covid19-compare(kaggle).ipynb

{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import numpy as np\nimport matplotlib.pyplot as plt\nimport keras\nfrom keras.layers import *\nfrom keras.models import * \nfrom keras.preprocessing import image","execution_count":1,"outputs":[{"output_type":"stream","text":"Using TensorFlow backend.\n","name":"stderr"}]},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"TRAIN_PATH = \"../input/chest-xray-pneumonia/chest_xray/train\"\nVAL_PATH = \"../input/chest-xray-pneumonia/chest_xray/val\"","execution_count":2,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = Sequential()\nmodel.add(Conv2D(32,kernel_size=(3,3), activation='relu', input_shape=(224,224,3)))\nmodel.add(Conv2D(64,kernel_size=(3,3), activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\nmodel.add(Dropout(0.25))\n\n\nmodel.add(Conv2D(64,kernel_size=(3,3), activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\nmodel.add(Dropout(0.25))\n\nmodel.add(Conv2D(128,kernel_size=(3,3), activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\nmodel.add(Dropout(0.25))\n\nmodel.add(Flatten())\nmodel.add(Dense(64,activation = 'relu'))\nmodel.add(Dropout(0.25))\nmodel.add(Dense(1, activation=\"sigmoid\"))\n\nmodel.compile(loss = keras.losses.binary_crossentropy, optimizer='adam', metrics=['accuracy'])","execution_count":6,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.summary()","execution_count":7,"outputs":[{"output_type":"stream","text":"Model: \"sequential_4\"\n_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nconv2d_13 (Conv2D)           (None, 222, 222, 32)      896       \n_________________________________________________________________\nconv2d_14 (Conv2D)           (None, 220, 220, 64)      18496     \n_________________________________________________________________\nmax_pooling2d_10 (MaxPooling (None, 110, 110, 64)      0         \n_________________________________________________________________\ndropout_13 (Dropout)         (None, 110, 110, 64)      0         \n_________________________________________________________________\nconv2d_15 (Conv2D)           (None, 108, 108, 64)      36928     \n_________________________________________________________________\nmax_pooling2d_11 (MaxPooling (None, 54, 54, 64)        0         \n_________________________________________________________________\ndropout_14 (Dropout)         (None, 54, 54, 64)        0         \n_________________________________________________________________\nconv2d_16 (Conv2D)           (None, 52, 52, 128)       73856     \n_________________________________________________________________\nmax_pooling2d_12 (MaxPooling (None, 26, 26, 128)       0         \n_________________________________________________________________\ndropout_15 (Dropout)         (None, 26, 26, 128)       0         \n_________________________________________________________________\nflatten_4 (Flatten)          (None, 86528)             0         \n_________________________________________________________________\ndense_4 (Dense)              (None, 64)                5537856   \n_________________________________________________________________\ndropout_16 (Dropout)         (None, 64)                0         \n_________________________________________________________________\ndense_5 (Dense)              (None, 1)                 65        \n=================================================================\nTotal params: 5,668,097\nTrainable params: 5,668,097\nNon-trainable params: 0\n_________________________________________________________________\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_datagen = image.ImageDataGenerator(\n    rescale = 1./255,\n    shear_range = 0.2,\n    zoom_range = 0.2,\n    horizontal_flip = True,\n)\n\ntest_dataset = image.ImageDataGenerator(rescale=1./255)","execution_count":12,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen = train_datagen.flow_from_directory(\n     \"../input/covid-dataset/CovidDataset/Train\",\n    target_size = (224,224),\n    batch_size = 32,\n    class_mode = 'binary',\n)","execution_count":16,"outputs":[{"output_type":"stream","text":"Found 224 images belonging to 2 classes.