs": [],
   "source": [
    "# y_label_pred,y_label_pred.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [],
   "source": [
    "t56 = np.where(ycls_pred_test > 0.56, 1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {
    "id": "GJAL6fOerG4h"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[14516   167]\n",
      " [  122   588]]\n"
     ]
    }
   ],
   "source": [
    "print(sklearn_metrics.confusion_matrix(Y_test, t56))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "id": "oEuhjX0qrI7L"
   },
   "outputs": [],
   "source": [
    "# tn, fp, fn, tp = sklearn_metrics.confusion_matrix(Y_test, y_label_pred).ravel()\n",
    "# print(tn, fp, fn, tp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "6mPj5izIrLBW",
    "outputId": "3431e591-eecb-44b7-84f9-ff302a34d7f5"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0      0.992     0.988     0.990     14683\n",
      "           1      0.774     0.834     0.803       710\n",
      "\n",
      "    accuracy                          0.981     15393\n",
      "   macro avg      0.883     0.911     0.896     15393\n",
      "weighted avg      0.982     0.981     0.981     15393\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(sklearn_metrics.classification_report(Y_test, y_label_pred, digits=3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "#### MCC\n",
    "from sklearn.metrics import matthews_corrcoef"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test MCC: 0.7934069857804233\n"
     ]
    }
   ],
   "source": [
    "#print ('Train MCC: ', matthews_corrcoef(y_true, y_pred))\n",
    "print ('Test MCC:',   matthews_corrcoef(Y_test, y_label_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Cohen Kappa: 0.792798348627036\n"
     ]
    }
   ],
   "source": [
    "#### Cohen Kappa Score\n",
    "from sklearn.metrics import cohen_kappa_score\n",
    "print ('Test Cohen Kappa:',   cohen_kappa_score(Y_test, y_label_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "######## F1-MCC Score\n",
    "from mcc_f1 import mcc_f1_curve, plot_mcc_f1_curve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "mcc, f1, thresholds = mcc_f1_curve(Y_test, y_label_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0.89670349, 0.5       ]),\n",
       " array([0.80271186, 0.08818233]),\n",
       " array([1, 0]))"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mcc, f1, thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "id": "13-ymqQyrNM8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((138531, 1), (138531, 1))"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#now regression\n",
    "freq_train.shape, yreg_pred_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "id": "82ZjTc0QrPLI"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((15393, 1), (15393, 1))"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "freq_test.shape, yreg_pred_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "id": "NMbxzcSgrTHo"
   },
   "outputs": [],
   "source": [
    "\n",
    "## R2 calculation\n",
    "from sklearn.metrics import r2_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "id": "93fOoupwrT7F"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training_r2 0.8991552270959285\n",
      "test_r2 0.6723188182096322\n"
     ]
    }
   ],
   "source": [
    "print('training_r2', r2_score(freq_train, yreg_pred_train))\n",
    "print('test_r2', r2_score(freq_test, yreg_pred_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "id": "3Ek33SuNas-p"
   },
   "outputs": [],
   "source": [
    "# #### calculation using TF addon function\n",
    "# metric_reg_train = tfa.metrics.r_square.RSquare(dtype=tf.float32, y_shape=(1,))\n",
    "# metric_reg_test = tfa.metrics.r_square.RSquare(dtype=tf.float32, y_shape=(1,))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "id": "tmZ1jHbNas-p",
    "outputId": "d8647ae6-588e-41ed-96f8-9dd2ec00433d"
   },
   "outputs": [],
   "source": [
    "# metric_reg_train.update_state(freq_train, yreg_pred_train)\n",
    "# result_trainr2 = metric_reg_train.result().numpy()\n",
    "# print('train R2',result_trainr2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "id": "pDZxLL_Was-q",
    "outputId": "a575f9e0-dce5-434b-c871-ce971992aa0b"
   },
   "outputs": [],
   "source": [
    "# metric_reg_test.update_state(freq_test, yreg_pred_test)\n",
    "# result_testr2 = metric_reg_test.result().numpy()\n",
    "# print('test R2',result_testr2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "id": "-R4zp-0XrV6K"
   },
   "outputs": [],
   "source": [
    "# #fig, ax = plt.subplots()\n",
    "# fig, ax = plt.subplots(figsize=(8,8))\n",
    "# ax.scatter(freq_train, yreg_pred_train,s=5)\n",
    "# ax.plot([freq.min(), freq.max()], [freq.min(), freq.max()], 'k--', lw=1)\n",
    "# ax.set_xlabel('Actual')\n",
    "# ax.set_ylabel('Predicted')\n",
    "# ax.set_ylim([0.0, 1001])\n",
    "# ax.set_xlim([0.0, 1001])\n",
    "# ax.set_yticks(np.arange(0, 1001, 100))\n",
    "# ax.set_xticks(np.arange(0, 1001, 100))\n",
    "# ax.set_title('Training Set R2: ' + str(r2_score(freq_train, yreg_pred_train)))\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "id": "K_ZhQMnEracQ"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x640 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#fig, ax = plt.subplots()\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "ax.scatter(freq_test, yreg_pred_test,s=6,alpha=0.5)\n",
    "ax.plot([freq.min(), freq.max()], [freq.min(), freq.max()], 'r--', lw=2)\n",
    "ax.set_xlabel('Actual')\n",
    "ax.set_ylabel('Predicted')\n",
    "# ax.set_yscale('log')\n",
    "# ax.set_xscale('log')\n",
    "ax.set_ylim([0.0, 1001])\n",
    "ax.set_xlim([0.0, 1001])\n",
    "ax.set_yticks(np.arange(0, 1001, 100))\n",
    "ax.set_xticks(np.arange(0, 1001, 100))\n",
    "#ax.set_title('Test Set R2: ' + str(r2_score(freq_test, yreg_pred_test)))\n",
    "plt.show()\n",
    "#####plt.savefig('seed10_june21_runs_test_r2_0p6723.png', dpi=1200, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "689fQSvzas-u"
   },
   "outputs": [],
   "source": [
    "# max(yreg_pred_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "id": "UNN4E4D2as-v"
   },
   "outputs": [],
   "source": [
    "########### how many of label 1 are having a regression value.\n",
    "test_zero_label = y_label_pred[y_label_pred == 0]\n",
    "test_one_label = y_label_pred[y_label_pred == 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "id": "R_mDunFtas-w"
   },
   "outputs": [],
   "source": [
    "# test_zero_label.shape, test_one_label.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {
    "id": "pLtsT_8Fas-x"
   },
   "outputs": [],
   "source": [
    "###### regression values according to the label\n",
    "test_reg_zero_label = yreg_pred_test[y_label_pred == 0]\n",
    "test_reg_one_label = yreg_pred_test[y_label_pred == 1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "hWEUI4LRas-y"
   },
   "outputs": [],
   "source": [
    "# test_reg_zero_label.shape, test_reg_one_label.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EfkZAEmcas-z"
   },
   "outputs": [],
   "source": [
    "\n",
    "# max(test_reg_zero_label),min(test_reg_zero_label )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "zoxQVlRmas-0"
   },
   "outputs": [],
   "source": [
    "\n",
    "# max(test_reg_one_label),min(test_reg_one_label )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "kDOlR3qIas-0"
   },
   "outputs": [],
   "source": [
    "\n",
    "# np.mean(test_reg_zero_label),np.mean(test_reg_one_label )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "OBm6ojzEas-1"
   },
   "outputs": [],
   "source": [
    "# np.median(test_reg_zero_label),np.median(test_reg_one_label )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "nFYjNKeias-1"
   },
   "outputs": [],
   "source": [
    "# import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "y0etpbgVas-1"
   },
   "outputs": [],
   "source": [
    "# plt.hist(test_reg_zero_label, bins='scott');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "2c_2Z_uyas-2"
   },
   "outputs": [],
   "source": [
    "# plt.hist(test_reg_one_label, bins='scott');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "id": "OKIoxw0Qas-2"
   },
   "outputs": [],
   "source": [
    "### class wise R2\n",
    "train_reg_zero_label = yreg_pred_train[y_label_pred_train ==0]\n",
    "train_reg_one_label = yreg_pred_train[y_label_pred_train ==1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "id": "N6YazQdjas-3"
   },
   "outputs": [],
   "source": [
    "# train_reg_zero_label.shape, train_reg_one_label.shape, "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "id": "JdoI9kVEas-3"
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import r2_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {
    "id": "DaXmn0qvas-3"
   },
   "outputs": [],
   "source": [
    "# freq_train[y_label_pred_train  ==1].shape, yreg_pred_train[y_label_pred_train  ==1].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "id": "5ySpEDgxas-4"
   },
   "outputs": [],
   "source": [
    "# freq_test[y_label_pred  ==1].shape, yreg_pred_test[y_label_pred  ==1].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "id": "IabnNunDas-4",
    "outputId": "3b4e787f-03f7-4e6b-8f8e-260891c8bf65"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "one_label_training_r2 0.8709026616857497\n",
      "one_label_test_r2 0.616442670581975\n"
     ]
    }
   ],
   "source": [
    "print('one_label_training_r2', r2_score(freq_train[y_label_pred_train ==1], train_reg_one_label))\n",
    "print('one_label_test_r2', r2_score(freq_test[y_label_pred ==1], test_reg_one_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "7SFmC_-vas-5"
   },
   "outputs": [],
   "source": [
    "# fig, ax = plt.subplots()\n",
    "# ax.scatter(freq_train[y_label_pred_train ==1], train_reg_one_label,s=5)\n",
    "# ax.plot([freq.min(), freq.max()], [freq.min(), freq.max()], 'k--', lw=1)\n",
    "# ax.set_xlabel('Actual')\n",
    "# ax.set_ylabel('Predicted')\n",
    "# ax.set_title('Training Set R2: ' + str(r2_score(freq_train[y_label_pred_train ==1], train_reg_one_label)))\n",