\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen.class_indices","execution_count":17,"outputs":[{"output_type":"execute_result","execution_count":17,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"validation_gen = test_dataset.flow_from_directory(\n    '../input/covid-dataset/CovidDataset/Val',\n    target_size = (224,224),\n    batch_size = 32,\n    class_mode = 'binary',\n)","execution_count":21,"outputs":[{"output_type":"stream","text":"Found 60 images belonging to 2 classes.\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"validation_gen.class_indices","execution_count":22,"outputs":[{"output_type":"execute_result","execution_count":22,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"hist = model.fit_generator(\n    train_gen,\n    steps_per_epoch = 8,\n    validation_data = validation_gen,\n    validation_steps = 2,\n    epochs = 10\n)","execution_count":23,"outputs":[{"output_type":"stream","text":"Epoch 1/10\n8/8 [==============================] - 21s 3s/step - loss: 0.2784 - accuracy: 0.8828 - val_loss: 0.2084 - val_accuracy: 0.9667\nEpoch 2/10\n8/8 [==============================] - 21s 3s/step - loss: 0.2877 - accuracy: 0.8984 - val_loss: 0.1454 - val_accuracy: 0.9833\nEpoch 3/10\n8/8 [==============================] - 19s 2s/step - loss: 0.1928 - accuracy: 0.9258 - val_loss: 0.1274 - val_accuracy: 0.9833\nEpoch 4/10\n8/8 [==============================] - 21s 3s/step - loss: 0.1594 - accuracy: 0.9453 - val_loss: 0.1479 - val_accuracy: 0.9667\nEpoch 5/10\n8/8 [==============================] - 20s 3s/step - loss: 0.1377 - accuracy: 0.9688 - val_loss: 0.0216 - val_accuracy: 0.9833\nEpoch 6/10\n8/8 [==============================] - 21s 3s/step - loss: 0.0959 - accuracy: 0.9570 - val_loss: 0.1340 - val_accuracy: 0.9667\nEpoch 7/10\n8/8 [==============================] - 20s 3s/step - loss: 0.1127 - accuracy: 0.9570 - val_loss: 0.1963 - val_accuracy: 0.9500\nEpoch 8/10\n8/8 [==============================] - 19s 2s/step - loss: 0.1006 - accuracy: 0.9688 - val_loss: 0.1035 - val_accuracy: 0.9667\nEpoch 9/10\n8/8 [==============================] - 21s 3s/step - loss: 0.1510 - accuracy: 0.9492 - val_loss: 0.1301 - val_accuracy: 0.9667\nEpoch 10/10\n8/8 [==============================] - 20s 2s/step - loss: 0.1360 - accuracy: 0.9648 - val_loss: 0.0832 - val_accuracy: 0.9667\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.save(\"model_adv.h5\")","execution_count":25,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.evaluate_generator(train_gen)","execution_count":26,"outputs":[{"output_type":"execute_result","execution_count":26,"data":{"text/plain":"[0.08884092420339584, 0.9866071343421936]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.evaluate_generator(validation_gen)","execution_count":27,"outputs":[{"output_type":"execute_result","execution_count":27,"data":{"text/plain":"[0.08543799072504044, 0.9666666388511658]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.plot(hist.history['accuracy'])\nplt.plot(hist.history['val_accuracy'])\nplt.title('model accuracy')\nplt.ylabel('accuracy')\nplt.xlabel('epoch')\nplt.legend(['train', 'test'], loc='best')\nplt.show()","execution_count":28,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = load_model('model_adv.h5')","execution_count":29,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import os","execution_count":30,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen.class_indices","execution_count":31,"outputs":[{"output_type":"execute_result","execution_count":31,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual = []\ny_test = []","execution_count":34,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for i in os.listdir('../input/covid-dataset/CovidDataset/Val/Normal'):\n  img = image.load_img('../input/covid-dataset/CovidDataset/Val/Normal/'+i, target_size=(224,224))\n  img = image.img_to_array(img)\n  img = np.expand_dims(img, axis=0)\n  p = model.predict_classes(img)\n  y_test.append(p[0,0])\n  y_actual.append(1)","execution_count":35,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for i in os.listdir('../input/covid-dataset/CovidDataset/Val/Covid'):\n  img = image.load_img('../input/covid-dataset/CovidDataset/Val/Covid/'+i, target_size=(224,224))\n  img = image.img_to_array(img)\n  img = np.expand_dims(img, axis=0)\n  p = model.predict_classes(img)\n  y_test.append(p[0,0])\n  y_actual.append(0)","execution_count":36,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual = np.array(y_actual)\ny_test = np.array(y_test)","execution_count":37,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual, y_test","execution_count":38,"outputs":[{"output_type":"execute_result","execution_count":38,"data":{"text/plain":"(array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]),\n array([1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], dtype=int32))"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.metrics import confusion_matrix","execution_count":39,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"cm = confusion_matrix(y_actual, y_test)","execution_count":40,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import seaborn as sns","execution_count":41,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sns.heatmap(cm, cmap=\"plasma\", annot=True)","execution_count":42,"outputs":[{"output_type":"execute_result","execution_count":42,"data":{"text/plain":"<matplotlib.axes._subplots.AxesSubplot at 0x7f5e89ea0990>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{},"cell_type":"markdown","source":"# Main Comparison"},{"metadata":{},"cell_type":"markdown","source":"Comparing with IEEE Dataset from https://github.com/ieee8023/covid-chestxray-dataset repo"},{"metadata":{"trusted":true},"cell_type":"code","source":"train_datagen = image.ImageDataGenerator(\n    rescale = 1./255,\n    shear_range = 0.2,\n    zoom_range = 0.2,\n    horizontal_flip = True,\n)\n\ntest_dataset = image.ImageDataGenerator(rescale=1./255)","execution_count":43,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen = train_datagen.flow_from_directory(\n    '../input/covid-dataset/CovidDataset/Train',\n    target_size = (224,224),\n    batch_size = 32,\n    class_mode = 'binary',\n    subset='training'\n)","execution_count":44,"outputs":[{"output_type":"stream","text":"Found 224 images belonging to 2 classes.\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen.class_indices","execution_count":45,"outputs":[{"output_type":"execute_result","execution_count":45,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"ieee_gen = test_dataset.flow_from_directory(\n    '../input/ieeeimagedataset/datasetcovid19/',\n    target_size = (224,224),\n    batch_size = 32,\n    class_mode = 'binary',\n)","execution_count":50,"outputs":[{"output_type":"stream","text":"Found 364 images belonging to 2 classes.\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"ieee_gen.class_indices","execution_count":51,"outputs":[{"output_type":"execute_result","execution_count":51,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"hist = model.fit_generator(\n    train_gen,\n    steps_per_epoch = 6,\n    validation_data = ieee_gen,\n    validation_steps = 2,\n    epochs = 5,\n)","execution_count":58,"outputs":[{"output_type":"stream","text":"Epoch 1/5\n6/6 [==============================] - 25s 4s/step - loss: 0.0392 - accuracy: 0.9948 - val_loss: 0.1912 - val_accuracy: 0.9531\nEpoch 2/5\n6/6 [==============================] - 29s 5s/step - loss: 0.0317 - accuracy: 0.9948 - val_loss: 0.0086 - val_accuracy: 0.9844\nEpoch 3/5\n6/6 [==============================] - 13s 2s/step - loss: 0.0762 - accuracy: 0.9688 - val_loss: 0.0040 - val_accuracy: 0.9844\nEpoch 4/5\n6/6 [==============================] - 37s 6s/step - loss: 0.0123 - accuracy: 1.0000 - val_loss: 0.0350 - val_accuracy: 0.9531\nEpoch 5/5\n6/6 [==============================] - 21s 3s/step - loss: 0.0502 - accuracy: 0.9844 - val_loss: 0.0575 - val_accuracy: 0.9531\n","name":"stdout"}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.save(\"model_adv.h5\")","execution_count":59,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.evaluate_generator(train_gen)","execution_count":60,"outputs":[{"output_type":"execute_result","execution_count":60,"data":{"text/plain":"[0.00458112545311451, 0.9821428656578064]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.evaluate_generator(ieee_gen)","execution_count":61,"outputs":[{"output_type":"execute_result","execution_count":61,"data":{"text/plain":"[0.02208774723112583, 0.9752747416496277]"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"plt.plot(hist.history['accuracy'])\nplt.plot(hist.history['val_accuracy'])\nplt.title('model accuracy')\nplt.ylabel('accuracy')\nplt.xlabel('epoch')\nplt.legend(['train', 'test'], loc='best')\nplt.show()","execution_count":63,"outputs":[{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 1 Axes>","image/png":"iVBORw0KGgoAAAANSUhEUgAAAYgAAAEWCAYAAAB8LwAVAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nO3dd3gU5dfG8e9JSAiBkEASegkl9N6kS++KomLvioi90hRFRbGhPyv2LoIoFqR3pLdQE3oLNfQSQtrz/jGLb4wb2ITdnc3mfK4rl8nO7M6dMezZmXnmPGKMQSmllMouwO4ASimlfJMWCKWUUk5pgVBKKeWUFgillFJOaYFQSinllBYIpZRSTmmBUAoQka9F5BUX190lIl08nUkpu2mBUEop5ZQWCKX8iIgUsjuD8h9aIFS+4Ti184yIrBORsyLyhYiUFpGpInJaRGaJSIks618tIhtF5ISIzBOR2lmWNRaR1Y7njQdCsm2rj4jEOZ67WEQauJixt4isEZFTIrJXRF7Mtryt4/VOOJbf5Xi8iIi8LSK7ReSkiPzteKyDiCQ62Q9dHN+/KCITReR7ETkF3CUiLURkiWMbB0TkAxEJzvL8uiIyU0SOicghERkmImVEJFlEIrOs11REkkQkyJXfXfkfLRAqv7kO6ArUAK4CpgLDgCisv+dHAUSkBjAOeByIBqYAf4pIsOPN8jfgO6Ak8LPjdXE8twnwJfAAEAl8AvwhIoVdyHcWuAOIAHoDD4rINY7XreTI+74jUyMgzvG8t4CmQGtHpmeBTBf3SV9gomObPwAZwBOOfdIK6AwMcmQIA2YB04ByQHVgtjHmIDAP6J/ldW8DfjLGpLmYQ/kZLRAqv3nfGHPIGLMPWAgsM8asMcacByYBjR3r3Qj8ZYyZ6XiDewsogvUG3BIIAt41xqQZYyYCK7Js437gE2PMMmNMhjHmG+C843kXZYyZZ4xZb4zJNMaswypSVzoW3wrMMsaMc2z3qDEmTkQCgHuAx4wx+xzbXOz4nVyxxBjzm2Ob54wxq4wxS40x6caYXVgF7kKGPsBBY8zbxpgUY8xpY8wyx7JvsIoCIhII3IxVRFUBpQVC5TeHsnx/zsnPxRzflwN2X1hgjMkE9gLlHcv2mX93qtyd5fvKwFOOUzQnROQEUNHxvIsSkStEZK7j1MxJYCDWJ3kcr7HdydOisE5xOVvmir3ZMtQQkckictBx2ulVFzIA/A7UEZGqWEdpJ40xy/OYSfkBLRDKX+3HeqMHQEQE681xH3AAKO947IJKWb7fC4wyxkRk+Qo1xoxzYbs/An8AFY0x4cBY4MJ29gLVnDznCJCSw7KzQGiW3yMQ6/RUVtlbMn8MJACxxpjiWKfgLpUBY0wKMAHrSOd29OihwNMCofzVBKC3iHR2XGR9Cus00WJgCZAOPCoihUSkH9Aiy3M/AwY6jgZERIo6Lj6HubDdMOCYMSZFRFoAt2RZ9gPQRUT6O7YbKSKNHEc3XwJjRKSciASKSCvHNY8tQIhj+0HAc8ClroWEAaeAMyJSC3gwy7LJQBkReVxECotImIhckWX5t8BdwNXA9y78vsqPaYFQfskYsxnrfPr7WJ/QrwKuMsakGmNSgX5Yb4THsa5X/JrluSuxrkN84Fi+zbGuKwYBL4nIaWAEVqG68Lp7gF5YxeoY1gXqho7FTwPrsa6FHANeBwKMMScdr/k51tHPWeBfo5qceBqrMJ3GKnbjs2Q4jXX66CrgILAV6Jhl+SKsi+OrHdcvVAEmOmGQUiorEZkD/GiM+dzuLMpeWiCUUv8QkebATKxrKKftzqPspaeYlFIAiMg3WPdIPK7FQYEeQSillMqBHkEopZRyyq8ae0VFRZmYmBi7YyilVL6xatWqI8aY7PfWAH5WIGJiYli5cqXdMZRSKt8Qkd05LdNTTEoppZzSAqGUUsopLRBKKaWc0gKhlFLKKS0QSimlnPJYgRCRL0XksIhsyGG5iMh7IrJNrCkkm2RZ1kNENjuWDfFURqWUUjnz5BHE10CPiyzvCcQ6vgZg9bC/0O/+Q8fyOsDNIlLHgzmVUko54bECYYxZgNW2OCd9gW+NZSkQISJlsfrybzPG7HC0Zf7Jsa5SqgBYn3iSuZsP2x1DYe81iPL8e6rERMdjOT3ulIgMEJGVIrIyKSnJI0GVUt5x5nw693yzgru/WsHbMzajveLsZWeBECePmYs87pQx5lNjTDNjTLPoaKd3iyul8okP524j6fR5OtUqxftztvHE+DjOp2fYHavAsrNAJGLNEXxBBax5hHN6XCnlx3YfPcsXC3fSr0l5vrizGU93q8Fvcfu5/YvlnEhOtTtegWRngfgDuMMxmqklcNIYcwBrysVYEakiIsHATY51lVJ+bNRf8RQKFAb3qIWI8HCnWP53UyPi9pyg38eL2XM02e6IBY4nh7mOw5ocvqaIJIrIvSIyUEQGOlaZAuzAmu/3M6x5dzHGpAMPA9OBeGCCMWajp3Iqpey3aNsRZmw6xEMdq1O6eMg/j/dtVJ7v7m3B0TOpXPvRIlbvOW5jyoLHryYMatasmdFurkrlL+kZmfR+72+S09KZ+cSVhAQF/med7UlnuPurFRw6lcK7NzaiZ/2yNiT1TyKyyhjTzNkyvZNaKWWrccv3sPnQaYb3qu20OABUiy7GpEGtqVOuOIN+XM1nC3boCCcv0AKhlLLNieRU3p65hVZVI+let8xF140sVphx97ekR90yjJoSz4jfN5KekemlpAWTFgillG3enbWVU+fSGHFVHUScjXD/t5CgQD68pQkPtK/Kd0t3M+C7VZw9n+6FpAWTFgillC22HjrNd0t3c8sVlahdtrjLzwsIEIb2qs3L19Rj3ubD9P9kCYdOpXgwacGlBUIp5XXGGF6avImiwYE82bVmnl7j9paV+eLO5uw8cpZrPlxEwsFTbk6ptEAopbxudvxhFm49wuNdalCyaHCeX6djrVJMeKAVmcZw/cdLWLBF2+24kxYIpZRXnU/P4JW/NlEtuii3t6p82a9Xr3w4kwa1oUKJItz99QrGr9jjhpQKtEAopbzs60W72HU0mRFX1SUo0D1vQeUiivDzwFa0qR7F4F/W8+b0BDIzdRjs5dICoZTymqTT53l/zjY61yrFlTXc21wzLCSIL+5sxk3NK/Lh3O08Nj6OlDRt9Hc5CtkdQClVcLw1fTPn0zMY3ru2R14/KDCA1/rVp1JkKG9M28zBk+f49PZmlLiM6xwFmR5BKKW8Yn3iSSas2stdrWOoGl3MY9sREQZ1qM77NzdmbeJJ+n28mF1Hznpse/5MC4RSyuOMMYz8cyMlQ4N5pHOsV7Z5VcNy/HDfFRxPTqXfx4tZtVsb/eWWFgillMf9ue4AK3cf55nuNSkeEuS17TaPKcmkQW0oHlKImz9byl/rDnht2/5AC4RSyqPOpWbw2pR46pYrzg3NKl76CW5WJaoovw5qQ/3y4Tz042rGzt+ujf5cpAVCKeVRnyzYzoGTKbxwVV0CAy7db8kTShYN5of7rqB3g7KMnprAc79t0EZ/LtBRTMAb0xJI1zHTuXJ1w3LUKx9udwzl4/adOMfY+dvp06AsLaqUtDVLSFAg79/UmIolQhk7fzv7Tpzjg1uaUKywvg3mRPcM8NOKvZxL1fHSrkrLyOSXVYlMf6I9UcUK2x1H+bDRUxMwBob28syw1twKCBCG9KxF5chQnvttAzeMXcJXdzWnTHjIpZ9cAOmMcirXthw6TZ/3/6Z9bDSf3dHUpTbNquBZvvMY/T9ZwqOdY3myaw274/zH/C1JDPp+FWEhQXx5V3PqlHO9o6w/0RnllFvVKB3Gs91rMiv+ED+vTLQ7jvJBGZnWsNay4SEMvLKq3XGcurJGND8PbA3ADWMXM2/zYZsT+R4tECpP7mlThVZVIxn550b2Hku2O47yMRNX7WXj/lMM6VmL0GDfPZNdp1xxfnuoDZUii3LvNyv5cZk2+stKC4TKk4AA4a3+DQkQ4ckJcWToRX7lcDoljTenb6ZZ5RJc3bCc3XEuqUx4CD8PbEXb6lEMm7Se0VO10d8FWiBUnpWPKMLIvnVZses4ny3cYXcc5SM+mLONI2dSXZ5G1BcUK1yIL+5sxq1XVGLs/O088tMabfSHFgh1ma5tXJ6e9crw9ozNbNqvM3oVdDuPnOXLRTu5oWkFGlSIsDtOrhQKDOCVa+oxtGct/lp3gFs/X8axs6l2x7KVFgh1WUSEUdfWJyI0mCcnxHE+XT91FWSj/tpEcGAAz/TI2zSidhMRHriyGh/e0oT1+07S76NF7CzAjf60QKjLVrJoMG9c34CEg6cZM2OL3XGUTRZsSWJW/GEe6RxLqbD8fV9B7wZlGXf/FZxKSaffR4tYseuY3ZFsoQVCuUXHmqW49YpKfLpwB0t3HLU7jvKytIxMXp68icqRodzdJsbuOG7RtHJJJg1qTURoMLd+tow/1+63O5LXaYFQbjO8d20qlwzlqQlrOZ2SZncc5UU/LN3N1sNnGN6rNoULBdodx20qRxbl1wdb07BiOI+MW8NH87YVqEZ/WiCU24QGF2LMjY04cPIcI//cZHcc5SXHzqYyZuYW2laPomud0nbHcbsSRYP57t4ruLphOd6Ytplhk9aTVkAa/WmBUG7VpFIJHupYnYmrEpm+8aDdcZQXvDNzC2dTM3i+T/4Z1ppbIUGBvHtjIx7uWJ1xy/dy7zcrC8RRshYI5XaPdo6lXvniDP11PUmnz9sdR3lQwsFT/LBsN7ddUYmaZcLsjuNRAQHC091r8vp19Vm07Qg3jF3C/hPn7I7lUVoglNsFBQbwTv9GnDmfzpBf1hWoc7YFiTGGl/7cRPEiQTzhg834POXG5pX4+u7m7Dt+jms/WsTG/SftjuQxWiCUR8SWDmNIj1rMTjjM+BV77Y6jPGDGpkMs3n6UJ7vWICI02O44XtUuNpqfH2xFoAj9xy5hboJ/NvrTAqE85q7WMbSpHslLkzex+2jBvdnIH6WkZTDqr3hqlC7GLS0q2R3HFrXKFGfSQ22IiSrKvd+s4Pulu+2O5HYeLRAi0kNENovINhEZ4mR5CRGZJCLrRGS5iNTLsuwxEdkgIhtF5HFP5lSeERAgvHl9QwIDhKcmrNWGfn7ky0U72XMsmRF96lIosOB+zixdPIQJD7SiQ81SPPfbBl6bEu9Xjf489n9WRAKBD4GeQB3gZhGpk221YUCcMaYBcAfwP8dz6wH3Ay2AhkAfEYn1VFblOeUiivBy33qs3H2cTxZstzuOcoNDp1L4YM42utYpTdvYKLvj2K5o4UJ8entTbm9ZmU8W7ODhcav9ptGfJ0t/C2CbMWaHMSYV+Anom22dOsBsAGNMAhAjIqWB2sBSY0yyMSYdmA9c68GsyoP6NipH7/pleWfmFr++oFdQvDFtM+kZhuE+Mo2oLygUGMBLfevyXO/aTN1wkJs/W8rRM/l/BJ8nC0R5IOvVyUTHY1mtBfoBiEgLoDJQAdgAtBeRSBEJBXoBFZ1tREQGiMhKEVmZlJTk5l9BuYOI8Mo19SgRGswT4+P85tNVQbR27wl+WZ3IPW2rEBNV1O44PkVEuK9dVT6+tQmb9p/i2o8Wsz3pjN2xLosnC4SzO2ayn5wbDZQQkTjgEWANkG6MiQdeB2YC07AKSbqzjRhjPjXGNDPGNIuOjnZbeOVeJRwN/bYcOsPbMzbbHUflgTGGF//cSFSxwjzcqbrdcXxWj3plGTegJWfPp9Pvo8Usy8e9yTxZIBL596f+CsC/ul0ZY04ZY+42xjTCugYRDex0LPvCGNPEGNMeOAZs9WBW5QUdapbitpaV+PzvnSzZnn//0RRUv8ftZ82eEzzboybFCvvuNKK+oEmlEkwa1IbIYsHc/sVyfo/bZ3ekPPFkgVgBxIpIFREJBm4C/si6gohEOJYB3AcsMMacciwr5fhvJazTUOM8mFV5ybBetYmJLMrTP6/lVAFoVeAvzp5P57Wp8dQvH871TSrYHSdfqBQZyq8PtqZxpQge+ymOD+ZszXc3jXqsQDguLj8MTAfigQnGmI0iMlBEBjpWqw1sFJEErNFOj2V5iV9EZBPwJ/CQMea4p7Iq7wkNLsSY/g05eCqFkX9oQ7/8Yuz87Rw6dZ4Xr65DQIB/9lvyhIjQYL69twXXNCrHWzO2MPiXdfmq0Z9HjxONMVOAKdkeG5vl+yWA0+Grxph2nsym7NPY0dDvvdlb6VqnFD3qlbU7krqIvceS+XTBDvo2KkfTyiXtjpPvFC4UyDs3NqJSyVDem7ONAydT+PDWJhQPCbI72iUV3DtclK0e6VSdBhXCGfrreg6fTrE7jrqI0VMTCBBhSM9adkfJt0SEJ7vV5M3rG7Bk+1Fu+HgJ+/JBoz8tEMoWQYEBjOnfiOTUDAZP1IZ+vmrpjqP8tf4AA6+sRtnwInbHyfduaFaRb+5pwf4T57jmw0Vs2Ofb9wVpgVC2qV6qGEN71mLu5iTGLdeGfr4mI9Mw8s9NlI8owoD2Ve2O4zfaVI/il0GtCQ4MoP8nS5gdf8juSDnSAqFsdUerGNpWj+KVvzax64g29PMl41fsJf7AKYb2qkWRYP+ZRtQX1CgdxqSHWlMtuhj3f7uSb5fssjuSU1oglK0CAoQ3b2hAoQDhyQlxpOejER7+7OS5NN6asZkWMSXpXV8HEXhCqbAQxj/Qkk61SjHi9428MnmTzzX60wKhbFc2vAgvX1OP1XtO8MmCHXbHUcD7s7dyPDmVEVf57zSiviA0uBCf3N6Mu1rH8PnfOxn0w2rOpfpOKxotEMon9G1Unj4NrIZ+vn7hzt9tTzrD14t3cWOzitQrH253HL8XGCC8eHVdRvSpw/RNB7nps6Uc8ZFGf1oglM945Zp6RBbThn52e2XyJooEBfJUt5p2RylQ7mlbhbG3NWXzwVNc+9Eith22v9GfFgjlMyJCg3nz+oZsPXyGN6drQz87zE04zNzNSTzaOZbosMJ2xylwutctw/gBrTiXmkG/jxax1OZGf1oglE9pXyOaO1pV5ou/d7J42xG74xQoaRmZvPzXJqpEFeXO1jF2xymwGlaMYNKgNpQqHsLtXyxj0ppE27JogVA+Z2jP2lSNshr6nTynDf285dslu9mRdJbn+9QmuJC+NdipYslQfhnYmmaVS/LE+LW8N9ueRn/6V6B8TpFgq3fNodPnGfnHRrvjFAhHz5zn3VlbaF8jmo41S9kdRwHhoUF8c08L+jUpz5iZW3hm4jpS0707DFwLhPJJDStG8HDH6vy6Zh9T1h+wO47fe3vmFpJTMxjRp7YOa/UhwYUCePuGhjzeJZaJqxK566vlXj2q1gKhfNbDnarTsEI4wyat5/ApbejnKRv3n2Tc8j3c0aoy1UuF2R1HZSMiPN6lBm/f0JAVu45x/ceLSTye7JVta4FQPisoMIAxNzYiJS2DZ3/Rhn6eYIzhpT83EVEkiMc717A7jrqI65pW4Jt7WnDwVArXfLiYdYknPL5NLRDKp1WLLsawXrWZtzmJH5btsTuO35m24SDLdh7jqW41CQ/1/fkJCrrW1aKYNKg1IUEB3PjJUmZu8myjPy0Qyufd3rIy7WKjGPVXPDu1oZ/bpKRlMGpKPLXKhHFzi0p2x1Euql4qjEmD2lCjdDEGfLeSrxbt9Ni2dOZxlXtxP8KuRV7bnACfFE9nRuAhtn/+OZVrliIgv11IrXsNxHa1O8W/fL5wB4nHz/Hj/VcQqNOI5ivRYYX5aUArHvtpDSP/3MSeY8k817uO2/8/aoFQubNvFfz2IIRGQiHvTSATCnQvksGx5FTOJhQirHA+Oh2Seho2ToJHVkFx3+iMevBkCh/O3U6PumVoXS3K7jgqD4oEB/LxbU0Z9Vc8K3cfIzU90+1t2bVAKNcZA1OHQNFS1ptdSHGvbr4I8Pq4NUxZf4BJt7ehfoV80kju2E74sAXMHgnXjr30+l7w+rQEMoxhWK/adkdRlyEwQBhxVR3OpWZ4ZM4OvQahXLd+IiQuh84jvF4cLni5bz2iihXm8fFr8k9Dv5JVoNVDsHYcJK60Ow2r9xxn0pp93N+uCpUiQ+2Oo9zAUxM6aYFQrkk9CzNHQNlG0OhW22KEhwbx5g0N2J50ltenJdiWI9faPQXFSsPUwZBp36RImY5pREuFFWZQh+q25VD5gxYI5Zq/34XT+6Hn6xBg759Nu9ho7modw1eLdvH31nzS0K9wGHR5EfathPUTbIsxac0+1u49weAetShaWM8wq4vTAqEu7cQeWPwe1LseKrW0Ow0Ag3vUolp0UZ6ZmI8a+jW4Cco1gVkvwnnv9/o/cz6d16cl0LBiBNc2Lu/17av8RwuEurSZIwCBriPtTvKPCw39kk6f54XfN9gdxzUBAdYR2OkD8Pc7Xt/8R3O3cfj0eV64qg4BOqxVuUALhLq4XYusIZptn4DwCnan+ZcGFSJ4pFMsv8XtZ/K6/XbHcU3FFtDgRlj8Phzf5bXN7jmazOcLd9KvcXmaVCrhte2q/E0LhMpZZgZMGwzFK0DrR+xO49RDHavRsGIEwydt4ODJfNLQr8uLEBAIM5732iZfnRJPoUDh2R61vLZNlf9pgVA5W/MdHFwP3V6CYN8cDlkoMIB3+jfkfHo+auhXvBy0fRLi/4CdCz2+ucXbjjBt40EGdahGmfAQj29P+Q8tEMq5lJMw+2Wo1Arq9rM7zUVVjS7G8F61WbAlie+X7rY7jmtaPwzhlWDaEOtIzUPSMzJ5afImKpQown3tqnpsO8o/aYFQzs1/A5KPQo/RkA/6Ht3WsjLta0Qzako8O5K8P0Io14KKQLeX4dAGWP2NxzYzbsVeEg6eZniv2oQEeeZmKuW/tECo/zqyFZaNhSa3Q7lGdqdxiYjw5vUNCAkK5IkJa0nPsO9mNJfV6QuV21pHaueOu/3lTySnMmbGZlpWLUmPemXc/vrK/7lUIETkFxHpLSJaUAqC6cOtRnydvHcR1R1KFw9h1DX1Wbv3BB/O3W53nEsTgR6vWcVh/htuf/l3Z23l5Lk0RvSpq9OIqjxx9Q3/Y+AWYKuIjBYRHQrhr7bOgq3T4cpnoVj+m7y+d4OyXNOoHO/N2cravZ6fceuylW0ATe+E5Z9C0ha3vezWQ6f5bulubm5RiTrl7OmbpfI/lwqEMWaWMeZWoAmwC5gpIotF5G4RybHvsoj0EJHNIrJNRIY4WV5CRCaJyDoRWS4i9bIse0JENorIBhEZJyI6/MLTMtJg+lAoWQ2uGGh3mjwb2bcepcIK88SEOM6l5oOGfp2eh6CiMH2YW17OGMNLkzcRGhzIk111GlGVdy6fMhKRSOAu4D5gDfA/rIIxM4f1A4EPgZ5AHeBmEamTbbVhQJwxpgFwh+M1EZHywKNAM2NMPSAQuMnl30rlzYrP4cgW6P4qFAq2O02ehRcJ4q0bGrIjvzT0KxoFHQbDtpmwZcZlv9ychMMs3HqEx7vUILJYYTcEVAWVS926RORXoBbwHXCVMeaAY9F4Ecmpf3ELYJsxZofjNX4C+gKbsqxTB3gNwBiTICIxIlI6S7YiIpKGNV9MPrlVNp86ewTmvgbVOkON7nanuWxtqkdxdxuroV/n2qVoFxttd6SLa34/rPzKOoKr2iHPBTo1PZOXJ2+iWnRR7mhV2a0R/VVaWhqJiYmkpOSTGy3zKCQkhAoVKhAU5PpkW662c/zAGDPH2QJjTLMcnlMe2Jvl50TgimzrrAX6AX+LSAugMlDBGLNKRN4C9gDngBnGmMv/aKVyNncUpJ6xjh785ILm4B61WLj1CM/8vI7pj7cnPNSHZ6ErFGzt+x9vsK5HtH44Ty/z9eKd7DqazNd3NycoUMeUuCIxMZGwsDBiYmL89mK+MYajR4+SmJhIlSpVXH6eq39BtUUk4sIPjmsHgy7xHGd7OvttrqOBEiISBzyCdeoqXURKYB1tVAHKAUVF5DanGxEZICIrRWRlUlKSi7+O+peDG2DV19DifijlP+MPQoICead/I46cOc/z+aGhX41uUL0rzH8dzuT+bznp9Hnen72NTrVK0aFm/htgYJeUlBQiIyP9tjiANQw8MjIy10dJrhaI+40x/wwJMcYcB+6/xHMSgYpZfq5AttNExphTxpi7jTGNsK5BRAM7gS7ATmNMkjEmDfgVaO1sI8aYT40xzYwxzaKjffw0gi8yxrqbNyQCOvxnHEG+V79COI91juWPtfv5Y20+OEvZ/VVIS4Y5L+f6qW/P2My5tAyG99ZpRHPLn4vDBXn5HV0tEAGS5dUdF6AvdZJ0BRArIlVEJBjrIvMfWVcQkQjHMrAufi8wxpzCOrXUUkRCHdvtDMS7mFXlRvyfsGshdBoORfyzy+eDHarRuFIEz01a7/sN/aJrQIsHYPW3cGCdy0/bsO8k41fu5a7WMVSLLubBgMrdTpw4wUcffZTr5/Xq1YsTJzw7lNvVAjEdmCAinUWkEzAOmHaxJxhj0oGHHc+NByYYYzaKyEARuTCGsjawUUQSsEY7PeZ47jJgIrAaWO/I+WmufjN1aWkpMGM4lKoLTe6yO43HWA39GpGWYXhm4loyM328od+Vz0JoSevIzoXmg8YYRv65kZKhwTzSOdYLAZU75VQgMjIuPkR7ypQpREREXHSdy+XqRerBwAPAg1jXFmYAn1/qScaYKcCUbI+NzfL9EsDpX7Qx5gXgBRfzqbxY8oE1W9wdf0Cgf08/GRNVlOG9a/Pcbxv4bulu7mwdY3eknBWJgE7PweQnYNNvUPfai64+ed0BVuw6zmv96hNexIcvxCunhgwZwvbt22nUqBFBQUEUK1aMsmXLEhcXx6ZNm7jmmmvYu3cvKSkpPPbYYwwYMACAmJgYVq5cyZkzZ+jZsydt27Zl8eLFlC9fnt9//50iRYpcdjaX3hWMMZlYd1N/fNlbVL7h1H5YOAZq9YGqV9qdxituvaISs+IP8drUeNpUj6J6KR8+FdPkTljxhTVnRI0eVnM/J86lZvDalHjqlC1O/2YVna6jXDfyz41s2n/Kra9Zp1xxXriqbo7LR48ezYYNG4iLi2PevHn07t2bDRs2/DPa6Msvv6RkyZKcO3eO5s2bc9111xEZGfmv19i6dSvjxo3js88+o3///nvx4XcAAB99SURBVPzyyy/cdpvTcT254movplgRmSgim0Rkx4Wvy966ss+skZCZBt1esTuJ14gIb1zXgCJBgTw5IY40X27oFxBoddI9udeafS4Hny7Ywf6TKbxwVR0CdRpRv9CiRYt/DUV97733aNiwIS1btmTv3r1s3br1P8+pUqUKjRpZjTWbNm3Krl273JLF1fMKX2Gd7nkH6AjcjfNhrCo/SFwJ636yJq0p6fqYaH9QqngIo66tz6AfVvPBnG084cutKKq0szq+/v0ONLoVwsv/a/H+E+f4eP42etcvyxVVI3N4EZUbF/uk7y1Fixb95/t58+Yxa9YslixZQmhoKB06dHA6VLVw4f+/Yz4wMJBz5865JYurF6mLGGNmA2KM2W2MeRHo5JYEyrsyM2HqYChWBto9aXcaW/SqX5Z+jcvzwdxtxPl6Q7+uL1sTCs168T+LRk9NwBgY0tN/7l0piMLCwjh9+rTTZSdPnqREiRKEhoaSkJDA0qVLvZrN1QKR4mj1vVVEHhaRawG9Eyc/Wj8B9q205kUuHGZ3Gtu82LcupcMK8+R4H2/oV6IytHnU+v+2Z9k/D6/YdYw/1u7ngfZVqVjSN6eDVa6JjIykTZs21KtXj2eeeeZfy3r06EF6ejoNGjTg+eefp2XLll7NJq7M4SsizbGGqkYALwPFgTeNMd4tZ5fQrFkzs3JlTq2hFOfPwPtNrVMV986CgILdimHx9iPc8tky7mhVmZf61rv0E+xy/gx80AzCysB9c8hE6PvhIpJOn2fO01cSGuzfI9A8LT4+ntq1C8bNhc5+VxFZlVPLpEu+QzhuiutvjDljjEl03Pl8na8VB+WCv8fAmYPWxc8CXhwAWleL4t62Vfh2yW7mb/HhNi2Fi0GXkbB/Dawdx8TViazfd5KhvWppcVAedcl3CWNMBtBUCsK96P7s+C5Y/AE0uBEqtrA7jc94pntNYksV45mf13IiOdXuODmrfwNUaE7mrBf5YGocTSpFcHXDcnanUn7O1Y+Ra4DfReR2Eel34cuTwZSbzXjeGjrZ5UW7k/iUkKBA3rmxEceTUxn+2wZcOeVqi4AA6PE6AWcPc9P5CbxwlU4jqjzP1QJREjiKNXLpKsdXH0+FUm62cwHE/2GNWiqunzqzq1c+nMe71OCvdQd8uqHfzpBa/JrZngFBU2lY9JjdcVQB4Oqd1Hd7OojykIx0mDYUwitBq7zNMVAQPNC+KrPjD/H8bxtoHlOSchGX36bA3Ub9Fc82buaaQiutI8KbfrA7kvJzrt5J/ZWIfJn9y9PhlBus/gYObYBuL+fYrkFZDf3G9G9EeqZvNvRbuDWJWfGHuLHTFQS0fxoSJsP2uXbHUn7O1VNMk4G/HF+zsYa5nvFUKOUm547DnFegclvrjlx1UTFRRXmudx0WbTvKN0t22R3nH+kZmbz05yYqlQzlnrYx0PIhiKhsHRlmpNsdT12mvLb7Bnj33XdJTk52c6L/51KBMMb8kuXrB6A/4MMDxxUA89+AlBPQ4zW/mUbU025uUZFOtUoxemoC2w47v7vV235Ytoeth88wvHdtChcKhKAQ6D4KkuJh1Vd2x1OXKd8XCCdigUruDKLcLGmzNbdxkzuhbAO70+QbIsLo6+oTGhzIE+PX2t7Q7/jZVMbM3EKb6pF0q1P6/xfU6gNV2ltziSfrBev8LGu772eeeYY333yT5s2b06BBA154wZrx4OzZs/Tu3ZuGDRtSr149xo8fz3vvvcf+/fvp2LEjHTt29Eg2ly5Si8hp/j2f9EGsOSKULzLGOv0QVNSaV0DlSqmwEF7rV5+B36/m/dlbebJbTduyvDNrC6dT0hjRJ9uwVhHrhsexbWHeaOj1hm0Z/crUIXBwvXtfs0x96Dk6x8VZ233PmDGDiRMnsnz5cowxXH311SxYsICkpCTKlSvHX3/9BVg9msLDwxkzZgxz584lKirKvZkdXD3FFGaMKZ7lq4Yx5hePJFKXb+sM2D4bOgyGop75w/F3PeqV5bomFfhw3nZW7zluS4bNB0/z/dLd3NayMjXLOOmbVbouNL0bVnwOh3VGXn8wY8YMZsyYQePGjWnSpAkJCQls3bqV+vXrM2vWLAYPHszChQsJDw/3Sh5XjyCuBeYYY046fo4AOhhjfvNkOJUH6anW0UNkLDS/3+40+doLV9dh6Y6jPDk+jimPtfNqWwtjDC9N3khYSBBPdLlIS/KOw2HDROv/+e2T9FrT5brIJ31vMMYwdOhQHnjggf8sW7VqFVOmTGHo0KF069aNESNGeDyPq9cgXrhQHACMMSfQ6UB90/JP4dh268J0oWC70+RrxUOCeOuGhuw+lsyrU7z7CX3mpkMs2naUJ7rEUqLoRf4/Fo2EDsNgx1zYctFp4pWPytruu3v37nz55ZecOWMNEt23bx+HDx9m//79hIaGctttt/H000+zevXq/zzXE1z9SOSskGiXMF9zJgnmvw6x3SC2q91p/EKrapHc17YKny3cSefapelY0/Nd7s+nZ/DKX/HElirGrS0rX/oJze+FlV/C9GFQrRMUKnzp5yifkbXdd8+ePbnlllto1aoVAMWKFeP7779n27ZtPPPMMwQEBBAUFMTHH1uzPw8YMICePXtStmxZ5s51/30xrrb7/hI4AXyIdbH6EaCEMeYutye6DAW+3fcfj0LcDzBoKUTF2p3Gb6SkZdD3g0UcS05lxuPtL/6J3g0+nred16cl8N29LWgXG+3ak7bNgu+vg64vQZvHPJrP32i778to9+3wCJAKjAcmAOeAhy4jp3K3A2th9bfQ4gEtDm4WEhTImBsbciI5leG/rfdoQ7/Dp1L4YM5WutQu7XpxAKjeBWr0gPlvwulDHsunChZXRzGdNcYMMcY0c3wNM8ac9XQ45SJjrOF5oSXhymftTuOX6pYL54muNZiy/iC/xe3z2HbenL6Z1IxMnuudh0+03UZBegrMecn9wVSB5GovppmOkUsXfi4hItM9F0vlyqbfYM9i6PQ8FIm49PoqTx5oX41mlUsw4veN7Dvhnknhs1q79wQ/r0rknjZViIkqeuknZBdVHVoOhDU/WJMLKXWZXD3FFOUYuQSAMeY4Oie1b0g7Z3X2LF0fmtxhdxq/FhggjOnfiMxMw9MT3NvQzxjDyD