    "# plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "id": "7uYBNCMfas-5"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x640 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#fig, ax = plt.subplots()\n",
    "fig, ax = plt.subplots(figsize=(8,8))\n",
    "ax.scatter(freq_test[y_label_pred ==1], test_reg_one_label,s=6,alpha=0.5)\n",
    "ax.plot([freq.min(), freq.max()], [freq.min(), freq.max()], 'r--', lw=2)\n",
    "ax.set_xlabel('Actual')\n",
    "ax.set_ylabel('Predicted')\n",
    "ax.set_ylim([0.0, 501])\n",
    "ax.set_xlim([0.0, 501])\n",
    "ax.set_yticks(np.arange(0, 501, 100))\n",
    "ax.set_xticks(np.arange(0, 501, 100))\n",
    "ax.set_title('Test Set R2: ' + str(r2_score(freq_test[y_label_pred ==1], test_reg_one_label)))\n",
    "plt.show()\n",
    "###plt.savefig('seed10_june21_runs_test_r2_0p6723.png', dpi=1200, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "id": "xxKJgUkHas-6"
   },
   "outputs": [],
   "source": [
    "### explained variance score\n",
    "from sklearn.metrics import explained_variance_score\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "id": "efnigcrtas-7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "exp var score:  train 0.919003297111685\n",
      "exp var score:  test 0.6474752587708971\n"
     ]
    }
   ],
   "source": [
    "print('exp var score:  train', explained_variance_score(freq_train[y_label_pred_train ==1], train_reg_one_label))\n",
    "print('exp var score:  test', explained_variance_score(freq_test[y_label_pred ==1], test_reg_one_label))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "id": "Wam3VKqAas-8"
   },
   "outputs": [],
   "source": [
    "##################### F1-MCC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "id": "bVYUxOBhas-9"
   },
   "outputs": [],
   "source": [
    "\n",
    "def mcc_f1(ground_truths, predicted_values):\n",
    "    \n",
    "    \"\"\"\n",
    "    The function mcc_f1 calculates MCC and F1 scores based on varying thresholds identifed by the PRC function\n",
    "    based on inputs -  ground truth vaules and predicted values.\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    ground_truths, predicted_values : numpy.ndarray;\n",
    "                        Input arrays; groundtruth values (True values) and predicted values.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    mcc, f1, thresholds: numpy.ndarray;\n",
    "                           Output arrays; Unit normalized MCC scores and F1 score values \n",
    "                           for every threshold.           \n",
    "    \"\"\"\n",
    "    \n",
    "#     mcc = []   \n",
    "#     f1 = []\n",
    "   \n",
    "    \n",
    "    # Generating thresholds with the PRC function\n",
    "    precision, recall, thresholds = metrics.precision_recall_curve(\n",
    "                                    ground_truths, predicted_values)\n",
    "    \n",
    "    mcc = np.zeros((thresholds.shape))\n",
    "    f1 = np.zeros((thresholds.shape))\n",
    "    # For every value of the thresholds (cutoffs), calculate MCC and F1 scores :\n",
    "    for index,T in enumerate(thresholds):\n",
    "        predicted_values_after_cutoff = predicted_values > T  # Positive classes are determined for every predicted value that exceeds the threshold. \n",
    "    \n",
    "        ##mcc.append\n",
    "        mcc[index]=(metrics.matthews_corrcoef(ground_truths, predicted_values_after_cutoff)+1)*0.5 # get normalised MCC: change the range of MCC from [-1, 1] to [0, 1] \n",
    "        \n",
    "        ###f1.append(\n",
    "        f1[index]=metrics.f1_score(ground_truths, predicted_values_after_cutoff)\n",
    "        \n",
    "    #mcc = np.array(mcc)\n",
    "    #f1 = np.array(f1)  \n",
    "    \n",
    "    # Resulting values are an array of MCC & F1 scores across different thresholds (cutoffs)  \n",
    "    return mcc,f1, thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tqdm import tqdm\n",
    "\n",
    "def MCC_F1_Metric(mcc,f1,thresholds, W_num_of_subranges = 100):\n",
    "    \n",
    "    \"\"\"\n",
    "    MCC_F1_Metric function computes the MCC-F1 metric i.e. the average distances between the predcition scores to the point of Perfect Performace (1,1) \n",
    "    and also results in the best threshold value\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    \n",
    "    mcc,f1, thresholds : numpy.ndarray;\n",
    "                        MCC and F1 scores across varying thresholds\n",
    "    \n",
    "    W_num_of_subranges : int, optional;\n",
    "        Number of subranges- default = 100; \n",
    "        larger values of 𝑊 will cause the MCC-𝐹1 metric to capture the performance of a classifier more accurately.\n",
    "    \n",
    "    Returns\n",
    "    -------\n",
    "    MCC_F1_Met : float;\n",
    "                The ratio of the average distance of the MCC-F1 score to the Point of Perfect Performace (1,1)\n",
    "                A metric to compare classifier performance.\n",
    "    \n",
    "    Best_threshold : int;\n",
    "                Index of the best prediction score threshold (𝑇), \n",
    "                Index of the point on the MCC-𝐹1 curve closest to the point of perfect performance (1,1).\n",
    "    \n",
    "    \"\"\"\n",
    "    \n",
    "     \n",
    "    subrange_intervals = np.linspace(np.min(mcc), np.max(mcc),W_num_of_subranges) # breaking into subranges\n",
    "    \n",
    "    # Computing MCC values per subrange\n",
    "    \n",
    "    values_per_subrange = np.zeros_like(subrange_intervals) # 'n' is the number of points per subrange\n",
    "    for i in range(W_num_of_subranges-1):\n",
    "        for j in mcc:\n",
    "            if j >= subrange_intervals[i] and j < subrange_intervals[i+1]:\n",
    "                values_per_subrange[i] = values_per_subrange[i] + 1\n",
    "\n",
    "    # Calculating the distances between points in a subrange to the Point of perfect performance (1,1)            \n",
    "    Distance_of_points_within_subrange = np.sqrt(((mcc-1)**2)+((f1-1)**2)) # calculating the Euclidean distance 𝐷𝑖            \n",
    "    sum_of_distance_within_subrange = np.zeros_like(subrange_intervals)\n",
    "    index = -1\n",
    "    for value in mcc:\n",
    "        index += 1\n",
    "        for i in range(W_num_of_subranges-1):\n",
    "                 if value >= subrange_intervals[i] and value < subrange_intervals[i+1]:\n",
    "                        sum_of_distance_within_subrange[i] = sum_of_distance_within_subrange[i] + Distance_of_points_within_subrange[index]\n",
    "\n",
    "    # Mean Distance across subranges\n",
    "    mean_Distance_per_subrange = np.array(sum_of_distance_within_subrange/values_per_subrange)\n",
    "    total_number_of_subranges = 0  \n",
    "    total_sum_of_mean_distances = 0\n",
    "    for i in mean_Distance_per_subrange:\n",
    "        if not np.isnan(i):\n",
    "            total_number_of_subranges += 1 \n",
    "            total_sum_of_mean_distances += i # addition of all the means across subranges that have atleast 1 MCC value.\n",
    "\n",
    "    grand_mean_distance = total_sum_of_mean_distances/total_number_of_subranges # P = total number of subranges that have atleast 1 MCC value\n",
    "   \n",
    "    \"\"\" \n",
    "    Compare the grand average distance to √2 (The distance between the point of worst performance (0,0) and \n",
    "    the point of perfect performance (1,1) is √2).That is the maximum possible distance between a point on the MCC-𝐹1 curve\n",
    "    The ratio between the grand avgerage distance and √2 is taken.\n",
    "    This ratio ranges between 0 and 1 (worst value = 0; best value = 1). To get the MCC-𝐹1 score, we subtract this ratio from 1\n",
    "    \"\"\"\n",
    "\n",
    "    MCC_F1_Met = 1 - (grand_mean_distance /np.sqrt(2))\n",
    "  \n",
    "    \n",
    "    # Finding the best threshold 𝑇 the point on the MCC-𝐹1 curve closest to the point of perfect performance (1,1).\n",
    "   \n",
    "    result = np.where(Distance_of_points_within_subrange == Distance_of_points_within_subrange.min())\n",
    "    Best_threshold = result[0][0]\n",
    "    \n",
    "    print(\"MCC-F1 metric = {0:.2f}\".format(MCC_F1_Met), \"Best Threshold = {0:.2f}\".format(thresholds[Best_threshold]))\n",
    "    \n",
    "    return MCC_F1_Met,Best_threshold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generating a plot (the overall canvas);\n",
    "axis = None\n",
    "def axis_plotting(total_plots): \n",
    "    global axis\n",
    "    if axis == None:\n",
    "        fig = plt.figure(figsize=(7,7))\n",
    "        axis = []        \n",
    "        for i in range(1,total_plots+1):\n",
    "            ax = fig.add_subplot(total_plots, 1, i)\n",
    "            axis.append(ax)\n",
    "            \n",
    "    \n",
    "def plotting(mcc, f1, thresholds,MCC_F1_Met,Best_threshold,classifer_name = \"\",total_plots = 1,ax = 0):\n",
    "    \n",
    "        \"\"\"\n",
    "        Plotting function - fuction to plot the MCC and F1 scores (across varying Thresholds), and highlight the the MCC_F1 metric with the best threshold\n",
    "        For multiple plots, 'total_plots' should be > 1 AND 'ax' (< total_plots) starting from  1\n",
    "    \n",
    "        Parameters\n",
    "        ----------\n",
    "        mcc,f1 : numpy.ndarray;\n",
    "            MCC and F1 scores across varying thresholds\n",
    "    \n",
    "        MCC_F1_Met : float;\n",
    "                The ratio of the average distance of the MCC-F1 score to the Point of Perfect Performace (1,1)\n",
    "                A metric to compare classifier performance.\n",
    "    \n",
    "        Best_threshold : int;\n",
    "                Index of the best prediction score threshold (𝑇), \n",
    "                Index of the point on the MCC-𝐹1 curve closest to the point of perfect performance (1,1).\n",
    "                \n",
    "        classifer_name : str, optional\n",
    "                    default = \"\"\n",
    "    \n",
    "        total_plots: int, optional\n",
    "                    Number of sub-plots to generate; default = 1\n",
    "    \n",
    "        ax : int, optional\n",
    "            Index of subplot ([1,2,3...]); default = None\n",
    "    \n",
    "                            \n",
    "        Returns\n",
    "        -------\n",
    "        \n",
    "        MCC-F1 curve: \n",
    "            for multiple sublplots, adjust total_plots and ax input value.\n",
    "                        \n",
    "        \"\"\"\n",
    "        \n",
    "        global axis\n",
    "        if total_plots > 1 and ax < total_plots: # if more than 1 total plot is mentioned;\n",
    "            axis_plotting(total_plots) #calling function to generate plot;\n",
    "            if ax == 0:\n",
    "                subplot = plt.gca() # plot on current axis value, \n",
    "            else:\n",
    "                subplot = axis[ax-1] # plotting additional graphs based on index of the axis.\n",
    "        elif ax > total_plots:\n",
    "            print(\"Index inconsistent with total_plots\")\n",
    "        else:\n",
    "            axis = None\n",
    "            subplot = plt.gca() # plot on base plot if no additional subplots are needed\n",
    "      \n",
    "        # Labelling the plot;\n",
    "        subplot.plot(f1,mcc,linewidth=4,color='r')\n",
    "        subplot.annotate(classifer_name, #+ \"MCC-F1-Metric - {0:.2f}\".format(MCC_F1_Met), \n",
    "                 (f1[Best_threshold],mcc[Best_threshold]), \n",
    "                 textcoords=\"offset points\", \n",
    "                 xytext=(1,1),\n",
    "                 ha='left') \n",
    "        \n",
    "        subplot.scatter(f1[Best_threshold],mcc[Best_threshold],color='g',linewidth=4)\n",
    "        subplot.set(xlabel='F1 score',\n",
    "               ylabel='unit−normalized MCC',title='Test Set MCC-F1 curve')\n",
    "        \n",
    "        # Inserting Random line\n",
    "        subplot.axhline(0.5,linewidth=2, color='blue',linestyle='--')\n",
    "        subplot.annotate('Random prediction line',xy =(0,0.5),xytext=(250,-15),\n",
    "                    color='b', xycoords = subplot.get_yaxis_transform(), \n",
    "                    textcoords=\"offset points\", va=\"bottom\")\n",
    "\n",
    "\n",
    "        # Displaying and labelling points of performances.\n",
    "\n",
    "        subplot.scatter(1,1,color = 'black',linewidth=4);subplot.scatter(0,0,color = 'black',linewidth=4)\n",
    "        subplot.annotate('Point of Perfect performance',(1,1),\n",
    "                    textcoords=\"offset points\",xytext=(-9,-2),ha='right',size=14)\n",
    "        subplot.annotate('Point of Worst performance',(0,0)\n",
    "                    ,textcoords=\"offset points\",xytext=(9,2),ha='left',size=14)\n",
    "        \n",
    "        subplot.set_xticks(np.arange(0, 1.1, 0.1)); subplot.set_yticks(np.arange(0, 1.1, 0.1))\n",
    "        subplot.set_xlim(-0.05,1.05); subplot.set_ylim(-0.05,1.05)\n",
    "        subplot.grid(alpha = 0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MCC-F1 metric = 0.55 Best Threshold = 0.94\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/xinfy/anaconda3/lib/python3.7/site-packages/ipykernel_launcher.py:53: RuntimeWarning: invalid value encountered in true_divide\n"
     ]
    }
   ],
   "source": [
    "####mcc,f1,thresholds = mcc_f1(Y_train, y_label_pred_train) ###train\n",
    "#mcc,f1,thresholds = mcc_f1(Y_test, ycls_pred_test)\n",
    "# MCC_F1_Met,T_index = MCC_F1_Metric(mcc,f1,thresholds,W_num_of_subranges=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.55782478, 0.55784697, 0.55785807, ..., 0.62980102, 0.62448453,\n",
       "       0.5       ])"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mcc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.11135543, 0.11137292, 0.11138167, ..., 0.13157895, 0.12169312,\n",
       "       0.        ])"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([6.255587e-19, 6.266191e-19, 6.285583e-19, ..., 9.999998e-01,\n",
       "       9.999999e-01, 1.000000e+00], dtype=float32)"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "thresholds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5558215445005983"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "MCC_F1_Met"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10131"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "T_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAi0AAAHHCAYAAABz3mgLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAMTQAADE0B0s6tTgAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOydZ1gU1xqA3wVpgmAFxK7EgrFiL1jB2LsJmhg1McUkqGiuMTG2xBTLtRtN9NpiSexdwa7RxBK9sRAT7AoKKopY6PfHXDYsdVeW2R393ufh0Z05c+bdswN8nPId3b59+1IRBEEQBEGwcmwsLSAIgiAIgmAMErQIgiAIgqAJJGgRBEEQBEETSNAiCIIgCIImkKBFEARBEARNIEGLIAiCIAiaQIIWQRAEQRA0QQFLCwjC80CrVq1yLePh4cHq1avNds8DBw5w584devbsaVT5GzdusGzZMs6cOcPdu3dxcXHBy8uL+vXr8+abb5p075s3bxIaGkr79u3x8PDItfzEiRPZt28fXl5erFixItP5H374gZUrV2JjY8OePXsMziUkJLBp0yb27t3LtWvXSExMxN3dHV9fX3r27EnZsmX1ZZOTk9m1axe7du3i0qVLPH36lGLFilGzZk26d+9OtWrVcvQcMmQIYWFhmY6n9w4PD2fjxo1cuHCBy5cvY2NjQ0hISK5tIAhC3pGgRRDMwNy5cw1ef/7551SqVIkBAwboj9nZ2Zn1ngcOHODChQtGBS03btzg3XffpVSpUgwYMAAPDw/u3bvH+fPnOXjw4DMFLUuXLsXX19eooAXAycmJiIgIzpw5Q40aNfTHU1JS2L17NwULFuTp06cG18TFxfHxxx9z5coVunbtyoABA3BwcODatWvs3LmTzz//nKVLlwKQlJTEmDFjOHnyJO3bt6dPnz64uLgQGRnJnj17GDZsGLt27crVs2rVqnz00UcGx+zt7fX/P3/+PCdOnKBy5coUKFCAixcvGvX+BUHIOxK0CIIZ8PHxMXhtZ2eHm5tbpuOWYuvWrSQmJjJ9+nScnZ31x1u3bk1KSooqDkWKFKFSpUqEhoYaBC2nTp3izp07tGnTJlMvy/Tp07l06RKzZs2iSpUq+uO1a9emS5cuHD58WH9s8eLFHDt2jEmTJtG4cWP98Vq1avHKK68YlM2JggUL5vi5derUiS5dugDw3XffWVXQEh8fj4ODg6U1BCHfkKBFECzAyZMnWb58ORcuXACUX6zvv/8+5cqV05c5evQoy5cv5+rVq6SmplKiRAnatWtH37599cMt8M/QVJkyZVi2bFmW93v48CFOTk44OTllOmdjYzi1LSkpiZUrVxIaGsrt27dxc3Ojbdu2DBo0CDs7O44dO8aoUaMACAoK0l83d+7cXIO0gIAAfvjhBz766CN9z1NISAh169alePHiBmUjIyPZt28fffr0MQhY0tOsWTMAnjx5wvr16/Hz8zMIWLIqm1cytpep3Lt3j8WLF/Prr7/y4MEDihQpQp06dfj444+xtbVl4sSJxMXFMXnyZIPrhgwZgre3N8HBwQBs3ryZ6dOnM3v2bFatWsXp06epXr061atXZ+XKlaxfv94gQAXo168fPj4+fPbZZwA8fvyYJUuWcPDgQe7evYu7uzudOnXitddeQ6fT5el9CkJ+IEGLIKjMwYMHmTBhAs2aNWPMmDGkpKSwcuVKhg4dyqJFiyhWrBjXrl1j7NixtGnThgEDBmBra8vNmzeJiooC4K233iI2NpZr164xfvx4wHAIIyNVq1Zl+/btTJo0iR49elClShUKFMj623/ChAmcOHGCfv36Ua1aNS5fvszixYuJjo5mzJgx+Pj48MEHHzB37lyCg4OpVKkSABUqVMj1vbdq1Yo5c+Zw9OhR/Pz8ePLkCYcOHWL48OFcvnzZoOypU6dITU3NNghJT1hYGE+fPjWqbG6kpqaSnJxscMzGxsYsv8QfPHjAkCFDSEhI4PXXX6d8+fLcu3ePw4cPk5KSgq2trcl1fvnllwQEBNCjRw8ASpYsyZIlSzhw4AAdOnTQlztz5gwREREMHToUgMTEREaMGMGtW7d44403KFeuHGfPnuU///kPjx494u23387z+xUEcyNBiyCoSEpKCnPmzKF+/fpMmDBBf7xWrVr07duXtWvX8u6773LhwgWSk5MJDg7WByO+vr768qVKlcLV1RU7OzujhqA6dOjA+fPn2blzJ3v37tVf5+fnR5cuXfQBzIkTJzh8+DBjx47V9+D4+vri5OTE1KlT9b/c0ia/li9f3qQhMBcXFxo3bkxoaCh+fn4cOnQIgObNm2cKWqKjowGMmjOTFswZO78mJ06dOkXbtm0Njn366af4+/vnue7Vq1dz584dFi1aZNCrlvF+puDv78+gQYMMjr388suEhoYaBC2hoaEUK1ZM/xzt2rWLCxcuMGfOHP1n6OvrS3JyMqtXr+bVV1+lUKFCz+wlCPmBLHkWBBW5cuUK0dHRtG3bluTkZP2Xs7MzVatW5cyZMwC89NJL6HQ6JkyYwOHDh3nw4EGe7mtra8uoUaP48ccf+fDDD2nevDnXrl1j9uzZfPTRRyQmJgJw7NgxHB0dadq0qYFf/fr1Afjjjz/y1gAoQ0S//fYbsbGxhISE4Ofnh6OjY57rNZa0npT0X+mpWrUq8+fPN/hq2LChWe594sQJatSoYRCw5JWmTZtmOubv789///tfbt++DSi9Kvv376dNmzb63pxjx45RunRpqlSpYtAW9erVIzExkT///NNsjoJgLqSnRRBUJCYmBoBJkyYxadKkTOfLlCkDKD0Y33zzDatXr2bixIkkJydTrVo13n33XYNJrKZSqlQpevbsSc+ePUlOTmb+/PmsXbuWkJAQOnbsyP3793n69Cnt2rXL8vrY2NhnvncaDRs2xNnZmTVr1nDq1CmmTJmSZbkSJUoASi+Kp6dnjnW6u7vry+bGli1bmD59uv51xrlABQsWzHYOTV558OAB5cuXN2udRYsWzXQsbRhu9+7d9OvXj6NHj/Lw4UOD3qL79+9z/fr1bHt5zPFZC4K5kaBFEFTEzc0NgPfff59atWplOp9+Xkr9+vWpX78+CQkJ/PHHHyxcuJDRo0ezevVqXFxc8uxia2urH5K6cuUKAK6urhQsWJB///vfWV6TcbLss963devWrFy5kuLFi1O7du0sy9WpUwedTseRI0eoWbNmjnVWq1YNR0dHjhw5QkBAQI5l/fz8DIKSnOYCmRs3Nzfu3LmTYxl7e3t9z1d6sutty2quTaFChWjUqBGhoaH069eP3bt3U7FiRby9vfVlXF1dKVOmjH5SbkZKliyZo6cgWAIJWgRBRSpUqEDx4sW5du0affr0Meoae3t76tWrx8OHD5k4cSJRUVG4uLhgZ2dHQkKCUXXcu3cvy7/Ir127BkCxYsUAaNCgAevWrSMhISHHHp20lT/x8fFG3T8jHTt2JCIigsaNG2e7GqdkyZK0atWKjRs30rZtW4NfuGkcPnyYZs2a4eTkRI8ePVi1ahW//fZblsM5aWULFy5M4cKFn8k7r9SrV481a9Zw9erVbIeIPDw8OH78OMnJyfqhnMuXL3Pr1i2T7hUQEMDnn3/OyZMn+fXXX3nrrbcMzjdo0IBjx47h4uJCqVKlnu0NCYLKSNAiCCpia2tLUFAQ48eP5+nTp/j5+eHq6sq9e/c4e/Ys5cqVo2vXrqxbt44LFy5Qv359SpQowYMHD1ixYgUeHh4Gk2BDQkLYtm0blSpVwsHBIdsVPIsWLeLy5cu0atWKihUrotPpuHjxIqtWraJIkSL63okGDRrQvHlzxowZQ+/evfU9EpGRkfz2228MGzaMEiVKULZsWWxsbNi2bRuOjo7Y2dlRtmzZLJdUZ0XFihX5+uuvcy03fPhwbt68SVBQEF27dqVOnTrY29tz48YNdu7cycOHD/VLmQcMGMDFixcZM2YMHTp0oFGjRjg7O+uXTp8+fZqdO3ca5ZcTjx8/5vjx44CStC81NZUDBw4ASubcl156KdtrX3vtNfbt28fw4cN54403KF++PDExMRw+fJjRo0djZ2dHq1at+PHHH/n2229p164dUVFRrFq1St9LZywNGzbE1dWVb775huTkZNq0aWNw/pVXXmHXrl0EBwfTu3dvKlSoQGJiIjdv3uSXX37JtrdNECyJBC2CoDLNmzdn+vTprFixgilTphAfH0+xYsWoXr06VatWBZSJuCdPnuT777/nwYMHuLq6UqtWLQYNGqRf6dOlSxcuXLjAd999x6NHj3LM05L2C2rbtm3cuXNHf89GjRrRv39/g16YcePGsW7dOnbu3MmyZcuwt7enZMmSNGjQQD8sVaxYMT788EN++uknhg4dSkpKilF5WkzFxcWFWbNmsXHjRvbu3cvGjRtJTk7G3d2devXq0atXL31ZOzs7Jk2axM6dO9m1axd79uzRv8/atWsza9YsszjduXNHv8w8jbTXnTt31udRyQo3Nzfmzp3LokWLWL58OXFxcRQpUoS6devqe5zKly/PmDFjWLp0KQcOHKBcuXIMHz6cH374wSRPOzs7WrZsyebNm6lXr16moT17e3umTZvGihUr2LRpE7dv38bJyQkvLy+zLB0XhPxAt2/fvlRLSwiCIAiCIOSGLHkWBEEQBEETSNAiCIIgCIImkKBFEARBEARN8NxNxE1JSeHu3bs4OTnJhl+CIAiCoBFSU1N58uQJxYoVyzYVwnMXtNy9e9fo/BeCIAiCIFgXP//8sz4jdkaeu6AlLU/E9evXcXV1NXv9Q4YMYd68eWavNz8RZ3UQZ3UQZ3XQojNo01ucFWJjYylTpkyO+Z6eu6AlbUjI1dU1X4KWwoUL50u9+Yk4q4M4q4M4q4MWnUGb3uJsSE5TO2QiromouU+JuRBndRBndRBnddCiM2jTW5yNR4IWE8lu91trRpzVQZzVQZzVQYvOoE1vcTYeCVpMpEGDBpZWMBlxVgdxVgdxVgctOoM2vcXZeCRoMZHY2FhLK5iMOKuDOKuDOKuDFp1Bm97ibDzP3d5Djx49olOnTvpN5gRBEARBsH5iY2Nxc3Nj69atODs7Z1lGeloEQbAqxo8fj06n038VK1aMZs2asW3btmeqy9HR0eTrTp8+zfjx44mLizP52pwYNWoUXl5e2NjY0K1btyzLXLlyxeD9Ozo6Uq1aNcaOHcvjx49V93lWxo8fz+HDh81aZ165e/cuPXv2pGjRouh0OmbMmGFpJcFEnrslz4IgaB97e3sOHDgAwO3bt5k2bRqdOnUiJCQEf39/o+t5++236dChg8n3P336NBMmTOC9997DxcXF5OuzYvv27UyePJlp06bRuHFjihUrlmP5L774grZt2/LkyRNCQkKYNGkSFy9eZMWKFRbxMZUJEybg6OhIs2bNzFpvXvj222/ZvXs3S5YsoWTJkpQvX97SSoKJSNBiImFhYVSrVs3SGiYhzuogzuZDp9PRqFEj/etWrVpRtmxZZs2aRenSpY12Ll26NKVLl84vTaMJCwsjLCwMgGHDhmWbojw93t7e+jZo1aoVt2/fZvHixcyaNStPAcaTJ09wcnLK1cdan43cyMo7/XuuVasW3bt3z/N9EhMTsbGxwdbWNs91abGtLeUsw0MmUrFiRUsrmIw4q4M45x+urq5UrlyZS5cu6Z1v3LhBYGAgRYsWxcnJiUaNGhEaGmpwXcbhof3796PT6di1axcDBgzA1dWVUqVKMXr0aJKTkwFYsmQJAwcOBKBkyZLodLpc/yI/f/48nTp1wtXVFWdnZ9q2bcvJkyf15999911GjhwJgK2tLTqdjv3795vUBvXq1QPg0qVL+mMrVqygTp06ODo64uHhwUcffcSTJ08yvd+tW7fSt29fChcuTIcOHWjZsmW2PrGxsQQFBeHv74+DgwMvv/wyP//8cyaf48eP07FjRwoXLoyzszO1a9dm1apVwD/JwUaPHq0f5srp/ep0Or766is+//xzPD09cXZ2pmfPnkRFRRmUS0hIYOzYsVSsWBF7e3u8vb2ZPXu2QZkVK1bg6OjIqVOnaN68OU5OTnz77bf6djh06JDeKY0jR47QsmVLChYsiJubG927dyc8PNyg3vLly/Pee+8xc+ZMKlasiKOjI9evX2fJkiXodDqOHTtG+/btcXZ2pmzZsixfvhyABQsWULFiRVxdXenRowf37t3T1/nkyROCgoLo3r07zs7OlClThr59+xIREWFw75YtW/LKK6+wadMmqlevjrOzM40bN+b06dOZ2nLVqlXUr1+fggULUrRoUVq3bs2pU6f059M+39KlS+f4+eaGpX52WCRoOXjwIMHBwXTq1IlWrVrpf1hkx5MnT/j222/p2LEjXbt2Ze7cublek184ODhY5L55QZzVQZzzj+TkZK5fv06pUqVwcHAgLi6OFi1acPjwYaZPn86aNWsoVqwY7du3Z9++fbnW9/777+Pl5cX69esZOHAg33zzDf/5z38A6NixI2PGjAFg27ZtHD16lA0bNmRb140bN2jevDlXrlxh4cKF/Pjjj3q/tN6M+fPnM2TIEACOHj3K0aNHqVu3rkltcPnyZQBKlSoFwKxZs+jfvz8tWrRg8+bNfPHFF6xatYq3334707XvvfceXl5erF27ljFjxjBv3rwsfRITEwkICGDdunV8/vnnbN26lRYtWvDaa68REhKir+/o0aM0b96ce/fuMX/+fDZt2kT//v25evWq/jwoqd6Nfb/z5s3j2LFjLFy4kFmzZnHw4EF69OhhUOa1115j1qxZfPjhh2zfvp3AwECGDx/ODz/8oC9ToEABkpKS6NOnD3369GHHjh106dKFo0eP0qhRI+rUqaN3Ajh16hStW7cmJSWFlStXsmDBAs6fP0/z5s0zBU2bN2/m559/ZurUqWzevJkiRYroz7355pv4+/uzceNGGjRowIABAxg9ejQ//fQTM2fOZPbs2ezZs0cfLILyu+3p06dMnDiR7du3M23aNC5evIifnx8JCQkG9z5z5gwTJ05k/PjxrFq1ivv379OtWzeSkpL0Zf7973/Tt29ffHx8+Omnn1i2bBkNGzbk5s2bAEZ/vrkRFRXFuXPnMrWPGlhkeCg+Pp66devi6+vLwoULcy0/Y8YM/vzzT6ZMmcLTp0/56quvcHJyYtCgQSrYCoJgCdJ+GEdHR/PFF19w69YtvvjiCwAWL17MpUuXOHXqFLVr1wagQ4cOvPzyy4wdO5ZDhw7lWHeXLl346quvAGjbti27d+/m559/ZvDgwZQoUYJKlSoBULduXTw9PXOsa/r06Tx+/Jjdu3fry/r7+1OuXDm++uorli9fjo+PD2XKlAEwGPbKiZSUFJKSknjy5Am7d+9m/vz5NG3aFC8vLx4+fMiYMWMYNmwY06ZN01/j7u5Ojx49+Oyzz/Dx8dEff+WVV5g6dapB/Vn5LFmyhOPHj3Py5El9u/r7+3Pz5k3Gjh1LQEAAAB9//DHly5fn4MGD2NnZ6dsxjbQ6y5QpY/T71el0bNmyRZ9ptWTJknTs2JHQ0FD8/f3Zv38/GzZsYPPmzXTu3Fl/z4cPHzJu3Djeeust/TBXcnIy48aN4/XXXze4h5ubG0+fPjVw+vLLLylSpAghISH6XrnGjRvz0ksvMXv2bP0zB/D06VN27txJoUKFMvm///77BAUFAUoOk40bN7JkyRIuXbqk30vnzJkzLFiwQB8gFy1alO+//15fR3JyMk2aNKFMmTLs3r3bYD7WvXv3OHHiBCVLlgSUeV/t27fnt99+o2nTpjx48ICxY8cyaNAgFi1apL+uU6dO+v+vWLHCqM83O86dO8fw4cMNejT9/f2ZMWOGwfOWn1ikp8Xf35/XX3+d6tWr51r24cOH7N69m48++ggfHx/q1q3LoEGD2LRpk0V6W6Kjo1W/Z14RZ3UQZ/MRHx+PnZ0ddnZ2eHl5sWzZMsaPH8/bb79NdHQ0hw4domrVqvofvAA2Njb06dOHX3/9lcTExBzrz5jN08fHh+vXrz+T66FDh2jZsqVBcOPi4kLnzp05ePAg8Gzt3K9fP+zs7PTDCs2bN2flypWA0pPx8OFDXn31VZKSkvRfrVu3JjU1lRMnThjUlf4XV06EhITg4+PDyy+/TGRkpL5ef39/fv/9d5KTk3n8+DFHjx7ljTfe0Acs5qBLly4GqeHbt29PwYIF+fXXX/Vurq6uvPLKKwbvuW3btkRGRnLjxg1ASXthyns+dOgQ3bp1MxhGLFeuHE2aNNF/fmn4+fllGbAABr/w3dzccHd3x8/Pz2Dzv8qVKxMXF8f9+/f1x3788Udq1apFoUKFKFCggD6Y/Ouvvwzqr1mzpj5gAfRBQtpze/ToUR49epTjH/PpP9/0bZj+882Oc+fO0aRJk0xDsKGhoTRu3Jjz589ne605sfqJuGkfXPofTnXr1iU2NpabN29StmxZVX1y+2FojYizOoiz+bC3t+eXX35Bp9NRtGhRypYtq5/wmJiYyL1797LsAfH09CQpKYnY2NgcJ6um79ZPu9/Tp0+fyfXevXtZ/gHm6empn7/wLO08adIkAgICcHR0pHz58garmNKCoIYNG2Z57bVr1wxee3h4GHXP6Ohozp49m20wEhkZiU6nIyUlRT9MZS7c3d0NXut0Otzd3fXzO6Kjo4mNjc12z5tr165RtmxZUlJSsLe3p3DhwkbdN6dn6dy5cwbHcmrHjPezt7fP8jkD9M/ahg0beOONN+jduzeTJk2iePHi+knoGZ/H3Oq6e/cuQI6fizGfb3YT14cPH55tQrnY2FiGDRtm0hDTs2L1QUtMTAwuLi4UKPCPatrDcf/+fdWDFi8vL1XvZw7EWR3E2XzodDr9xNOMeHl5UbRoUc6cOZPp3K1btyhQoICqiSWLFi3KrVu3snQpWrQo8GztXLFixWzbIC0g+/nnn6lQoUKm8xnvl9OuuRnrrV69OkuWLMnyvLu7O0lJSdjY2OjnSZiLjPMjUlNTiYqK0vcuFCtWTD+MkxVVqlQBoFChQka/XzDu80vDlHqNYe3atVSvXt1gIuyVK1eeqa7ixYsDcPPmzWwnjhvz+WZFVFRUph6WjISGhhIVFZVtHebC6lcPpaZmTthrzIMTFBREcHAwwcHB7Nixg3PnzpGSkqI/HxERYdBlGx8fr580l8bVq1eJiYnRv46Li8s0ozw8PNwgAVVMTIx+MloaYWFhxMfH619HR0cbzA5PSUkRP/ETv/+T1r2fk1/NmjX5888/9asn4uPjOXfuHGvWrKFRo0bY2dlx9epVg5U0cXFxmX7RZueX/q/Y3NqvWbNm7Nu3z+BYeHg4W7Zswc/PT++X8ZdyXtqvSZMmuLi4cO3aNZydnalRowb16tWjXr16lCtXjoxcunQp0+ebvp3TPt+AgAAuX76Mh4eHfm5PWr1Vq1bl2rVrFCxYkCZNmrBs2TLCwsKy/Xzt7OyMar80r82bN+snn0ZERLB69WoeP35Mo0aNiI+Pp0qVKsTExOgD2ox+Op0u1/ZLTEzM1INRu3ZtNm7cqD8eHR3N8ePHOXLkCH5+fvrnLz0Zvz8g83BOUlKSwfuOi4vj9u3bBmWio6MNej1iYmIyrYYKCwsz+Oyy+/5t2LAhzs7OLFq0KNvvX2M+36zaL23oLTdu3Lhh8s+XXbt2MXToUAYMGMCnn36a6z2svqelaNGixMXFkZSUpO9tSftGz6n7b9asWTn+tZXxLxEHB4dMa84zfvO7uLhkWubl7e1t8LpIkSKZuvEy1luiRAmD1zY2Npm6l5/VL6OPt7e3wcNijX7pSfNLSUnRT6qzRr/0pNWb5mytfmmk90tztia/jOm7M/qlpKQwbNgwFi9eTOfOnZk0aRLFihVj/vz5XLhwQf8XYbly5QzmE7i4uGTqOs/OL21YZ86cOfTu3Zu//vqLGjVqAJmfv+DgYJYuXUq7du0YO3YsdnZ2TJ48madPn+p/CNvZ2WX6CzS79kv/l3ZO7ff1118zYsQIbt68SZs2bXB0dOTKlSts27aNKVOmGFxXsWJFg1wsXl5eBu2c9vlWqlSJhQsX0rJlS4KDg/Hx8eHhw4ecPXuWK1eu6CeNfvvtt7Rq1Yq33nqLoKAg3N3dOXv2LE+ePGHUqFGA8rlu2rSJli1b8ujRI6pUqUKhQoWy/P4A5Q/Uzp07ExQUxO3btxk1ahSNGzfWzxUZOHAg27Zto0OHDnz88cfUqVOH+Ph4Lly4wKFDh1i/fj3e3t4Gf+hmbD87O7tMGZK//fZbGjduTLt27QgODubp06eMGzeOokWL8tFHHxn1/QHKfJX0FChQwGB1nouLS6bhpR49evD+++8zZswYWrVqxcGDB9m6datBmWrVqhl8dmntl9Y7lN5v4sSJjBgxguTkZHr37o2trS1HjhyhYcOGdOrUyeDzHTFiBNWqVcvy883YfsbmOipdurTJP//atWunn2MWGxvL3Llzc7yH1fe0vPTSSwD897//1R87deqUPr+C2mT8a1MLiLM6iLM6hIWF4eLiwoEDB2jWrBlDhw6lZ8+eREdHs2PHDlq1apXne9SpU4fx48fz008/0aRJE/1qlawoXbo0hw4domzZsgwaNIh+/frh5OTEgQMH9AFZfrTzhx9+yKpVq/j111/p06cP3bt3Z+bMmXh7e2f6xWEs9vb2hIaG0rt3b7755hvatWvH4MGD2bt3Ly1atNCXS5uk6uLiwuDBg+nSpQuLFy82GJaYO3cuNjY2dOjQgfr16xvkrcmK999/n7p16zJw4EA+/PBDmjdvzvr16w3KrF69muHDh7No0SI6dOhA//79Wbt2LS1bttSXMXXSc506ddizZw8AgYGBDB48mKpVq3L48OF8H+oYPHgwo0aNYsGCBXTt2pXffvvtmbarSCM4OJhly5bxxx9/0KtXL/r168eRI0f0QUf6z3fq1KnZfr4ZcXd3zzUTtb+/f763F1how8TY2FiioqK4cOECU6dOZf78+dja2lKqVCni4uIYMWIEo0eP1n/Df/XVV/z111+MGjVKv+S5ffv2Wc6Szu8NE9P3AGgFcVYHcVYHcVYHNZ11Oh1ff/01n3zySZ7rkrbOH86fP0/jxo2znMDOeQUAACAASURBVIzr5ubGkSNH8rzs2Wo3TDxy5AiDBw/W5w147733GDx4MBcuXNAnkUo/Fjh8+HCqVq3KyJEjGTduHC1atODNN9+0hLrVP1hZIc7qIM7qIM7qoEVn0Ka3Fpx9fHw4evRoph6XgIAAswQsxmKRnpb8JL97WgRBEATzYs6eFiH/iYqK4saNG5QuXdqsQ0JW29OiZTLuCaEFxFkdxFkdxFkd1HROTU01W8AibZ3/uLu74+npqcocloxI0GIi5swAqRbirA7irA7irA5adAZteouz8UjQYiLPOivfkoizOoizOoizOmjRGbTpLc7GI0GLIAiCIAiaQIIWE0m/qkkriLM6iLM6iLM6aNEZtOktzsYjQYuJXLp0ydIKJiPO6iDO6iDO6qBFZ9Cmtzgbjyx5FgRBEATB4siSZ0EQBEEQnhskaBEEQRAEQRNI0GIiaduuawlxVgdxVgdxVgctOoM2vcXZeCRoMREtzpMRZ3UQZ3UQZ3XQojNo01ucjUeCFhMpUqSIpRVMRpzVQZzVQZzVQYvOoE1vcTYeCVoEQRAEQdAEErSYSFxcnKUVTEac1UGc1UGc1UGLzqBNb3E2HglaTOTWrVuWVjAZcVYHcVYHcVYHLTqDNr3TnHU6Hbt378623Pjx42nWrJlaWjliqXaWoMVEvL29La1gMuKsDuKsDuKsDlp0hmf3vnLlCgMGDMDLywtHR0cqV65MUFAQN27cAJSAQafT8emnnxpct3DhQsqXL69/PWDAAHQ6Hd9//71BuTFjxtCyZcs8OY8cOZLNmzfrX7ds2ZIxY8ZkW3727NlUr16dggULUrx4cZo3b86mTZu4cuUKOp0ux6+09tDpdIwdOzZT3e3bt881yMoPJGgRBEEQXmguXLhAvXr1uHv3Lj/99BN//fUXS5cuJSkpienTp+vLOTo6MmvWLKKionKsz9HRkS+++MLs+/O4uLhQtGhRo8p+//33fP7553z22WecO3eO3bt3ExgYyL179yhTpgyRkZH6r549e9KnTx+DY2XKlAGgdOnS/Pjjj6Sm/pM8/8iRIzx9+tSs781YJGgRBEEQXmg++OADKlWqxObNm2nevDlly5alcePGzJs3j88//1xfztfXl+rVq/PVV1/lWF+nTp3Q6XR89913JrtcvnyZ5s2b4+joSL169Thz5oz+nCnDQ9u3b+eNN96gb9++VKhQgdq1azNkyBAGDhyIra0tnp6e+i9HR0ecnJwMjtna2gLg5+dHcnIyhw8f1te9dOlSXn/9dZPfmzmQoMVEwsPDLa1gMuKsDuKsDuKsDlp0BtO979y5w969ewkODkan02U6X7hwYYPXX375JfPnz+f69evZ1ung4MDnn3/O119/zaNHj0xyHjt2LEFBQfz+++9UqFCB7t27k5ycbMI7UvDw8ODIkSNERkaafG16dDod/fr1Y/ny5YCyu/PPP/9MixYt8lTvsyJBi4l4enpaWsFkxFkdxFkdxFkdtOgMpntfvHiR1NRUqlSpYlR5f39/GjVqxBdffJFjuYEDB+Lq6srMmTNzrTO985tvvknv3r3x8fFh4cKFREZGsmPHDqPc0jNmzBhSU1MpVaoUNWrU4IMPPmD//v0m1wPQv39/1qxZQ3x8PJs2bcLb29tiE4IlaDERFxcXSyuYjDirgzirgzirgxadQR3vSZMmsXjxYi5evJhtmQIFCjB+/HimTJnC/fv3Dc6tWLECFxcX/de9e/f05xo0aKD/v5ubG1WqVOHChQsmO5YpU4aTJ09y7NgxBg4cyJUrV2jdujXjx483ua6qVavy0ksvsXnzZpYtW0b//v0t9nxI0CIIgiC8UEQ9iuL3yN+JehRFpUqV0Ol0JgUGTZs2xd/fn3HjxuVYLjAwkNKlSzN16lSD4126dOH06dP6Ly8vL/25rIaonhWdTke9evUIDg5m27ZtfPnll0yaNOmZJgi/8cYbTJ8+nb179/Laa6+ZzdFUJGgxkZiYGEsrmIw4q4M4q4M4q4MWnSFn73NR5whYHoDHVA98v/fFY6oHfXf0pUGzBsyYMcNghUwaDx48yLKuL7/8ktWrV3Pu3Lls72djY8PEiROZOXMm0dHR+uOFChXC29tb//Xw4UP9uWPHjun/Hxsby19//WX00FVuVK1alaSkpGcKWgIDAzlx4gRt27alRIkSFns+CljkrhomNjZWc/tEiLM6iLM6iLM6aNEZsvc+F3WOJv9pQmx8LHZJ4JwIsQ4QeikU53rO2C62pW3btowaNYrKlStz+/ZtfvzxR+zt7Zk2bVqm+urWrUu3bt2YP38+Hh4e2fp0796dqlWrsmzZMho2bJir89KlS/WrlMaPH4+HhwevvPJKtvXfvn2b06dPGxzz9vZm5MiRVKhQgRYtWlCyZEn+/PNPPvvsM1q0aPFMmx0WL16c27dv4+joqHe2BBYLWlauXMn69euJi4vD19eXESNGZLv+/Pz58yxYsIC///6bggUL0rlzZ/r372/WbjRjKVeunOr3zCvirA7irA7irA5adIZsvFNT2TLuNR7MN/xFG2cHlT+CSLdHNB3TlNJ/lObNN98kJiaGMmXK0L59e4KDg7O918SJE9mwYUOuTl9++WWOgUd65/Hjx/Pvf/+bkydPUr16ddavX0+BAtn/ql64cCELFy40OLZv3z7atGnDokWLmD59Ovfv38fDw4MOHTowYcKEXH2zI30waKnnQ7dv377M/WH5zI4dO5g1axajR4/Gy8uLOXPmkJqamuUs6+joaAYOHEiXLl3o2LEjERERTJkyhV69etGnT59M5R89ekSnTp148OCBJrf7FgRBEMxAXBx06ACHDuVYbEILGN9K+f/tkbdxd3ZXQU7IitjYWNzc3Ni6dSvOzs5ZlrHInJYNGzbQs2dP/Pz88Pb25l//+hd//PFHluvrf/31V1xdXXnnnXcoVaoU9evXJzAwkDVr1mQ5/igIgiC8oGzcCDqd8lWoUK4BC4BduhQoN2Jv5KOcYA5UD1oSEhK4ePEiderU0R/z8vLC09OT8+fPZyqfmJiIvb29wTEHBwfu3LljkQ2bwsLCVL9nXhFndRBndRBnddCMc2QkzJypBCrdu2Oq9cK6//y/tGtps6oZi2baOh2WclY9aImNjSUlJSXTRKnChQtnWssOULt2ba5fv87WrVtJTk4mMjKStWvXAhisbVeLihUrqn7PvCLO6iDO6iDO6mDVzvfuwQ8/QOvWUKoUDBumP2WKtftIuPz/qZT+Ff0tNjRk1W2dDZZyVj1oMXVIp2LFiowcOZLvv/+egIAA3n77bX0mvpwm4gYFBREcHExwcDA7duzg3LlzpKSk6M9HREQYLEGLj4/PFDlevXrVYFlXXFxcptTN4eHhxMXF6V/HxMRw9epVgzJhYWEGS8yio6OJiIjQv05JSTGbX8YhtvDwcBITE63aL6v2c3BwsGq/9KT5pTlbq18a6f0cHBys2g8yt5+Dg4NV+0Hm9kurx1r9smq/W7duWZffiROwYgV06gQeHlx95x1i9u2D//9OiQPCgX9+ciiv49K9jgGuAi6jQTceol2AaHC1dWXGKzPM2n6mfL5pPzu08vMZ0Dvn1W/Xrl0MHTqUAQMGZNpBOytUn4ibkJBA+/btmTx5Mr6+vvrjgYGBBAYG0qVLlyyvS01N5e7du7i6unL69GlGjRrFunXrMq04kom4giAIzwnHj8OsWfDjj2apLurTobxe9Tyhl0L1xwIqBTC93XR8SviY5R7Cs2PMRFzVlzzb29tTqVIlTp8+rQ9aIiMjuXXrFj4+2T80Op2O4sWLA7B//36qVq1q9Bbd5iQ6OpoSJUqoft+8IM7qIM7qIM7qYBHniAgIDYWzZyFDFlljiQb01n5+8O670KMHODriDoSgZMS9EXuD0q6lrWK1kDwfxmORPC3dunVjzpw5VK5cmZIlSzJv3jxq1qyJt7c3YWFhfP3110ybNk3fIOvXr6dWrVrY2NgQGhrK7t27s0z2owbph1q0gjirgzirgzirg2rOFy/CmjUwerRZqkssXBgGDoR33oGqVbMs4+7sbhXBShryfBiPRYKWDh06EBMTw4wZM/TJ5UaOHAkoY2/Xr1832Ir7jz/+YPHixSQkJPDSSy8xefJkatSoYQl1gz0itII4q4M4q4M4q0O+OaemKj0p69crX3/8YZ56mzeHd9/Fq2dP+H/WVq0gz4fxWCS5XH4ic1oEQRCsiNRUOHUKvvwSjMgeazLDhsHgwZDD9AJBG1jlnBatk5KSgo2NtvaZFGd1EGd1EGd1yJNzair897+wejV8+615xdLz9Ck4OBgceuHa2kJYyllbrWQFSBIgdRBndRBndXghnFNT4cQJ+OQTeOklqFPHfAGLt7cymfbbbyE6WrlXamqmgOWZvK0AcTYeGR4yEYmI1UGc1UGc1eG5dU5JgWPHYO1a5StDjpE8M3Qo9OsH9eopGW+N4LltaysjP5xleCgf0NqDBeKsFuKsDuKsDtk6p6TAkSNKkLJuHdww8349BQvCW2/BtGlgZ2fy5c9VW1sxlnKWoEUQBEHImeRkOHz4n0AlMtL895gzB95/HzT4C1xQD3k6TCR9emKtIM7qIM7qIM7qEHHtGuzZowQSXl7QsqUSWJgjYClSRFnxs3cvJCUp81M++MAsAYsm21qcjUZ6WkzE7hm6Ky2NOKuDOKuDOOcjiYlKILF2LXbr1kG6vWnMQvfuMGgQBASAvb156/4/mmnrdIiz8UjQYiJaS7UM4qwW4qwO4mxm4uNh925l6GfTJn2gYnbjmBgoXNjctWbCqts6G8TZeCRoEQRBeNF48AB27FCSvW3fDnFxuV9jDAUKQM2aymTa99+HwECjV/0IgjFI0GIi8fHx+i25tYI4q4M4q4M4PyORkUpPysaNyhBQLnvHxAMmGX//vZJLpVixvFjmGatoaxMRZ+ORibgmcunSJUsrmIw4q4M4q4M4G0lSkrK3T6lSSm+Hl5fS+7FrV64BC0Cuxt26KRNzIyKUibSDB1s8YAF5PtTCUs6SXE4QBOF54dEjCAlRelSWLVOCCXPz3nswZQq4uJi/buGFRpLLCYIgPO/cvg1btiiByu7dyn485uTDD6FXL2jWDGxtzVu3IJiIBC2CIAhaIjlZyUj74Yfwxx/5cw87O2Wzw27dJNmbYFXI02giV829t4YKiLM6iLM6vJDOT5/Ctm3w9ttQsiT4+Zk/YPn3v+HyZWVIKSGBq76+mgxYXsjnwwJYyll6WkxEi/NkxFkdxFkdXhjnBw+UQGXjRmV5srmWJQM4O0OHDtCkCfTtC+7umYposZ1Bm97ibDwStJhIkSJFLK1gMuKsDuKsDs+tc2qqMuwzciT8+qv5JVxcoEsXZX5Ku3ZKLpUc0GI7gza9xdl4JGgRBEGwFImJcOCAMol20ya4ft289bu6QteuSqASEACOjuatXxBURoIWE4mLi8NFY0v9xFkdxFkdNO/88CHs3KkM+2zfDvfvm/dmhQsrE2h79YK2beEZE4BpsZ1Bm97ibDzam2VlYW7dumVpBZMRZ3UQZ3XQpPMff8CCBdC+PRQvDn36wMqV5gtYihaFt95SgqHbt2HxYujY8ZkDFtBmO4M2vcXZeCS5nCAIgrlJTYU//1R6UzZtgt9+M/89SpRQdk3u3RtatFCWKQuChpHkcoIgCGqRnKxMoE3b3+fvv81/Dw8PZX+f3r2heXNlg0JBeIGQJ14QBOFZefIE9uxRgpQtWyAqyvz38PKCnj2VOSpNm0pWWuGFRua0mEh4eLilFUxGnNVBnNXBos6pqXDihDI3RadTlg137gyLFuUYsJhsXLo0DBsGhw8rK4pmzVISyqkYsGjx2QBteouz8Vi0p2XlypWsX7+euLg4fH19GTFiBEWLFs2y7JkzZ1iwYAEXL17EycmJBg0a8MEHH1CoUCFVnT09PVW9nzkQZ3UQZ3VQ3TkxEQ4d+mdZ8jNkAjXKuFw5pTelVy9o0MDi2Wi1+GyANr3F2Xgs9l2xY8cOli9fTlBQEHPmzOHRo0dMmDAhy7KPHz/m008/pUqVKixcuJAvv/ySv//+m1mzZqlsjeaWpYE4q4U4q4Mqzg8ewE8/Qb9+SrbYNm2U3o5nTF2eo/Fbb8GxY0oK/alToVEjiwcsoM1nA7TpLc7GY7Gelg0bNtCzZ0/8/PwA+Ne//kW/fv0IDw/H29vboOy1a9eIi4tj4MCBuLi4UKpUKTp27MjGjRstoS4IwvPItWuwebPytX+/0sNibgoUUAKgVq2UNPovv6wMMwmCYBQWCecTEhK4ePEiderU0R/z8vLC09OT8+fPZypfpkwZChUqxM6dO0lOTubBgwccPnyYevXqqakNQExMjOr3zCvirA7irA5mc46Lg7VrYfRoqFNHGZ756CMIDTVvwGJnR0zbtvCf/yg5VHbuhFGjoEYNqw5YtPhsgDa9xdl4LBK0xMbGkpKSkmnvgsKFC3M/i2RLzs7OTJ06lTVr1hAQEEC3bt1wcHBgyJAhainriY2NVf2eeUWc1UGc1eGZnZOSlJT5acFCoULK0uFvvoHTp80rCUrK/CVL4PZtYhcuhIEDlSRwGkGLzwZo01ucjcciQUtqqmn57J48ecLUqVNp2rQp8+fPZ+rUqdy9e5fZs2dne01QUBDBwcEEBwezY8cOzp07R0pKiv58REQE0dHR+tfx8fGEhYUZ1HH16lWDaDIuLo7EDH+BhYeHE5du99WYmJhMW3aHhYURHx+vfx0dHU1ERIT+dUpKitn8Ms7oDg8Pp1ixYlbtl1X7lStXzqr90pPml+ZsrX5ppPcrV66cVftB5vYrV66c8X7Xrim9KW++SZyrK+EtW8LZs//4Aen3To4BMs5aCQPi072OBiLSvU4Bzv3/XwYNgvnziTh7luhr1+DNN6FIETw9Pa2m/cC4zzftftbql137pf/ZYY1+6UnzS3O2Vr800vulOefVb9euXQwdOpQBAwbw6aefkhsWyYibkJBA+/btmTx5Mr6+vvrjgYGBBAYG0qVLF4Py27ZtY+XKlaxYsUJ/7OzZswQFBbF582aDCUGSEVcQXnCuXVNypmzZAvv2QUJC/txHp4Pq1ZU8KoMHK7lUrHi4RxCsHavNiGtvb0+lSpU4ffq0PmiJjIzk1q1b+Pj4ZCr/9OlTdBl+GNjY2JjcYyMIwnNISgr8/rsSpGzenD9DPWnY2kLr1kpW2m7dQINLVQVBy1hsXV23bt1Yt24dhw4dIjw8nClTplCzZk28vb0JCwujf//++u4jX19fbt++zQ8//MDNmzcJCwtj7ty51KxZU/VlV1l1oVo74qwO4qwOYWFhSibabdvgvfegTBmoXx8mTszfgOWHH5QEciEhyn1NCFg0284aRIve4mw8Flvy3KFDB2JiYpgxY4Y+udzIkSMBZfzt+vXrJCcnA1C+fHkmTpzI0qVLWb9+PU5OTtSpU4f33ntPde+KFSuqfs+8Is7qIM75SHKysung3LlUXLky/+9XtqzSm9KjBzRpkudMtJpp53Ro0Rm06S3OxiO7PAuCYJ3ExSm9Glu2KL0q6SYW5guVKyvzUnr0AF9fmZ8iCCpjtXNaBEEQsuT6ddi6VZmbsndv/k2iTaNOnX96VKpVk0BFEKwcy+eK1hjR+f3XXj4gzuogzs9ASoqyAeG4cUoAUbYsDBmiJGDLJmAxi/H48XDxojKBd8wY8PHJ14DF4u38DGjRGbTpLc7GIz0tJpIxT4sWEGd1EGcjefIE9uz5Z1lyZKRJl5tsXKgQdOqkbET4yivKzswqI8+GemjRW5yNR4IWE/Hy8rK0gsmIszqIcw7cuqUM+2zZoqTJf/LkmasyyrhwYejaVZmj4u+vZKe1IPJsqIcWvcXZeCRoEQTBvKSkwN9/KwFKSAjcvw/Hj+f/fYsXV3Kn9Oyp5FKxt8//ewqCoCoStJhISkoKNlawbbwpiLM6vNDOjx/D7t3KBNqtW5WNAfOJFNJNxvP0hO7dlaEfPz9lF2Ur5IV+NlRGi97ibDzaaiUrQJIAqYM4q0OenCMjlYRrXbpAsWLKcMyiRfkasACEeXjA0KFw8CDcuAHz5ik9K1YasMAL+GxYEC16i7PxSJ4WE5GIWB3EWR1Mck5NhTNn/kmXf+xY/splZORI6NmTlHr1sLHiACUrnvtnw4rQorc4K0ielnxAaw8WiLNaPJfOCQlKj8bmzcrX1Yx7IOczZcoovTkBAfolydpr5ef02bBStOgtzsYjQYsgCIbcuwc7dihBys6dEBtr/LUFCkDdukrulXRb0RtNgwbK/JSePUGDqc0FQchfJGgxkYiICM0tTxNnddC0c3i4EqRs2QKHDil7/RhL4cLQsKHy/0eP4PBh0yQaN/4nUClXznhnDSHO6qFFb3E2HglaTMTOzs7SCiYjzuqgKef/b0Bot3q1surH1El1FSsqgUpysjLxdvdu4wMdnQ6aNVMClR49oHRpk26tqXb+P+KsHlr0FmfjkaDFREqUKGFpBZMRZ3Wweue4OCWxW9qy5Dt3MNpYp1N6RGrVgsREJchZvVqZnGsMNjbQooUSqHTvDiVLPuu7sP52zgJxVg8teouz8UjQIgjPMzdv/rPaZ+9eiI83/tqCBZUJsD4+8PSpMmz03XemO0yfDn37gru76dcKgiCkQ4IWE4mPj8fBwcHSGiYhzupgFc6pqXD69D/zU06ezLF4PGBg7OUF7duDmxskJcGBA7Bxo/H3T9vnp0cPpZ5sli3mBatoZxMRZ/XQorc4G48ELSZy6dIlqlWrZmkNkxBndbCYc3w87N//z7LkGzeMvvQSUK12bejYEYoUUXY9XrTItPsXLfrPPj9t2uT7Pj/ybKiDFp1Bm97ibDySXE4QtMidO7B9uxKk7NqlzFcxFjs7aNUKOneGEiXgyBFYs8a03ZbT0uf37KnMVdFYsjdBEKwPSS4nCM8T4eGwaZPy9csvpuVBKVpU6U3p3FnZWHD7dpgyBa5dM76O0qWVIKVXL2VSrq2t6e9BEAQhD0jQIgjWSkqKMidl0yZlXsm5c6Zd/9JLyrBNp07KjscbNsCoUXD5smn1fPyxEqzUr6+sAhIEQbAQ8hPIRK6qncbcDIizOpjFOSFBGe4ZMgTKllUyxE6aZFzAYmMDzZvD5MnKkuTly5WJuW++CU2aKD0rGQKWLI2LF1fq2blTCZwmT1ZyslhJwPLCPhsqo0Vn0Ka3OBuP9LSYiBbnyYizOjyTc2Kikjtl8mRl+OfhQ9PS5ru4QLt2yk7Lr7wCf/4JffrAv/5lnHPaf+rVg1dfVa4tW9bkt6EmL8yzYWG06Aza9BZn45GgxUSKFCliaQWTEWd1MNo5Lk7pTdm4EX780fQbeXkpQUrXrtCypbLz8urV8OmnSl4WY6lRgyKvvaYEKt7epntYiOf62bAitOgM2vQWZ+ORoEUQ1CA6WsmbsnGj0rPy9Klp11evrgQp3bqBr68y/LNqFXz4IVy8aHw9tWopE2l79YKqVU1zEARBsDAStJhIXFwcLi4ultYwCXFWh0zOly8rQcrGjcomgqas9rGxgaZNlUCla1elJ+TiRfjpJxg0CM6eNb6umjXhtdeUybSVK+fsrAHEWR206Aza9BZn47Fo0LJy5UrWr19PXFwcvr6+jBgxgqJFi2Yqd/r0aYYPH57puIODAzt37lRDVc+tW7fw1lBXOoizWtw6dQrv06eVvCf//a/pmxCCkkW2d29lxU+JEnD1Kvz8sxKs5JLdNhONGsH8+UrvSnbOWmxncVYFLTqDNr3F2Xgsllxux44dzJo1i9GjR+Pl5cWcOXNITU1l5syZmcomJiby8OFDg2Njx46lZMmSfPbZZwbHJbmcoCp//w3r18O6dXD8uGnX2ttD27bK/j6NGkGNGsp+PydOwJgxyoohE7LbUqeO0qPy6qtQrpxpLoIgCBbGqpPLbdiwgZ49e+Ln5wfAv/71L/r160d4eHim6M3Ozs6gByYqKoqwsDAGDBigprIgKEuIz5z5J1AxZZgGwNVVSfLWrZuy2ictsI6OhsWLISjItGGkKlUgMFAJVGSOiiAIzzkWCVoSEhK4ePEi7777rv6Yl5cXnp6enD9/Ptcup5CQEIoXL07dunXzW1UQlCDi+HElSFm/3rSJrwAlS/4zibZVK6WHBeDxY2XVz48/KquJkpKMq69sWaVH5bXXoHZt0OlM8xEEQdAouWaLio6OZv78+Tx69CjTuUePHjF//nzu3r1r0k1jY2NJSUnJtGSqcOHC3L9/P9frQ0JC8Pf3x8YCya7Cw8NVv2deEednIClJ2YTwo4+UIKFRIyU5Ww4Bi97Y3V0JUsaMgaNHlSGe775T8qnY2sLu3TBgAHh4KL0k27blHrB4ecGwYUp9V67At98qw0F5DFgs3s7PgDirgxadQZve4mw8uf7WX716NfHx8VmOLzk7O5OQkMCqVatMumlq6rNPozl79izXr1+nXbt2OZYLCgoiODiY4OBgduzYwblz50hJ1+0eERFBdHS0/nV8fDxhGSZOXr16lZiYGP3ruLi4TMFbeHg4cek2q4uJicmUKTAsLIz4+Hj96+joaCIiIvSvU1JSzOaX8UEKDw83mOFtjX5ZtZ+np6f6fnfvEr5oEbz9ttI70qoV4XPmEJcu90kMmbPIhgHxZcrgOWQIHDxI9H//S8S8efDFF9CoESmpqZybM4cUJydlY0F/fyKWLiU63fuO/389Bn5ATLducPAgXL9O3BdfEF68uEGgktfnz9PT0yKfr7F+kPnz9fT0tGo/yNx+RYoUsWq/rNovISHBqv2ya7/0Pzus0S89aX5pztbql0Z6vzTnvPrt2rWLoUOHMmDAAD799FNyI9eJuP379+eTTz7Bx8cny/NhYWF8/fXXLFu2LNebpZGQkED79u2ZPHkyvr6++uOBgYEEBgbSpUuXbK+dOnUqV65cYc6cOVmel4m4gkk8sszwFQAAIABJREFUeqQMzaxbB1u3mpaNtnJlZRlxjx5K7pSMvR7Xryup9L//XlkFZAxFiyq9NF26KCuIZPdkQRBeEMwyEff27duUKFEi2/NFihQhKirKJDF7e3sqVarE6dOn9UFLZGQkt27dyjY4AiXY2b9/v8FcGEEwmbg4ZUhmzRplt+MnT4y/tnZtJUjp0QN8fDIHKo8fK3lZlixRhoGM6VV0cFAClddfV4aQ0ua8CIIgCAbkGrS4urrmGLjcuHGDQoUKmXzjbt26MWfOHCpXrkzJkiWZN28eNWvWxNvbW997M23aNIP7Hjp0iMTERFq1amXy/cxFTEyM5lIuizPKnj5btyqByo4dpmWkbdz4n0ClYsXM51NTYft2Yn76iSKbNhnfW9O6tRKo9OgBbm7G+5gReTbUQZzVQ4ve4mw8uQYt9erVY9WqVUyaNCnL86tWraJevXom37hDhw7ExMQwY8YMfXK5kSNHAsr42/Xr10lOTja4ZteuXTRr1syimQNjY2M193C9sM5xcbB5sxKo7NxpfKBiawstWijBRLduUKpU5jKpqfD778qy5Tt3FGfAKOM+fWDaNChd2th3km+8sM+GyoizemjRW5yNJ9c5Lbdu3eK9996jdOnS9O7dm9L//0F7/fp11q5dy/Xr11mwYIHB5CdLInNaXnDi45UAZdUqJWAxdujH3h78/ZVApUsXKF4863JRUbBihZJT5cwZ4+ouVQr694c331TyqgiCIAiZMMucFk9PT2bNmsXMmTOZOHGiwbnatWsze/ZsqwlYhBeU5GRlefKqVcqEWiOWzQPg6Kikze/VS5n0ml2Q++QJfPYZTJ9uvJOjoxIAvfkmtGmj9N4IgiAIecKopQlly5Zl2rRpPHjwQL+UycvLCzcLjcMLAqmpcOyYEqj89BPcumXcdY6O0KHDP/v75DTUeO6csvJn1izT3L7/XhkCku8PQRAEs5Jrnpb4+Hh++eUXHj9+jJubG9WqVaNatWq4ubnx6NEjfvnlFxISEtRwtQoyrnXXAs+V87lzStI2b28l4dvMmbkHLE5OytLk1auVdPnr1inZZLMKWB4/hmXLoFkzePllkwKWsNWrlWBq8GDNBCzP1bNhxYizemjRW5yNJ9eelk2bNnHixAmaNm2a6ZyzszObN28mMjKSXr165YugtVExq9UjVo7mna9cUQKOlSuNn0dSoICyfDgwUFlOnNvk7TNnlB6S5cvhwYPc6y9RQln5M2AA1KypOKdLsKQVNP9saARxVg8teouz8eQatOzevZu333472/O9evXihx9+eGGCFgcHB0srmIwmne/fh59/VoZ/jh417iKdDvz8oG9fpWelWLGcy8fFKUNLCxfCr78aL7dhgzLElCGfiibbWZxVQZzVQ4ve4mw8uQYtN27coEKFCtmeL1u2LDfTpTkXhGfmwQNlQ8JVq2DPHuN3O/b1VQKVV1/NenlyelJTlQBl0SKl9yaLPbUy8fLLSo/KwIFKxlpBEATBIuQatBQoUIDY2Nhsk8vduXMH2xdoZUR0dHSOGYKtEat2fvJESfq2apWSnfb/QyzRQI7GVaoogcprrynp9HMjNlaZqzJ/vjIvJjecnJS633kHGjY0amNCq27nbBBndRBn9dCitzgbT65Bi4+PD3v27KFSpUpZnt+zZw/VqlUzu5i1kpiYaGkFk7E658REJcX9qlXKUEu6Dbn0RbK6rkwZJZAIDFTS6Ruzw/HZszBvnjJXJYv7ZKJmTXj3XejXz+TJtFbXzkYgzuogzuqhRW9xNp5cg5bAwEA+/vhjdDodvXv3pnDhwgDcv3+fNWvWsGXLFqZOnZrvotaCl5eXpRVMxiqcHz9WNibcsAG2bMk1l4reuHhxZXlyYCA0bQo2uS54U4KiDRtg7lxlh+TccHb+p1elfn3jgqGsnK2hnU1EnNVBnNVDi97ibDy5Bi21atXik08+YcaMGaxevVqfpe7Ro0cULFiQTz75hFq1auW7qKBBYmKUoZ8NG5QstcZmp3Vxge7dleGfNm3Azs64627fVoZ/FiyAyMjcyzduDG+/reRUseDWEIIgCIJxGJVcrnXr1jRu3Jjjx4/rJ92WKlWK+vXr4+TklK+C1kZKSgo2xvy1b0Wo6pyQoKTP/89/IDQUkpKMu87BQVmR07cvdOxIioOD8c5nzyrZales0M+JyRZXV2VS7TvvQPXqxtVvJPJsqIM4q4MWnUGb3uJsPEbf0cnJCT8/PwIDAwkMDMTPz++FC1hAkgBly7lzEBysrN7p3VvZSTm3gMXGBgIClH18bt9WVg716gVOTrk7p6YqvTft2kGNGkqQlFPAUrOm0gMTEaEkpDNzwALybKiFOKuDFp1Bm97ibDy5bpi4fft2oyrq0KGDWYTySn5vmCgRcTpiY5Vlw4sWKSn1jcHODtq2VYZ/unYFd/csi2Xr/OSJ0qMyfTqcP5/7vXr1gg8+gCZNnnmuirHIs6EO4qwOWnQGbXqLs4JZNkycOnUqbm5uODk5kZqadXyj0+msJmjJb7T2YIGZnVNT4fBhJVBZs0aZYJsbzs7K0E/37sq/RqzKyeR85w7MmaOsBIqOzvnikiVhyBBlvoqKm3m+8M+GSoizOmjRGbTpLc7Gk2vQUqdOHcLCwvD19SUgIID69eujy+e/WAUr5NYtWLJEGYb5++/cy9vZQbduSqp7f38l78mzcP06TJ0KP/yQ+0Te2rWVIapXX82UrVYQBEHQPrmGStOmTWPp0qVUrFiRefPm0adPH+bPn8+lS5fU8LM60na51hLP7JyaCvv2KatrypSB0aNzD1heflkZuomIUNLwd+nyTAFLxMGDSgbaihWVTQtzClg6d4a9e+H33+GNNywWsLxQz4YFEWd10KIzaNNbnI3HqNVDJUqUoG/fvvTt25ewsDBCQ0MZMWIExYsXZ/bs2Tg6Oua3p9VgZ+zyWyvCZOf792HpUmX58J9/5l7e1VXJozJoUJ7ynABw4gR8/TV269fnXM7JSVkFNHSokh3XCnghng0rQJzVQYvOoE1vcTYeo4KW9Hh4eODh4UHhwoW5e/dutvNcnle0lmoZTHA+cQK++07JVGtMThU/P3jrLWWya8GCeZM8dAgmTlQy5ZJDCn93dwgKgvfey31DRJV5rp8NK0Kc1UGLzqBNb3E2HqOClvj4eA4ePEhISAhnz56lcePGvPPOOzRo0OCF2nfouWXjRpg0SQlacsPTU+nhGDQIXnop7/c+fBjGj1c2SMyJChXg44+Ve7+AS+0FQRAEI4KWb775hsOHD1OpUiUCAgIYN24cLi9w9tD4+HjNbSOerXNCgjJxde7c3Ctp2xbef1+ZP2KObsEjR2DcOH3PSkbiAQdQcrB88okyr6aAyR2DqvJcPRtWjDirgxadQZve4mw8uf4WCAkJoXjx4qSmphISEkJISEiW5WbOnGl2OWvk0qVLmtsgMkvnyEglCdwvv2R/YZEiymTYd981bidlY9i7F77+OttgJY1LtWtT7X/s3XlcVOX+wPEPoCK5gLghalqS5ZpLpblVlpZL7plbRlqmdV1C01xKvV7NJRO9tJjWTTFyxRYVl8xS/GmluSW4DCouiFKOImSDMvz+OM3IyIBnWJ6ZY9/368XrXmaGOR+OGI/nPOc506ZBp05Fvr5KYbljfjY8nDSrYcRmMGa3NOt328XllixZouuNXnzxxUIJKqiiXlzujvB//6fNQ8nt/jzNmmlHVXr3LrxTMTt3QqtWt3/dU0/B229D69aGGawIIYQouEJZXM5TBiOiEGRlaRNtR43S7oR8q6efhhkzoEmTwtvm0aMwcSKsWZP369q21ea2tG5deNsWQghxR/HsSQKi8Pz1l3b05PPPnT//9tvaHJPCmlh9/rx2NdCiRZCZmfvrHn8cpk7VrkQSQggh8uDWtYOjoqLo1asXzzzzDBMnTuTSpUt5vn7NmjUMGDCA9u3b06dPH7Zs2aKo9KbExETl2yyoxP/7P+3UjLMBS5ky2tVD//534QxY0tO1wU9IiLbOS14Dlm3btA8nAxZD7mdpVkKa1TBiMxizW5r1c9uRlpiYGCIjIxk/fjzBwcFEREQwderUXCf0Ll26lA0bNvDaa69x3333YTabFRdrDDdP5vvvKfvcc+BsQFinDqxdWziLs2Vlaeu7jB0L587l/doPPtCO+uQxZ8Vw+xlpVkWa1TBiMxizW5r1c9ugZe3atfTs2ZM2f/8re+zYsfTv3x+TyURISIjDa69cucKyZcuYPXs2jRo1AqBKlSrKmwHKlSvnlu26LCsL5s6FceMoZ7XmfL5HD+3IS5kyBd/Wr79qC77ldSVS3bowcyZ07qxrgq1h9nM20qyGNKthxGYwZrc06+eW00MZGRkkJCTQuHFj+2PBwcEEBQURFxeX4/V79+7Fy8uLc+fO0b9/f/r168d///tf/vrrL5XZxpGeri2r/+abcOuAxdtbu+R49eqCD1hSUmDIEHjoodwHLFWraneEPnhQW+NFrggSQgiRT3keaXnjjTd039H5/fff173R1NRUrFZrjpFaQEAAly9fzvH65ORkrFYr0dHRvPnmm2RmZjJv3jwsFgtjxozRvd3CkJaW5tmL65lM0L07/Pab/aE0oDRAYKB2Cqd9+4JtIytLO0ozZozz004ApUvDhAnalUr5uGza4/ezE9KshjSrYcRmMGa3NOuX55GWBg0aUL9+ferXr0/t2rU5evQoFouFe+65h3vuuYeMjAyOHj1KbRcXHnP1fkVWq5UbN24wfPhwGjVqRNOmTRk6dCibNm0iM5eJniNGjCAsLIywsDBiYmI4fPgw1mxHHZKSkkhJSbF/brFYiI+Pd3iPxMREh7kzaWlp/PTTTw6vMZlMpKWl2T83m805JijFx8djsVjsn6ekpDjcIdNqtRZO3+rVmBo3dhiwmIAEgEaNYM8ezA8/XLC+o0ehbVuSBg0iJduAxQLY6wYOhKNHSezXD3O2o2FpaWmYTCaHbee2/5KTk9XvPxf6srP12Zo9tc8me5/tHwSe2gc5919ycrJH90HO/ZeYmOjRfc723969ez26L7f9l/2/HZ7Yl52tz9bsqX022ftszQXt27RpEyNHjiQ0NJQJEyZwO7ddXM5m5syZVK1alRdeeMHh8WXLlnH69GldG7PJyMigQ4cOzJ49m6ZNm9of79u3L3379qVLly4Or1+/fj3vvfcea9asITAwEIDTp0/z4osvsmrVKipUqGB/7T9ycTmrFf7zH22dE2cDwgEDYOHCgt3U0GKBWbO0exRlZDh/zcMPw4IF0Lx5/rcjhBDiH0nP4nK657Rs376dJ554Isfjjz/+OLGxsS6FlShRglq1arF//377Y+fPnyc5OZm6devmeL3tsXPZrko5d+4cxYsXN+QEpkJ15Yp2Omjy5JwDlmLFtEHE0qUFG7AcOKANSCZPdj5gqVgRPvsMdu+WAYsQQogio3vQUrZsWX744Yccj//www/5OqLRrVs31qxZw44dOzCZTMyZM4eGDRsSEhJCfHw8AwcOtB8+uueee3jooYeIiIjg6NGjxMXFsXDhQjp06PDPvsv04cPaYOKbb3I+V6mSdufk4cPzP/k1M1O74ufhh+HQIeevGTQI4uO1exR5u3XZHyGEEHc43Zc8Dx06lOnTp7N7924eeOABvLy8iI+P5+jRoy6dGrLp2LEjZrOZ8PBw0tLSaNq0qX1SrcVi4cyZMw7zVSZNmkR4eDijRo2iVKlSPPbYYwwZMsTl7RaUs0uy3WL1aggN1a4UulWzZtqy+VWrAvlsNpngxRe1+xQ5U7u2dsrp8cdde1/dm/eQ/ewCaVZDmtUwYjMYs1ua9dM9pwW0UzLr1q3j7NmzZGVlUb16dTp37kzVv385eoKintPi9lnemZnavXxmzXL+/JAh2imhbLcMd7n5yy+198k2AcuuWDF46y2toWRJF+P1c/t+zgdpVkOa1TBiMxizW5o1hXLDxOyqVq3Kq6++WihxRuXWH6w//oA+feC773I+V6KEttLsyy/neEp387Vr2iXKn3zi/Pk6dSAyErJNni4qRvsLDNKsijSrYcRmMGa3NOvn0qAlPj6e9evXk5yczPjx4ylfvjw//PADlStXpk6dOkXVKEBbdbZHD3B2v4dq1bTTQY88kv/3P34cnntOm3R7Ky8veOMN7QqlfKy5IoQQQhQG3TMnf/zxR8LCwvDy8uLgwYP267LNZjNLliwpskBP45Z7Hi1dCi1bOh+wPPYY7N2b54Dlts1btmhf72zAUq2adlPDuXOVDljcdW+pgpBmNaRZDSM2gzG7pVk/3YOWpUuX8uabbzJ69GiKFbt5gKZhw4YcO3asSOI8UWpqqrqNXb+uXf3z4ovg7JYFo0ZpA45KlfJ8mzybP/gAOnQAJysR07Ej7N+vDYwUU7qfC4k0qyHNahixGYzZLc366T49dO7cOadrqPj6+pLu7AqWO1SNGjXUbCg5WTtd42wNHD8/WLwY+vXT9VZOmzMztZscfvhhzud8fLRF5N58022XMSvbz4VImtWQZjWM2AzG7JZm/XQPWoKCgjh27BhBQUEOj+/evduQO9yj7doFPXvC+fM5n7vnHli7Fh58MP/vb7HACy/AqlU5n6tQQZsf8/fdt4UQQghPoXvQMnDgQMLDw7l06RJZWVns2bOHpKQkvvrqKyZNmlSUjf8cWVna2icjRminhm719NMQFaXd+DC/0tK0Cb1btuR8rn59baG6e+7J//sLIYQQRUT3sf+2bdsyadIktm/fTsmSJfnwww85dOgQ77zzDq1atSrKRo9y602nCs1ff2mXKw8b5nzAMnEirF+frwGLvfnSJWjXzvmApXNnbSE5DxmwFNl+LkLSrIY0q2HEZjBmtzTr59LickZQ1IvLWSwWfLMt3FYoTp/WTgft2ZPzuTJlYMkS7f5C+WSxWPD19oaHHoKDB3O+4OWX4eOPtbksHqJI9nMRk2Y1pFkNIzaDMbulWVOoN0ycNWsWGzduzPF4eno6s3JbnfUOVOg/WNu2aYu1ORuwPPAA/PxzgQYs8Hdz8eLwr3/lfHLcOG0xOQ8asEAR7GcFpFkNaVbDiM1gzG5p1k/3oGXTpk3897//Ze7cuQ73BLJYLGzevLlI4u5oWVnw/vva6Zrff8/5fPfu8NNP2sClsLzyinYDRJtZs7TP83tDRSGEEEIhl65nnTNnDkeOHGHkyJFcunSpqJo8mu3O0wWSnq5drjx6tHbpcXZeXtrlxqtXQyGd3nJoHjdOu3fQokUwdmyhvH9RKJT9rJg0qyHNahixGYzZLc36uTRoqVKlChEREVSuXJlXX32VuLi4ouryWNedTZJ1RUICPPooLF+e87ly5SAmBiZMKNT1UXI0v/uu03sUeZIC72c3kGY1pFkNIzaDMbulWT/dvxm9/j6F4Ovry9tvv81zzz3HmDFj2LBhQ5HFeaLg4OD8f3FMjDYZ9tChnM89+KA2r+Xpp/P//rkoULObSLMa0qyGNKtjxG5p1k/3Oi1ZWY4XGfXu3Zv77ruPf//734UedcexWrVTPpMna3NZbtWvn3a65q671LcJIYQQBqF70BIVFYW/v7/DY40bN+bTTz/l9OnThR7mqaxWK96unLq5ckW7d9DXX+d8zscH3nsPRo4s0smwLjd7AGlWQ5rVkGZ1jNgtzfrp3mJQUJDTwMDAQBo1alSoUZ7MpQV14uK0uyc7G7BUqgRbt2o3PSziq3dk4SI1pFkNaVbDiM1gzG5p1i/PxeWef/55PvnkE/z9/endu7d9XoszK1asKJJAVxX14nK6R5dr1kBoqLZs/q0eeUR7vlq1Qu9zRkbxakizGtKshhGbwZjd0qzRs7hcnqeHXnrpJfz8/AAYNGhQocYZ1W3/kDIzYdIkx/VQsnv5ZYiIAIUL8xjtLwNIsyrSrIY0q2PEbmnWL89ByzPPPOP0/4tc/PEH9O3r/N4+JUpog5VXXlHfJYQQQtwB8hwqJSUl6f74p8j1e923T7uc2dmApWpV+PFHtw1YjPjnI81qSLMa0qyOEbulWb88j7QMGDDAPo/l1kueQVu7JSsrCy8vL7Zu3Vo0hR6mePHiOR9ctkwbkPz1V87n2rSBlSuhcuWij8uF02YPJ81qSLMa0qyOEbulWb88By1RUVGqOgyjYsWKNz+5fh3GjIEFC5y/eORImDNHu1mhGzk0G4Q0qyHNakizOkbslmb98hy0BAUFqeownuRk6N0bduzI+Zyfn3bn5AED1HcJIYQQdyjdi8uBdq+BI0eOcPHiRW7cuOHw3NP5WH4+KiqK6Oho0tLSaNq0KaNHjyYwMNDpa/v06cOFCxccHps2bRqtWrVyebsFYbFY8N23D3r2BGfn9GrWhLVrwYPWrrFYLIa79bk0qyHNakizOkbslmb9dA9aTCYTkyZNIjU1FYvFQunSpbl69Sq+vr6ULVvW5UFLTEwMkZGRjB8/nuDgYCIiIpg6dSrz58/P9WuGDRvGU089Zf+8dOnSLm2zMJyYOZM606drp4Zu1b49fPkl5DLwcpcTJ05Qp04dd2e4RJrVkGY1pFkdI3ZLs366L7T+73//S7Nmzfj222/x9fXlo48+Yvny5dSpU4ehQ4e6vOG1a9fSs2dP2rRpQ0hICGPHjuXgwYOYTKZcv6ZUqVIEBgbaP0qUKOHydvPtr7/glVeoM2WK8wHL+PGwYYPHDVgAw/1lAGlWRZrVkGZ1jNgtzfq5dKRl7Nix+Pj44OPjQ0ZGBsHBwQwdOpSpU6fyxBNP6N5oRkYGCQkJvPrqq/bHgoODCQoKIi4ujpCQEKdf99lnn7Fo0SKqVKlC165d81w75tw5SE11fMzPD8qXd/76P/6Aa9ecP1ct6wz06gU//+zw+J/4cemu6hAeDh06wPmbzwUG5n7/w7NnnT+e775cFtb980+4dMn5c9InfdInfdInfZ7Ud/Wq8+ez0z1oKVmyJNf/PsJQvnx5zp49S82aNfHy8uJSbt9ZLlJTU7FarZQrV87h8YCAAC5fvuz0a3r16sX999+Pn58fe/fu5f333yczM5NOnTo5fX3dus7eA1atct40dCisXu38uayKTSElJcfj64NfoXfSfBiS82tWroTnnnP+ftWrO38833253Ihh/XptrrAz0id90id90id9nt53K92nhxo0aMC+ffsAaNmyJQsWLOCDDz5g+vTpPPjgg3rfBnC+5svt9OrViwYNGhASEsLzzz9Pnz59WJ3ndzkCCPv7IwY4TFaW1f5sUlISKdkGIpmZFuDWG0AlAmZo1w6ANOBH21Ndu8KMdwHT38/YmP/+upvi4+OxWCzZHkkBsk/itRas729paWlOTq+ZgDgP78u5/1JSbjZ6Yl92tr7ERNvjntl3U/a+RA/vI0dfYmKiR/dBzv2XmnrMo/uc77+dHt7nfP/d/HvomX2ObH22Jk/ts7nZZ9vPBe3btGkT+/aNBEKBCdxOnjdMzM5sNmOxWAgKCiIzM5OoqCji4+MJDg7mhRdewN/fX8/bANrpoQ4dOjB79myaNm1qf7xv37707duXLl263PY9YmNjmT59OjExMQ6P226YGBd3hTJlHG+YmO/DZ4F/wqOPwsGDmIFy06bBhAn8+Ze3IQ7vXb5sJiDA8aiWJ/U54+VlpmrVck6f84Q+Z/vPbDZTrlw5j+2zyd6X/WfDE/uys/XZ9rOn9jmTkGDG19f5z7Mn9Dnbf7afDU/ts7m1z/bz4al9Ntn7bv1vtKf13SowECyWnH8PC9p39WoqdevmfcNE3YOWwjZkyBCaNWvG4MGDATh//jz9+vVj0aJFuc5pyW7JkiVs27aNzz//3OHxIrvL84kT2hGXiAht/ooQQgghCo2euzy7dJvGK1eusGvXLjZu3MiGDRscPlzVrVs31qxZw44dOzCZTMyZM4eGDRsSEhJCfHw8AwcOtB8+Onz4MNHR0SQkJHDu3Dm++eYbli9fTvfu3V3ebr7dey8cPSoDFiGEct99B3/fUeWOVa0a2P4NeuqU9v3mcTHpbT3+OEyaVAhhubi18fPPcz9qIQqP7om4W7ZsYe7cuXh7e1O2bFn7PYlAuwdRx44dXdpwx44dMZvNhIeH2xeXGzNmDKAtWnPmzBkyMzMB7R4HW7Zs4dNPP8VqtRIcHMzrr7+e6yTcIlOsGGlpaW5ZH6YgpFkNaVbDnc2PP67d+xS0w/f33QcTJ+Y+KdFG9rNrqleH8+dBz0rxN25od0rZtk3787F1R0eDylUxnn8e8vsrSX4+9NM9aFm0aBEDBgygX79+eHu7dIAmV/3796d///45Hm/UqBHbtm2zf167dm0++uijQtlmQSUnJ+s6feVJpFkNaVbD3c2jRsG4cdo5+FWroG9fqF0b8roewd3N+ZGfZosFCmORVB8fyO9dZGzdqpfM8vPTPvLjn/LzURh0jz4sFgtPPPFEoQ1YjMpoP1ggzapIsxrubi5VSvuFes89MHYs+PvDDz/cfP6bb6B5cyhTBoKD4bXXoEqVm81TpkCrVtr0uCpVoEIF7X2yX1R56BA89BCULKm99tSpnB1z5mhHJHx9te1lX0bKdqpi+XKts3RpGD4cMjPh7be1CZHVqmk3qM9NSEgIXl6waBG0bq21PPSQ1nbr9zJvHlStqj0P2iTO11/XjpQEBEDnzo7fQ0YGDBmidVWvDpGRjtt2dnpoyxZo1kzrqFxZ269ap/a/Tzyhfc1//qM9cOvpoePHtUXL/fygUiV4803tKI1NzZowd6521KxUKahTB77/Pvf9c6tbTw+Fhmq3n5s0SZu4GhwM77/v+DUnTsCzz0KjRiEEB8O//qXtOyNw199D3SOQrl27smnTpqJsEUIIw7BaIToazGbH0xB//aWdMjpwQBs0bNsGU6c6fu3Bg/DLL9ovxcWLtfUp163TnsvMhB494O67Ye9e7Wbxb7/t+PVRUdqAYeZM2L8fGjaEjh1v2SwAAAAgAElEQVQdF9T84w/tdd9+CytWaPdw7dBB6961C4YNg1decboElYN33oERI+DXX7UBUPfuWqPN/v3w00+webO2fgdo624cPw4xMdpzlSppv5xtX/fuu1pXdLT2fX/6qdabm7g47dTLU0/Bvn3a+9oWZN29W/vfNWu0U0rO7gSTmamtUuHrqw3uliyBpUth9mzH182erXXu368N1AYM0AZY+fXNN9oC6rt3a39eo0drf/agve/TT2unGPfuha+/1n4mRo/O//b+CXSfHgoNDWXcuHEMHjyYmjVr4uPj4/D8hAm3v75aCCGMbvZsbZBhsWj/Uq9e3XFOS/YFue69FyZPhgkTHH9BFisGCxdqRw3q1NGOEvz4o/YLc/NmbUXvn3+GcuWgXj3tl9qsWTe/fsEC7UiG7ez6hx9qv8gjI7XHQetbtEg7KlG/vraNs2dh+nTt+bfe0gYPu3dr283Niy/e/P4WL9aOGMTEaEdPQDu6sXixdtQEtKMkK1ZAcrLWD9r3GhCgDZZatdJ6p07VjnwAfPzxzUGIM7Nmaa+1tQM0aaL9b4UK2v8GBuZ+SmnLFjh5EmJjtdc1aKBt/+23tT8bm549YeBA7f9Pnartv2PHtP2XH9Wr3/xzq11bO5Kzfbs2yFy+XDtKl/3oy7x52p9TRIR2ikzkpPtIywcffMCBAweoUKECxYsXty/nb/v4p8jr3kieSprVkGY13N38yivav8S3btVOhyxcePMXJ2hHBbp3146UlCkDL70EZ844rixx333agMUmKAguXtT+/9Gj2imP7EtgPPKIY8PRo9opIZtixbSWo0dvPlaxojZgsalcWRsA2fj4aKeJcjvSYtvP2bft7w/33++4nfvuuzlgATh8WDu6UL269njp0tr3cu2adjrkyhXte83+vg88oO2r3Pz2m3a6Rw9nPx9Hj2qd2ee5PPoo/P6741okDRrc/P+2AZDtzyU/bh3sZP9zPnRIOxpXujSUKmWldGltVY2MDG3Q6unc9fdQ95GWmJgYpk2bRrNmzYqyx+MF5Xd2mBtJsxrSrIa7m8uV0wYVISHwxRfQsqX2C8iW1aWL9i/pL77QTots3w5Dhjher1y8uON7enndPHWSlVU4lzc724azx6xWnLLt59u13LoIWlqaNm9k//6cr61U6eb2XPkeXVlE3dnPh96vz75/bH257R9X38/2nrb3S0uDNm20QW96+jWHdUmqVMn/NlVx199D3UdaAgICqJx92P4PZbTL0kCaVZFmNTypuXZt7QjAf/6jff7775CQoM0Dad1aOyqRnOzae95/vzYfJPtt2H75JedrbHM5QDtNtWePdsSisNj2c/YJvqmp2umS++/P/esefFCbTHrt2s3Bne2jbFntNFGlSo7ve/Ro3jfLa9DAcbJzdj4+4O19c9Dn7OfjgQe0fZr9qMquXdrRKNVXGdk8+CAcOaJN3n3wwVIO++nWwY4nctffQ92DlpdffpnFixdz5cqVouwRQghD+de/tImk589rR2HKldPmQpw4oc3tWLjQtfd7+mntX9ovv6ydalq9Wps4mt3Ikdq8kKgo7Rffa69pg4QBAwrv+7JZskRriI/XTo1VrgzPPJP76x94QJtI3KcPbNqkzSXZvl27esk22XboUG3OyHffaadIhg1zPF12q3HjtLk+Eydq3++BA9q8D9COXlSvrk1qvnhRO4Jxq/bttUnEoaHaqaaYGG2u0ahR+d4tBda/vzaB+/nntUGpyaRNTv57uTKRC92Dlo8//pg9e/bQo0cPevbsyfPPP+/w8U9hNptv/yIPI81qSLMantb82GPaEZf33tP+1f/FF9ov2Hr1tAHLv//t2vv5+GhX1Zw8CY0baxM1p0xxfE3fvtov3bFjtX+xHzwIGzZoRzIKi20/T5miNTRqpB2tiI7W5tDk5YsvtIHNoEHaICY0VJvnYjuVNGGCdiVT167aVU8DB+Z+XxqAunW1X+gbN2rf79NPO86rmT1b22aVKjBkiCXH13t7a1fnXLsGDz+sTS4eOFDbf+5Spox29KhECXjqKSsPPqhdHm2EU0Pgvr+Huu89tHHjxjyffyavobdCRXbvob8lJiZSo0aNQn/foiTNakizGtKsRmJiIjVr1mDLFu1SY6Mw6r6WZn33HtI1ETczM5O77rqLhg0bEhAQUKiRRmO0HyyQZlWkWQ1pVsOIzWDMbmnWT9fpIR8fH2bMmMFff/1V1D1CCCGEEE7pvuS5Xr16HDt2zO2XGwohhFDDlUuNhVBB90Tctm3b8sEHHxAZGcmuXbv49ddfHT7+KeLj492d4DJpVkOa1ZBmNYzYDMbslmb9dE/Ebdu2be5v4uXF1q1bCy2qIIp6Iq7FYsG3MG5jqpA0qyHNakizGkZsBmN2S7Om0CbiAnzvyu0u72BG+8ECaVZFmtWQZjWM2AzG7JZm/XSfHhJCCCGEcCeXBi3ff/89w4YNo3PnznTu3JnXXnvtH3cEJuV293H3QNKshjSrIc1qGLEZjNktzfrpPj20fPlylixZQo8ePRjw91rRhw8fZs6cOfz+++/0zn4/9jvY9evX3Z3gMmlWQ5rVkGY1jNgMxuyWZv10T8Tt06cPQ4YMyTEhd+vWrXzyySesWLGiSAJdVdQTcYUQQghR+PRMxNV9eujSpUuEhITkePy+++7jcvbbkQohhBBCFAHdg5batWuzatUqrFar/TGr1crKlSu57777iiTOE2X//o1CmtWQZjWkWQ0jNoMxu6VZP92Dltdff50dO3bQp08fJk2axKRJk+jTpw87d+5k+PDhRdnoUWQRIDWkWQ1pVkOa1TFitzTrp3tOC8C1a9fYsmULZ86cAaB69eq0a9cOPz+/Igt0VVHPabFarXh7G+tKcWlWQ5rVkGY1jNgMxuyWZk2hLi4H4OfnR5cuXQolDiAqKoro6GjS0tJo2rQpo0ePJjAwMM+vuXDhAoMHD8bPz49Vq1YVWoteRvvBAmlWRZrVkGY1jNgMxuyWZv1cGrScPHmSAwcOcPny5RznswYNGuTShmNiYoiMjGT8+PEEBwcTERHB1KlTmT9/fq5fk5WVxcyZM6lbty4nT550aXtCCCGEMDbdQ6VVq1YxePBgoqOj2b9/P4cOHbJ//Pbbby5veO3atfTs2ZM2bdoQEhLC2LFjOXjwICaTKc+GMmXK5HkfpKKWlJTktm3nlzSrIc1qSLMaRmwGY3ZLs366j7SsXLmSsLAwOnfuXOCNZmRkkJCQwKuvvmp/LDg4mKCgIOLi4pxeWn3q1CnWrFnDxx9/zE8//VTghvwqXry427adX9KshjSrIc1qGLEZjNktzfrpPtKSkZFB48aNC2WjqampWK1WypUr5/B4QECA0zVfbty4wYwZMxg2bFiOr1GtYsWKbt1+fkizGtKshjSrYcRmMGa3NOune9DStWtXNmzYUCgbzcrSfcESAJGRkVSrVo3HH3+8ULYvhBBCCOPRPWhJTk7m66+/ZtCgQUybNo0ZM2Y4fLjC398fb29vzGazw+OXL18mICAgx+sPHDjAjz/+yJNPPsmTTz5pv9/Rk08+yS+//OJ0GyNGjCAsLIywsDBiYmI4fPiww+ThpKQkhxs+WSyWHNedJyYmOjSmpaURFxfn8BqTyURaWpr9c7PZTGJiosNr4uPjsVgs9s9TUlIczgdardZC67t1TpDJZOKPP/7w6D5n+y97jyf2ZWfrszV6ap9N9j6LxeLRfZBz/1ksFo/ug5z7LzU11aP7nO2/Y8eOeXRfbvsv+7Y8sS87W59t+57aZ5O9z9ZU0L5NmzYxcuRIQkNDmTBhAreje52WWbNm5fn8uHHj9LyN3ZAhQ2jWrBmDBw8G4Pz58/Tr149FixblmNNy/vx5/vrrL/vnO3fuJDo6mrlz5xIUFOSwTkxRr9MSHx9PnTp1Cv19i5I0qyHNakizGkZsBmN2S7NGzzotLi0uV5g2bNhAREQE48ePp0qVKnz44YdkZmYyf/584uPjeffdd5k7d67T82YbN27k008/dbpOi9wwUQghhDCeQl9crjB17NgRs9lMeHi4fXG5MWPGANphpzNnzpCZmemuPCGEEEJ4mHwdaQkLC2Py5Mn4+/sXRVOByJEWIYQQwnj0HGnJ1zq8+/fv5/r16wWKM6pbJ0kZgTSrIc1qSLMaRmwGY3ZLs37Gu+GBmxnx6I00qyHNakizGkZsBmN2S7N+MmhxkbsXt8sPaVZDmtWQZjWM2AzG7JZm/WTQIoQQQghDkEGLi7IvtGMU0qyGNKshzWoYsRmM2S3N+smgxUXJycnuTnCZNKshzWpIsxpGbAZjdkuzfjJocZGzO1B7OmlWQ5rVkGY1jNgMxuyWZv3yNWh59tlnHZbOF0IIIYQoavlaEfeNN94o7A4hhBBCiDzpPtIya9Ys/vzzzxyPX7t27bY3U7yT3HoXTCOQZjWkWQ1pVsOIzWDMbmnWT/egZfPmzQ63n7axWCxs2bKlUKM8WVBQkLsTXCbNakizGtKshhGbwZjd0qzfbU8PHThwAICsrCwOHz5MmTJl7M9ZrVb27dvn9E7Md6rSpUu7O8Fl0qyGNKshzWoYsRmM2S3N+t120GKbv+Ll5cU777zj8JyXlxeVKlVi6NChRVMnhBBCCPG32w5abKd+BgwYwAcffOBwZ2cfH5+iK/NQZrPZcEsuS7Ma0qyGNKthxGYwZrc063fbOS0+Pj74+Pjw5ZdfEhgYaP/8nzhgAe3W2UYjzWpIsxrSrIYRm8GY3dKsn9e2bduycnvys88+o1+/fpQsWZLPPvsszzcaNGhQocflR3p6Op07d+bKlSuGvHOmEEII8U+UmpqKv78/69ato1SpUk5fk+fpoUOHDnHjxg37/8+Nl5dXATKFEEIIIW4vz0HLvHnznP5/IYQQQgjV5N5DLoqPj3d3gsukWQ1pVkOa1TBiMxizW5r1072Mf1paGsuWLePAgQOYzWayshynwqxYsaLQ4zzRvffe6+4El0mzGtKshjSrYcRmMGa3NOune9Aya9YsTpw4QadOnQgMDCzKJo/m6+vr7gSXSbMa0qyGNKthxGYwZrc066d70PLrr78yb948ateuXZQ9QgghhBBO6Z7TUrFixRynhP6JUlJS3J3gMmlWQ5rVkGY1jNgMxuyWZv10D1pGjBjBwoULOXr0KBkZGVitVoePf4rr16+7O8Fl0qyGNKshzWoYsRmM2S3N+uW5uFx2bdu21b4glzVZtm7d6vLGo6KiiI6OJi0tjaZNmzJ69Gin82WuXbvGlClTSEhIIDU1lcDAQNq1a0doaGiOlXllcTkhhBDCeAq8uFx277//fqGFAcTExBAZGcn48eMJDg4mIiKCqVOnMn/+fKevb968OYMGDcLf359Tp04xZ84cSpQowQsvvFCoXUIIIYTwTLoHLY0aNSrUDa9du5aePXvSpk0bAMaOHUv//v0xmUyEhIQ4vNbPz4/u3bvbPw8KCuLJJ5/kt99+K9QmPaxWK97exlreRprVkGY1pFkNIzaDMbulWT/dW/z111/z/HBFRkYGCQkJNG7c2P5YcHAwQUFBxMXF3fbrk5KS+Pnnn2nQoIFL2y0MsgiQGtKshjSrIc3qGLFbmvXTfaRlzJgxTh+3zXFxZU5LamoqVqs1x22tAwICuHz5cq5fN23aNGJjY8nIyODZZ5+lf//+urdZWOrUqaN8mwUlzWpIsxrSrIYRm8GY3dKsn+5By/fff+/weWZmJgkJCXzyyScMGDDApY3m99Lp119/nRdffJGEhAQ+/vhj1qxZQ69evfL1XvlltEN4IM2qSLMa0qyGEZvBmN3S7MJ28/uFPj4+1K5dm1deecXlmyn6+/vj7e2N2Wx2ePzy5csEBATk+nWBgYHcfffdPPHEE7z88stERkbm+toRI0YQFhZGWFgYMTExHD582OHS7KSkJIfrzC0WS47DXYmJiQ6NaWlpmEwmh9eYTCbS0tLsn5vNZhITEx1eEx8fj8VisX+ekpJCUlKS/XOr1Sp90id90id90veP69u0aRMjR44kNDSUCRMmcDu6L3nOzdGjRxk9ejTr1q1z6euGDBlCs2bNGDx4MADnz5+nX79+LFq0KMdEXGc2b97MRx99xNq1ax0eL+pLnpOSkggODi709y1K0qyGNKshzWoYsRmM2S3NmkK95HnDhg0On2dlZWE2m4mJiaFJkyYux3Xr1o2IiAhq165NlSpV+PDDD2nYsCEhISHEx8fz7rvvMnfuXCpWrMjBgwc5ffo09erVo2TJkhw/fpzFixfb145RqXjx4sq3WVDSrIY0qyHNahixGYzZLc366R603HoqxtvbG39/f1q3bu3ynBaAjh07YjabCQ8Pty8uZ5vsa7FYOHPmDJmZmQCUKFGCzZs38/HHH3Pjxg0qV65Mly5deP75513ebkFVrFhR+TYLSprVkGY1pFkNIzaDMbulWb8Cnx7yNLIirhBCCGE8ek4PGW/Ksptln3BkFNKshjSrIc1qGLEZjNktzfrJoMVFJ06ccHeCy6RZDWlWQ5rVMGIzGLNbmvWT00NCCCGEcDs5PSSEEEKIO4YMWoQQQghhCDJocdGtqwEagTSrIc1qSLMaRmwGY3ZLs34yaHGREefJSLMa0qyGNKthxGYwZrc06yeDFhfdemdqI5BmNaRZDWlWw4jNYMxuadZPBi1CCCGEMAQZtLgo+x0vjUKa1ZBmNaRZDSM2gzG7pVk/GbS4KDk52d0JLpNmNaRZDWlWw4jNYMxuadZPFpcTQgghhNvJ4nJCCCGEuGPIoEUIIYQQhiCDFheZTCZ3J7hMmtWQZjWkWQ0jNoMxu6VZPxm0uCgoKMjdCS6TZjWkWQ1pVsOIzWDMbmnWTwYtLipdurS7E1wmzWpIsxrSrIYRm8GY3dKsnwxahBBCCGEIMmhxkdlsdneCy6RZDWlWQ5rVMGIzGLNbmvWTQYuLUlNT3Z3gMmlWQ5rVkGY1jNgMxuyWZv1kcTkhhBBCuJ0sLieEEEKIO4YMWoQQQghhCDJocVF8fLy7E1wmzWpIsxrSrIYRm8GY3dKsXzG3bPVvUVFRREdHk5aWRtOmTRk9ejSBgYE5XpecnMySJUvYt28fZrOZoKAgevToQdeuXZU333vvvcq3WVDSrIY0qyHNahixGYzZLc36ue1IS0xMDJGRkYwYMYKIiAjS09OZOnWq09eePn0ab29v3nzzTf73v/8xYMAAPvroIzZt2qS4Gnx9fZVvs6CkWQ1pVkOa1TBiMxizW5r1c9uRlrVr19KzZ0/atGkDwNixY+nfvz8mk4mQkBCH1z7yyCM88sgj9s+Dg4M5ePAgO3fu5Omnn1baLYQQQgj3cMuRloyMDBISEmjcuLH9seDgYIKCgoiLi9P1HleuXKFMmTJFlZirlJQU5dssKGlWQ5rVkGY1jNgMxuyWZv3cMmhJTU3FarVSrlw5h8cDAgK4fPnybb8+Li6O3bt306FDh6JKzNX169eVb7OgpFkNaVZDmtUwYjMYs1ua9XPLoCUrK//r2Z05c4ZJkyYRGhpK/fr1C7FKn+DgYOXbLChpVkOa1ZBmNYzYDMbslmb93DJo8ff3x9vbO8e9Cy5fvkxAQECuX5eUlMTo0aPp0KED/fr1y3MbI0aMICwsjLCwMGJiYjh8+DBWq9XhvbIf3rJYLDku4UpMTHRoTEtLw2QyObzGZDKRlpZm/9xsNpOYmOjwmvj4eCwWi/3zlJQUkpKS7J9brVbpkz7pkz7pk75/XN+mTZsYOXIkoaGhTJgwgdtx2zL+Q4YMoVmzZgwePBiA8+fP069fPxYtWpRjIi7AhQsXGDlyJC1atGDEiBG5vm9RL+NvtVrx9jbW8jbSrIY0qyHNahixGYzZLc0aj17Gv1u3bqxZs4YdO3ZgMpmYM2cODRs2JCQkhPj4eAYOHGgfiaWkpBAWFkatWrXo378/ly5d4tKlS265YZMsAqSGNKshzWpIszpG7JZm/dx2yXPHjh0xm82Eh4fbF5cbM2YMoB3KOnPmDJmZmQDs3buXpKQkkpKS+L//+z/7ezz44IOEh4cr7a5Tp47S7RUGaVZDmtWQZjWM2AzG7JZm/eQuz0IIIYRwO48+PSSEEEII4QoZtLgo+6xoo5BmNaRZDWlWw4jNYMxuadZPBi0uKl68uLsTXCbNakizGtKshhGbwZjd0qyfDFpcVLFiRXcnuEya1ZBmNaRZDSM2gzG7pVk/GbQIIYQQwhBk0OKi7Cv/GYU0qyHNakizGkZsBmN2S7N+Mmhx0YkTJ9yd4DJpVkOa1ZBmNYzYDMbslmb9ZJ0WIYQQQridrNMihBBCiDuGDFqEEEIIYQgyaHHRrbflNgJpVkOa1ZBmNYzYDMbslmb9ZNDiIiPOk5FmNaRZDWlWw4jNYMxuadZPBi0uKleunLsTXCbNakizGtKshhGbwZjd0qyfDFqEEEIIYQgyaHFRWlqauxNcJs1qSLMa0qyGEZvBmN3SrJ8MWlyUnJzs7gSXSbMa0qyGNKthxGYwZrc06yeLywkhhBDC7WRxOSGEEELcMWTQIoQQQghDkEGLi0wmk7sTXCbNakizGtKshhGbwZjd0qyfDFpcFBQU5O4El0mzGtKshjSrYcRmMGa3NOsngxYXlS5d2t0JLpNmNaRZDWlWw4jNYMxuadZPBi1CCCGEMAQZtLjIbDa7O8Fl0qyGNKshzWoYsRmM2S3N+rl10BIVFUWvXr145plnmDhxIpcuXcr1tQsWLGDw4ME8+eSTTJ8+XWGlo9TUVLdtO7+kWQ1pVkOa1TBiMxizW5r1c9ugJSYmhsjISEaMGEFERATp6elMnTo119d7eXnRtWtXmjZtqrAypxo1arh1+/khzWpIsxrSrIYRm8GY3dKsXzG3bBVYu3YtPXv2pE2bNgCMHTuW/v37YzKZCAkJyfH64cOHAxAXF0dmZqbSViGEEEK4n1uOtGRkZJCQkEDjxo3tjwUHBxMUFERcXJw7koQQQgjh4dwyaElNTcVqtVKuXDmHxwMCArh8+bI7knSLj493d4LLpFkNaVZDmtUwYjMYs1ua9XPLoCUry7j3aLz33nvdneAyaVZDmtWQZjWM2AzG7JZm/dwyaPH398fb2zvHJVOXL18mICCgULYxYsQIwsLCCAsLIyYmhsOHD2O1Wu3PJyUlkZKSYv/cYrHkGDkmJiY6NKalpXHmzBmH15hMJtLS0uyfm81mEhMTHV4THx+PxWKxf56SkkJSUpL9c6vVWmh9ty6tbDKZuH79ukf3Odt/vr6+Ht2Xna3P1uypfTbZ+3x9fT26D3LuP19fX4/ug5z7z/Y+ntrnbP8lJyd7dF9u+y/7fzs8sS87W5+t2VP7bLL32ZoL2rdp0yZGjhxJaGgoEyZM4Ha8tm3b5pbDHkOGDKFZs2YMHjwYgPPnz9OvXz8WLVrkdCKuzcyZM8nMzGTixIlOn09PT6dz585cuXKFsmXLFkm7EEIIIQpXamoq/v7+rFu3jlKlSjl9jdsuee7WrRtr1qxhx44dmEwm5syZQ8OGDQkJCSE+Pp6BAwc6jBTPnTuHyWTi6tWrXL16FZPJxKlTp5R33/qvJyOQZjWkWQ1pVsOIzWDMbmnWz22XPHfs2BGz2Ux4eDhpaWk0bdqUMWPGANqhrDNnzjhc2jxnzhwOHDhg//ynn36icuXKLF++XGl39lMtRiHNakizGtKshhGbwZjd0qyf204PFRU5PSSEEEIYj0efHhJCCCGEcIUMWlyUfQa0UUizGtKshjSrYcRmMGa3NOsngxYXySJAakizGtKshjSrY8RuadZP5rS4yGq14u1trLGeNKshzWpIsxpGbAZjdkuzRua0FAGj/WCBNKsizWpIsxpGbAZjdkuzC9t1y1aFEEIIIVwkgxYXZV+e2CikWQ1pVkOa1TBiMxizW5r1k0GLi4oXL+7uBJdJsxrSrIY0q2HEZjBmtzTrJ4MWF1WsWNHdCS6TZjWkWQ1pVsOIzWDMbmnWTwYtQgghhDAEGbS4KPstuI1CmtWQZjWkWQ0jNoMxu6VZPxm0uOjEiRPuTnCZNKshzWpIsxpGbAZjdkuzfrK4nBBCCCHcThaXE0IIIcQdQwYtQgghhDAEGbS4KDEx0d0JLpNmNaRZDWlWw4jNYMxuadZPBi0ucmWezJQpU/Dy8rJ/lC9fnlatWrF+/XqXtztlyhRKlizp8tft37+fDz/8kLS0NJe/Ni/jxo0jODgYb29vunXrluP5jIwM/Pz8GD16dI7nKleuTLFixbh69arD4wsWLMDLy4uTJ08qmY/01VdfERERUWjvV1jN8+bNo2bNmhQrVoxGjRoVynvmxojzvqRZDSM2gzG7pVk/GbS4qFy5ci69vkSJEuzatYtdu3bx2Wef4e3tTefOndmyZYtL7/Pyyy+zfft2l74GtEHL7NmzC3XQsmHDBmbPns2YMWPYuXMns2fPzvGaEiVK8Mgjj7Bjxw6Hx48dO8bFixfx9fVl165dDs/FxsZStWpV7rnnHpf3c34U9qClMJoPHz5MWFgYvXv3Zvv27URGRhZCWe5U7OfCJs1qGLEZjNktzfoVc8tW/0G8vLxo3ry5/fMnnniCu+++mwULFtCuXTvd71OtWjWqVatWFIkui4+PB2DUqFF53umzVatWzJ49m/T0dPtM8NjYWGrVqkXt2rWJjY2lffv29tfv3LmTNm3aFLjv2rVr+Pn5Ffh9VLI1HzlyBIChQ4dy7733Ftr7CiHEnUCOtLiooEcsypYtS+3atR2ucT979ix9+/YlMDAQPz8/mjdvnuNIzK2nh3744Qe8vLzYtGkToaGhlC1blqpVqzJ+/HgyMzMB+Pzzz3nppZcAqFKlCl5eXtSsWTPPvri4ODp37kzZsmUpVaoUTz31FHv37rU///jjjzNmzBgAfHx88PLy4ocffnD6Xq1bt+bGjRvs3r3b/lhsbCwtW7akZcuWDkdhTpw4QQHaCuoAACAASURBVFJSEq1atQK0/ZyVlcXs2bMJCQnB19eXmjVr8s4773Djxo0c+2HdunX069ePgIAAOnbsCMCuXbt44oknCAgIoFSpUtx///28++67AISGhrJkyRKOHj1qP30XGhqa634JDQ3lgQceYOvWrTRq1IiSJUtSv359Nm7caH+N7Wdj48aNtGzZkrvuuoty5crxwgsv8Pvvv9tfd+rUKby8vPjss894/fXXqVixInXr1iU0NJRevXoBUKtWLby8vPj8888BuHz5MkOHDqVy5cqULFmSBx98kOXLlzs02n5G9u3bR+vWrfHz82PWrFmANnieMWMGkydPJigoiDJlyjBw4EB+//13Dh8+zOOPP06pUqWoX78+33//vcP7Llu2jMcee4wKFSrg7+9P8+bNiYmJcXjN559/jpeXF7/++itdu3alVKlS3HPPPcybNy/Hvjx27Bi9e/emQoUK+Pn5UadOHebPn+/wmi+++ILGjRtTsmRJKleuzPDhw7l27ZrDfjYSaVbHiN3SrJ8MWlyUnJxcoK/PzMzkzJkzVK1aFdD+4B977DFiY2OZN28eq1atonz58nTo0IFt27bd9v2GDRtGcHAw0dHRvPTSS8ycOZPPPvsMgE6dOjFp0iQA1q9fz65du1i7dm2u73X27Flat27NqVOnWLx4McuWLbP32Y6ufPjhh7z22msA9tNeTZo0cfp+LVq0wMfHh9jYWPtjsbGxtGrVihYtWvDzzz9z/fp1++OgDXRA28/jxo1j/Pjx9OrVi2+//ZbBgwczc+ZMhgwZkmNbQ4cOJTg4mNWrVzNp0iSuXr1Kx44dKVmyJF988QXffvsto0ePJj09HYC3336bjh07UqNGDfv38fbbb+e5ry9evMigQYMYPnw4q1evJjg4mC5duhAXF2dv/uqrr+jUqRO1atVi9erVfPDBB8TGxjqd9zNp0iSuXbvGsmXLiIiI4O2332batGkAREdHs2vXLjp16kRmZiYdO3ZkxYoVTJ48ma+++opGjRrRt29fli5d6vCeN27coHfv3vTu3ZuYmBi6dOlif+6jjz7i2LFjfP755/znP/9h+fLlvPbaa/Tu3Zu+ffsSHR1NhQoV6NGjB6mpqfavO3XqFH369OHLL79k1apVNGvWjE6dOrF169Yc31P//v1p2bIlX3/9Ne3atSMsLMxhAJ6QkECzZs04dOgQc+fOZf369YwYMYKzZ8/aX7NgwQIGDhzIY489xjfffMO0adP48ssvefnll+372WikWR0jdkuzfnJ6yEUhISEuf43tyEBKSgrTpk0jOTnZ/svpf//7HydOnGDfvn32SZcdO3akfv36vPPOOznmhNyqS5cuzJgxA4CnnnqK7777jpUrV/LKK69QsWJFatWqBUCTJk0ICgrK873mzZvHn3/+yXfffWd/bbt27ahRowYzZswgMjKSunXrUr16dQCH017OlC1blgYNGtgHJBcvXuT48eO0bNmSmjVrkpGRwa+//kqzZs2IjY0lICCA+vXrA9r50vnz5zNixAhmzpwJQPv27SlWrBgTJ05kwoQJDn8WzzzzDO+995798z179nD58mVmzZpFw4YNAWjbtq39+Vq1alGxYkVKlix52+/Dxmw288UXX9ChQwd7z7333sv06dP54osvqFWrFu3bt6dbt24Og4k6derQpEkTNm7cyDPPPGN/vG7duvYBpo3te2rcuLH9qNi3337Lrl27+Oqrr+jatav9+01OTmbSpEkMHDjQ/vWZmZlMnjyZAQMG5OivVKkSX375pf3rd+zYwapVq1i7dq19UFW1alUaNGjAli1b6NmzJ4B94AtgtVpp27YtR44cYeHChTz55JMO23j11VcZNWoUoJ0KXb9+PStXrrSfCp08eTLe3t7s2rWLgIAAwPHP5erVq0yaNIlRo0Yxd+5ch/YePXowceJE6tat6+yPx6Pl578b7mbEZjBmtzTrJ0daipjFYqF48eIUL16c4OBgli5dypQpU+z/atyxYwcPPPCAw1Ui3t7e9O7dm927d9uPROTm6aefdvi8bt26nDlzJl+tO3bs4PHHH3cY3JQuXZpnn302X5OAQTtysmvXLm7cuEFsbCyBgYHUqVOHu+66i0aNGtkHZbGxsbRo0cI+R+ann34iIyODPn36OLxf3759ycrKyjGY69y5s8PnISEh+Pv7M3ToUFauXFko/yooVaqUfcAC2mTjLl262E9/HT9+nJMnT9KnTx9u3Lhh/2jQoAGVKlXi559/zrM5Nzt27OCuu+5yOGoC2r44c+YMJ0+e1PW+t86hql27NoDDvCLbY9l/huLj4+nZsydBQUEUK1aM4sWLs3nzZo4dO5ZjG9l/Hn18fLj//vsd3uu7776jR48e9gHLrXbt2sXVq1d5/vnnHfZh27ZtycrKYs+ePU6/TgjxzyCDliJWokQJfvnlF/bs2cOJEye4cuUKkydPtj9/6dIlp0dAgoKCuHHjhsNhemduncFdokQJ/vrrr3y15tVy6dKlfL1n69atSU9PZ//+/faBiZeXFwAtW7YkNjaWP/74gyNHjthPDdlabNu+tSX78zaVK1d2+DwgIICtW7dSvnx5QkNDCQ4OplmzZvkefIHzW7FXrlyZpKQkQDuSBtC7d2/7QNX2cfHiRU6fPp1nc24uXbpEpUqV7PvNxtm+KFGiRK4DglsfL1GiBL6+vtx1110OjwH2n6GrV6/Svn17jh07xqxZs/jhhx/45ZdfeOaZZ5z+nN3u5/GPP/6wnxp1xrYPmzVr5rD//P39AXLsQyHEP4tbTw9FRUURHR1NWloaTZs2ZfTo0QQGBjp97aVLl3j//ffZs2cPpUqVonv37k4PgRc1k8nk0mExLy8vHnrooVyfDwwM5NChQzkeT05OplixYkqvhQ8MDHR6RCI5OTnXP5fbsQ1EYmNjiY2NtZ9yAG3OyxdffEFsbCxZWVkOgxbbL7rk5GRq1Kjh0GJrze7WX+gATZs25dtvvyUjI4PY2FgmTZpE586dOXv2bL72q+0XanYXLlygSpUqwM2JafPnz6dFixY5XluhQoXbNjsTGBjIhQsXyMrKcvgaZ/tC73vaZGXlfeux3bt3c/bsWaKjo3n44Yftj9smxbqqQoUKnDt3Ltfny5cvD8DKlSu55557cjwfHBzs8t9BTyDN6hixW5r1c9uRlpiYGCIjIxkxYgQRERGkp6czderUXF8/depUrl69SkREBKNGjSIqKooNGzYoLNbmZCQnJ3Px4sVCe8/WrVtz5MgR9u/fb3/MarWyatUqmjdvTvHixQv0/rf+y/l2LT/88IPDwCU9PZ1169bl+1LkoKAgatWqxaZNm9i3bx8tW7a0P9eyZUt+//13Fi9ejK+vr8Pgrl27dpQoUSLHFTIrVqzAy8vLYYBzOyVKlKBt27a89dZbXL161T7p09WjUunp6Q5XzWRkZPDNN9/Y58S0aNGCu+++m+PHj/PQQw/l+LjdlVu5ad26NdeuXeObb75xeHzFihVUr17d6S93vW43yPnzzz8BHK5cO3nyJDt37szX9p566imio6O5fPmy0+dbtGhB6dKlOX36tNN9GBwcfNu5WZ5ImtUxYrc06+e2Iy1r166lZ8+e9l+GY8eOpX///k5HbwkJCRw8eJClS5dSvXp1QkJCOH78ONHR0fbLW4vS4cOHeeONNxyugmjXrh3h4eEFnhT40ksvER4ezrPPPsv06dMpX748H3/8MUePHnV5ATpn6tSpA0BERATPPfccd911Fw0aNHD62jfeeIPPP/+cdu3a8c4771C8eHFmz57NtWvXmDBhQr4bWrduzZIlSyhevLjDv9arVq1KjRo1WL9+Pa1atcLX19f+XI0aNRg1ahTvvfceJUuWpG3btvz8889MnTqV0NDQ247w161bx6JFi+jevTs1a9bEbDYzY8YM7r77bvu8jTp16rB48WIiIyOpU6cOFSpUyHNgUa5cOYYOHcrkyZOpXLkyCxYs4MKFC0ycOBGAMmXKsGDBAnr16sWff/5Jly5dKFu2LGfPnmXz5s0MGTLEpcGWTceOHXn00UcJDQ1lxowZ3HvvvSxfvpyNGzeyZMkSl9/PFY8++ihlypThtddeY+LEiVy6dIl33nkn32sGTZkyhXXr1vHoo4/y1ltvcffdd2MymTh+/DizZ8+mbNmyvPvuu4wePZpz587x5JNPUrJkSU6dOsX69euZM2eOfXK5kZQuXdrdCS4zYjMYs1ua9XPLkZaMjAwSEhJo3Lix/THbv6Bsl49md+TIESpWrGi/agW0q2FOnjyJxWIp0tbDhw/TokWLHAOILVu28OijjzrtdUXp0qX58ccfadWqFSNHjqRnz56kpKQQExPDE088UaD3Bu0qlClTprBixQpatGjBs88+m+trq1Wrxo4dO7j77rsZNGgQ/fv3x8/Pjx9//NE++MmP1q1bk5WVxUMPPeQwMAHtaEtWVpZ9fZbsZs6cyYwZM1i5ciWdOnVi4cKFvPXWW3zyySe33eZ9992Hr68vkydP5plnnmHYsGHUrFmT7777jmLFtLH64MGDef7553njjTd4+OGHmTJlSp7vWalSJRYvXkx4eDg9evTg7NmzfP3119SrV8/+mq5du7JlyxZOnz7NCy+8QKdOnZg+fTply5bN92JxPj4+bNiwgd69ezN58mS6dOnC3r17iYqKcrhyqChUqlSJNWvWcOXKFbp3786///1vpkyZwmOPPZav96tVqxa7du2iTp06jBo1io4dOzJv3jyHv9v/+te/+PLLL9m9eze9e/eme/fuzJ8/n5CQEKfzioQQ/xxe27Zty/ukdhH4/fffee655/j0008d/kM+bNgwHn300Rz/IV62bBnbt293+GV16tQpXnrpJVasWEGlSpXsj6enp9O5c2euXLlSKPNB2rdvn+cRj3bt2rF58+YCb6comc1mwy0T7WnNoaGh7N69275irTOe1qyHNKshzeoYsVuaNampqfj7+7Nu3Tr7Kuq3csuRlttN/vMUFy9evO0pmi1bthTqHJeicLsrkDyRNKshzWpIszpG7JZm/dwyaPH398fb2xuz2ezw+OXLl51erlmuXLkcE/cuX76Mt7e3/VLIW40YMYKwsDDCwsKIiYnh8OHDWK1W+/NJSUkOV4NYLBb7qq82v/zyi67vJ/t6FWazOcctu+Pj4x1OY6WkpNgvkwVt4m1++hITEx32YVpaGiaTyeE1JpPJfkWGp/ZlXw7a1pf9iiFP7MvO1mdr9tQ+m+x9NWrU8Og+yLn/atSo4dF9kHP/BQUFeXSfs/1n256n9uW2/7L/t8MT+7Kz9dmaPbXPJnufrbmgfZs2bWLkyJGEhobqmjvpltNDAEOGDKFZs2YMHjwYgPPnz9OvXz8WLVrkdCLuyy+/TGRkpH0C4P/+9z927tzJ4sWLHV5bmKeHLl68qGstjQsXLjicohJCCCGEazz29BBAt27dWLNmDTt27MBkMjFnzhwaNmxISEgI8fHxDBw40D4Sq1WrFg0bNuS9997DZDIRGxvL6tWr6dGjR5E2VqpU6bZ3Ym7Xrp0MWIQQQggF3DZo6dixI/379yc8PJzXX3+dkiVL2leKtVgsnDlzxn63YtDuWVKqVCn+9a9/8f7779O3b18llzuHh4fnesTG39+f8PDwIm8oKGeHfT2dNKshzWpIszpG7JZm/dx2eqioFPbVQwBxcXGMGjXKYVJu+/btmTdvniFu3maxWHJcauzppFkNaVZDmtUxYrc0a/ScHpK7POtQt25dNm/ezMWLFzl79izVqlUz1Ckho/1lAGlWRZrVkGZ1jNgtzfrJoMUFlSpVMtRgRQghhLiTyF2eXeTspnmeTprVkGY1pFkNIzaDMbulWT8ZtLjo+vXr7k5wmTSrIc1qSLMaRmwGY3ZLs34yaHFRcHCwuxNcJs1qSLMa0qyGEZvBmN3SrJ8MWoQQQghhCDJocVH2pYiNQprVkGY1pFkNIzaDMbulWT8ZtLhIFgFSQ5rVkGY1pFkdI3ZLs34yaHHR6dOn3Z3gMmlWQ5rVkGY1jNgMxuyWZv1k0OKi7KviGoU0qyHNakizGkZsBmN2S7N+MmgRQgghhCHccSviZmVpt1JKTU0tkve/dOlSkb13UZFmNaRZDWlWw4jNYMxuadbY3s/2e9yZO+6GiSkpKfTu3dvdGUIIIYTIh5UrV1KxYkWnz91xgxar1coff/yBn58fXl5e7s4RQgghhA5ZWVlcu3aN8uXL4+3tfPbKHXd6yNvbO9cRmhBCCCE8V+nSpfN8XibiCiGEEMIQZNAihBBCCEOQQYsQQgghDOGOm9NSUFFRUURHR5OWlkbTpk0ZPXo0gYGBTl976dIl3n//ffbs2UOpUqXo3r07AwYMUFzsWvOCBQs4cOAAp06dom3btkycOFFxrUZvc3JyMkuWLGHfvn2YzWaCgoLo0aMHXbt29djma9euMWXKFBISEkhNTSUwMJB27doRGhqKj4+Px3Znd+HCBQYPHoyfnx+rVq1SVHqTK819+vThwoULDo9NmzaNVq1aqUi1c3U/r1mzhrVr13Lx4kUCAwMZPHgw7dq1U1isv3n//v288cYbOR739fVl48aNKlLtXNnPhw4dYuHChSQkJODn58cjjzzC66+/TpkyZTy2OS4ujoULF3L8+HHuuusunn32WQYOHKj0wpLt27fz1VdfcezYMdLT0/nuu+/y/G/XtWvXWLBgAdu3b6dYsWK0b9+eoUOHFsl/7+RISzYxMTFERkYyYsQIIiIiSE9PZ+rUqbm+furUqVy9epWIiAhGjRpFVFQUGzZsUFjserOXlxddu3aladOmCisdudJ8+vRpvL29efPNN/nf//7HgAED+Oijj9i0aZPHNgM0b96c6dOns3TpUkaNGsWGDRuIiopSWKxxtRu0GfwzZ86kbt26iiod5ad52LBhrFmzxv7xyCOPKKrVuNq8dOlSVq1axZAhQ1iyZAnvvPMOVatWVVjsWnO9evUc9u+aNWuoV68erVu39tjmP//8kwkTJnD//fezePFi/vOf/3D8+HEWLFjgsc0pKSmMHTuWevXqsWjRIsaNG8f69euV/8PBYrHQpEkT+vbtq+v14eHhxMXFMWfOHCZPnsy2bdtYsmRJkbTJoCWbtWvX0rNnT9q0aUNISAhjx47l4MGDmEymHK9NSEjg4MGDjBkzhpCQEFq3bk2vXr2Ijo722GaA4cOH06VLl9v+S7soudL8yCOP8Oabb9K0aVOCg4Np164d7dq1Y+fOnR7b7OfnR/fu3bn//vsJCgqiefPmPPnkk/z2229Km13ttlm1ahVlypShbdu2Cktvyk9zqVKlCAwMtH+UKFFCYbFrzVeuXGHZsmW89dZbtGnThipVqlC3bl3lg0RXmosXL+6wf2/cuEF8fDxPP/20xzafPn2atLQ0XnrpJapWrUrdunXp1KkTR48e9djm3bt3U7ZsWYYMGULVqlV5+OGH6du3L6tWrcpzwbXC1q5dOwYMGEC9evVu+9qrV6/y3XffMXz4cOrWrUuTJk0YNGgQX3/9NZmZmYXeJoOWv2VkZJCQkEDjxo3tjwUHBxMUFERcXFyO1x85coSKFStSvXp1+2NNmjTh5MmTWCwWj2z2BIXRfOXKFaWHdwvanJSUxM8//0yDBg2KMjOH/HSfOnWKNWvWOD0VoEJ+9/Vnn31Gt27dGDZsmPLTFa427927Fy8vL86dO0f//v3p9//t3XtM0+cex/G3IGBhWGu4iZMIyEXnlOncvJBIVNxQZ7KERZEN2AWH0xFQiFsCu5G5zTHlIgronMwlkgz4Y8GomCgLBEc2BhPBxbDNhDvIWsg2KJRy/nD0rAeOxyKtbc73lfSP39P2108foHz7PM/v99u1i9zcXIaGhqw283+qqKjAzc2NFStWmDOmEVMzL1iwAFdXVy5evMjo6Cj9/f1UV1fz5JNPWm3mkZGRCQW3k5MTd+7coaury+x5p+LWrVsAhISEGNpWrFjBwMAA7e3t0/56sqblbwMDA+j1elQqlVH7nDlz0Gg0Ex6vVquZM2fOhMfq9Xr6+/vx8PAwa14wPbM1eNDMzc3NfPfddxw5csRcESeYauaMjAyqq6sZHh7mueeeIzo62txRjZiaW6fTcejQIfbs2TPhOZYylb6OjIwkKCgIhUJBXV0dR44cYXR0lK1bt1oissmZu7q60Ov1lJWVkZqayujoKEePHkWr1ZKSkmKVmf9TRUUF4eHh//UEYOZgamYXFxcyMzNJT0/nxIkT6PV6Vq9ezRtvvGGpyCZnDgkJIS8vj/LyciIiIujp6aGkpAS4u4Zy3rx5FsltCrVazSOPPMLMmf8uJ8b/N2o0Gnx8fKb19WSk5W+WHHqbLv9vmVtbW0lLSyMuLo6lS5dOY6p7m2rmvXv3cvLkSd555x1qa2spLS2d5mT3Zmrus2fP8uijjxIWFmaeQPdhKn0dGRnJ448/zqJFi9ixYwc7d+40fNBbgqmZ9Xo9Op2ON998k5CQEFauXElCQgKXLl0yy3D6ZB7k7/DGjRu0trZafGrI1MyDg4NkZmaybt068vPzyczMpK+vj9zcXDMlnMjUzH5+fqSkpFBYWMjmzZt57bXXDAvKrfUM75O9R3NmlZGWvymVSuzs7FCr1UbtGo1mwogKgEqlmlApazQa7OzsUCqVZs06ztTM1mCqmTs6Ojhw4AARERHs2rXL3DGNTDXz+Py/j48POp2OY8eOERkZae64Bqbm/umnn2hsbGTjxo2GNr1ez8aNG/n4449ZtWqV1WWeTGBgoEUXLk7lswMw+gY6/juiVqtxc3Mzb2AerJ8vXrzIY489ZjQ1bgmmZr5y5Qp//vkniYmJhrbExEQSExOJj4//n2defRiZASIiInj22Wfp6+tj9uzZNDQ0AODl5WX2vFMxd+5c/vjjD3Q6nWG0Zfz9muP/kIy0/M3R0RF/f3/DLwhAZ2cnXV1dky6QCw4Opre3l7a2NkNbfX09vr6+ODk5WWVmazCVzN3d3ezfv5/Q0FDi4+MtFdVgOvp5bGzMokPpYHrugwcPcurUKcPt5ZdfRqVScerUKYuNbE1HX//yyy94enqaK+IEpmYeb/vnfH97ezsODg4Wm5abaj8PDw9TWVlp8VEWMD3z0NDQhG/8dnZ2Fh2hnmo/z5gxAzc3NxwdHamsrCQ4OPihHjxxLwEBAcDdLz3j6uvrmT17tlmOiLOPi4t7b9r3aqPs7e0pKirCx8eHkZERsrOz8fDwIDo6mps3b3LgwAFCQ0MNRyrU19dz7do1Fi1aRHNzM4WFhcTGxhp+iNaWGe5+OPb09FBbW8vQ0BALFy5kYGDAoiMzpmTu7e0lOTkZPz8/Xn/9dYaGhhgcHGRkZMRixaGpma9fv87333+Pg4MDQ0ND1NfXk5+fT1hYGE8//bTFMpua29XVFZVKZbh1dHTQ0NDA7t27cXBwsMrMTU1NVFdX4+joyODgIJWVlZw5c4aXXnqJ4OBgq8ysUqloamqiqqqKwMBAenp6yMnJITQ0lLVr11pl5nGVlZVUVVVx8OBBix+hZWpmZ2dnzp07h1arxcvLi/b2dvLy8pg3b55Fz/Nkaj+XlZXh6OhIf38/X3/9NeXl5aSlpVm0EB8YGKC1tZXffvuNmpoa1q5di0ajQaFQoNFo2LNnD8HBwbi7u+Pk5ER7ezsXLlwgKCiItrY2cnJy2LZtm1lOrSHTQ/+wZcsW1Go1WVlZhpMAjS+M02q1tLa2Gs05v/vuu3z22Wfs27cPZ2dnoqKi2LJli1Vn/vTTT40q4traWjw9PSkuLrbKzHV1dXR0dNDR0UFNTY1hH8uXLycrK8sqMzs6OlJRUUF+fj46nQ5PT0+2b9/Ojh07LJZ3KrmthSmZHRwcuHz5Mp9//jl6vR5vb2/27t1rsUW4U8kMkJaWRlZWFklJSbi4uLB+/Xp2795t1ZkBLl26RGhoqEWmViZjSuaFCxfywQcfUFRURFlZGQqFgieeeIKEhASrzQxw/fp1vvjiC4aHhwkICODw4cMWP/KwpqaGTz75xLA93mdHjx7Fy8uL1tZWo6Nkk5OTyc7OJiUlBXt7ezZv3kxsbKxZss24evWq7a3mFEIIIcT/HVnTIoQQQgibIEWLEEIIIWyCFC1CCCGEsAlStAghhBDCJkjRIoQQQgibIEWLEEIIIWyCFC1CCCGEsAlStAghhBDCJkjRIoQQQgibIKfxF0I8sKSkJKPLQwAsXryY48ePMzw8TGZmJj///DNtbW1ER0fz6quvPqSkQghbJkWLEGJaREZGEhUVZdgev0z96OgoCoWCqKgoysrKHla8Sen1esbGxrC3t3/YUYQQ90GKFiHEtJg1axZz586d0K5QKEhOTgbuXnDvfty6dYvc3FxaWlqYOXMmfn5+fPjhh4YL9ZWWllJWVkZvby9ubm5ER0cbLpJYW1tLQUEBbW1tuLu7ExMTwzPPPANAV1cXUVFRvPfeexQXF9PS0sKxY8cICgriypUrfPnll3R2duLt7U1cXBzr16+fjq4RQkwTKVqEEFbn0KFDrFmzhrfffhutVsuNGzcM95WXl3P69Gn27dvHsmXL6OnpQaPRAHeLkvT0dHbu3MmmTZuoq6vj8OHDeHt7G10pd/z53t7eKJVKfvzxR3JyckhKSiIwMJDm5mY++ugj3N3dWbJkicXfvxBiclK0CCGmRXFxMSUlJYbt/fv3Ex4ePqV99fT0sGbNGry9vQHw9fU13PfVV18RExNDREQEAPPnzzfc98033+Dv788rr7wCgI+PD42NjZSWlhoVLVFRUaxatcpon7GxsYSFhQHg7e1NQ0MD58+fl6JFCCsiRYsQYlps3bqVF154wbCtUqmmvK/nn3+e1NRUVq1axcqVK9mwYQNKpZK//vqL7u5uQkJCJn1ea2srixcvNmpbsmQJFy5cMGoLCAgw2v71119pamqi+dLrdgAAAgVJREFUsLDQ0KbT6YwKHSHEwydFixBiWri6uhqNejyI+Ph4Nm3axLVr17h8+TJnzpwhLy/vfxZCY2Nj97X/WbNmGW0PDg6SkJDAU089ZdTu6OhoWnAhhFnJeVqEEFbJ19eXXbt2cfz4cVQqFVVVVbi4uODp6UlDQ8Okz/Hx8eHmzZtGbc3NzSxYsOCer+Xv709nZyfz5883urm7u0/b+xFCPDgZaRFCmN3t27fR6XQMDg6iVqtpaWlBoVBMOjKj1WopKCggLCwMDw8Pbt++TU9Pj6HwePHFFzlx4gRKpZJly5Zx584dfv/9d8LCwti+fTslJSWcPn2a8PBwfvjhB7799luys7PvmS86Opr3338fd3d3Vq9ejVarpbGxEaVSyYYNG8zSJ0II00nRIoQwu7feeovu7m7g7uHM58+fZ/ny5WRlZU14rJ2dHRqNhoyMDPr7+3FzcyMmJobQ0FAAtm3bxuDgIEVFRfT19eHu7k50dDQAXl5eZGRkUFBQwLlz5/Dw8CA1NZWlS5feM9+6detIT0/n7NmznDx5EmdnZwIDA4mLi5vejhBCPJAZV69evb9JYCGEEEKIh0jWtAghhBDCJkjRIoQQQgibIEWLEEIIIWyCFC1CCCGEsAlStAghhBDCJkjRIoQQQgibIEWLEEIIIWyCFC1CCCGEsAlStAghhBDCJvwLZk5txcAUdRsAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plotting(mcc,f1,thresholds,MCC_F1_Met,T_index, classifer_name='CNN-biLSTM ')\n",
    "plt.savefig('MCC-F1-CNNbiLSTM-best-model.png', dpi=1200, bbox_inches='tight')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "multitasker_crispr.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}