83ElUsmIc7Vc/7C7V/xrr3ZeoQ68hSucRn5wFxo7z8jq4WiEwR+ae1hojE8O87q5VdFr8PJ/da47cDAu1O4/cqRYbyfJ86LNlxlK8W73Lb6/6xdj+r95zg2e61CAsJyvsLhYRD5xGwdyls0HtZXRESEsLRo0f9ukgYYzh69CghISG5ep6rQ1WHA3+LyHzHz+2BAbnaknK/k4mwcIzVqTWmrd1pCowbm1dkVvwhXp+WQPvYKGJLO7nLOReSU9N5bUoC9cuHc33TCpcfsNGtsPwzmDkCavaC4NDLf00/VqFCBRITE0lK8uF5yd0gJCSEChVy9/fl6oRB00SkGVZRiAN+xxrJpOw060UwmdD1ZbuTFCgiwmv9GtD93QU8Pj6OSYPaEFwor30vYey87Rw8lcIHtzQmIMANd0IHBELP1+GrnrDof9Bx6OW/ph8LCgqiSpUqdsfwSa5epL4Pax6Ipxxf3wEvei6WuqQ9y2D9z9DmUSjhws1Uyq2iwwrzWr/6bNx/ivdmb83z6yQeT+aTBTu4qmE5msWUdF/Ayq2hbj9Y9C6c2Ou+11UFiqsfex4DmgO7jTEdgcaAfx+P+bLMTJg2GMLKQdsn7E5TYHWvW4Ybmlbgo3nbWLU7bw39XpuagAgM6VnLzemwbpoDmKVdcVTeuFogUowxKQAiUtgYkwDY1wO5oFs7zhrG2HUkBOdhOKRymxFX1aFcRBGenBDH2fO5m91t2Y6j/LXuAAOvrEZ5T8yBHVER2jxuXazevdj9r6/8nqsFItFxH8RvwEwR+R3Y77lYKkcpp6xrDxVaQP0b7E5T4IWFBPH2DQ3ZcyyZUblo6JeRaRj55ybKhYfwQPtqngvY5jEoXh6mDobMDM9tR/klV++kvtYYc8IY8yLwPPAFcI0ng6kcLHwbzh62hrVqa2efcEXVSAa0q8qPy/YwJ8G1NhcTVu5l04FTDO1VmyLBHhyeHBxqnWo6uM7q06VULuR66IUxZr4x5g9jTKonAqmLOLYDln4EDW+B8k3tTqOyeLJbDWqVCePZies5dvbi/zROpaTx1vTNNI8pQZ8GZT0frt51ULElzH7JOgJVykV5H5vnAhHpISKbRWSbiAxxsryEiEwSkXUislxE6jkerykicVm+TonI457Mmi/MeB4Cg6GLXnT0NYULBTKmfyNOnktl2K8Xb+j3/uytHEtO5YWrsk0j6iki1hHn2SOw4E3Pb0/5DY8VCBEJxGoP3hOoA9wsInWyrTYMiDPGNADuAP4HYIzZbIxpZIxpBDQFkoFJnsqaL2yfCwmTod1TEFbG7jTKiTrlivNUt5pM23iQX1c7b+i3PekMXy3aRf+mFalX3jvTRgJQrjE0vhWWfgxHt3tvuypf8+QRRAtgmzFmh+N01E9A32zr1MG6vwLHyKgYESmdbZ3OwHZjzG4PZvVtGenWlJIlYqDlILvTqIu4v11VWsSU5MU/nDf0G/VXPCFBgTzd3YZBgJ1GQKEQmD7c+9tW+ZInC0R5IOsdOomOx7JaC/QDEJEWQGUg+73gNwHjctqIiAwQkZUistJvb5Vf9RUkxUO3VyAod71UlHcFBghv929IpjE8NSHuXw395m0+zJyEwzzauTrRYTbM+hZWGto/DVumWhMMKXUJniwQzk6uZj8xOxooISJxWJMSrQH+GUwuIsHA1cDPOW3EGPPphXkqoqNzMcFKfpF8DOaOgirtoVYfu9MoF1QsGcoLV9Vl6Y5jfLloJwBpGZm8PHkTVaKKcldrG9s6tHwQSlSBacMgI82+HCpf8GSBSAQqZvm5AtnunTDGnDLG3O241nAHEA3szLJKT2C1MabgTpE1bzSknIQeOqw1P7mhWQW61C7NG9M3s/ngab5bspvtSWcZ3qv2ZfVtumyFCkP3V+HIZmsea6UuwpN/qSuAWBGp4jgSuAn4I+sKIhLhWAZwH7DAGJN1HN7NXOT0kt87HA8rPodm90DpunanUbkgIoy+rj5hhQvx2E9reHfWFtrFRtG5tg9Mo1KzJ1TtaB2Znj1qdxrlwzxWIIwx6cDDwHQgHphgjNkoIgNFZKBjtdrARhFJwDpa+Ge2dREJBboCv3oqo08zxrowXTgMOupFxfwoqlhhRl/XgISDpzmbmsGIPnW8M6z1UkSgx2tw/gzMe9XuNMqHuTofRJ4YY6YAU7I9NjbL90uA2ByemwxEejKfT9s8FXbMhR6vW3NNq3ypa53SDO5Ri6KFAy973gi3KlUbmt+rR6jqomw8GapylH4epg+DqJrWP2KVrz3YoRp3tIqxO8Z/dRhqzUA3dbBOT6qc0gLhi5aNheM7ocerEHgZ008qdTGhJa3Tl7sWWjdhKpWNFghfc/oQzH8TavSE6l3sTqP8XdO7Ibo2zHgO0lLsTqN8jBYIXzPnJUhPge6j7E6iCoLAQlafpuO7rEaQSmWhBcKX7FsNa36wbmaK9OAcAUplVbUD1OwNC96CUwfsTqN8iBYIX3FhWGvRKGj/jN1pVEHT7WXITLNagivloAXCV2z4BfYuhc4jIKS43WlUQRNZzWoEufZHSFxldxrlI7RA+ILUZJg5Aso2hEa32p1GFVTtn4ZipWGaDntVFi0QvmDR/+DUPuumuAAPTj+p1MUUDoPOL0DiClifY39MVYBogbDbib2w6F1rWsjKrexOowq6hjdbkwvNHGG14lAFmhYIu80cAQh0GWl3EqUgIMA6kj19wPrgogo0LRB22r0YNv4KbR6DiIqXXl8pb6h0BdS/ARa9B8cL7kSOSguEfTIzrB44xctbBUIpX9JlpHU9bOYIu5MoG2mBsEvcD3BwHXR9CYJD7U6j1L+Fl4e2T8Cm32DX33anUTbRAmGHlJPWDUkVW1oXp5XyRa0fgfCKMHWIdcSrChwtEHZY8CacPWL1wPGFCWSUciaoiHWEe2g9rP7W7jTKBlogvO3odlg6Fhrfag0nVMqX1b0WKrWGOS/DuRN2p1FepgXC26YPh0Ih0Ekv/ql8QMQ60k0+BvPfsDuN8jItEN60bRZsmQpXPgNhpe1Oo5RryjaEJnfA8k/gyFa70ygv0gLhLRlpMG0YlKwKVwy0O41SudPpeQgKtabCVQWGFghvWfEFHNkM3V+FQoXtTqNU7hSLhiufha0zYMsMu9MoL9EC4Q1nj8K8V6FqR6jRw+40SuVNiwegZDXrKCI91e40ygu0QHjDvFetxmc9XtNhrSr/KhRs/Q0f3QorPrM7jfICLRCedmgjrPwSmt8HpWrbnUapyxPbDap3gXmvW/fyKL+mBcKTjLH6LYWEQ4chdqdR6vKJWNfR0s7CnFfsTqM8TAuEJyVMhl0LoeNwCC1pdxql3CO6JrQYAKu+hgPr7E6jPEgLhKekpVg3xZWqA03vtjuNUu515bNQpARMG6rTk/oxLRCesvQjOLHbuqgXWMjuNEq5V5ES0Ok52P03bPrd7jTKQ7RAeMKpA7DgLajVB6p2sDuNUp7R9C4oXQ9mPg9p5+xOozxAC4QnzH4JMtOg28t2J1HKcwICrSPkE3tgyQd2p1EeoAXC3RJXwdofodVDVlsNpfxZlfZQ+2pYOAZO7bc7jXIzLRDuZAxMGwzFSkO7p+xOo5R3dHvZmlBo1ot2J1FupgXCndb/DIkroPMLUDjM7jRKeUeJGGj9MKwbD3tX2J1GuZFHC4SI9BCRzSKyTUT+c6eYiJQQkUkisk5ElotIvSzLIkRkoogkiEi8iLTyZNbLdv6MNcF7ucbQ8Ga70yjlXW2fhGJlrCPozEy70yg38ViBEJFA4EOgJ1AHuFlE6mRbbRgQZ4xpANwB/C/Lsv8B04wxtYCGQLynsrrFonfh9AHo+QYE6IGZKmAKF4OuI2HfKutIQvkFT76TtQC2GWN2GGNSgZ+AvtnWqQPMBjDGJAAxIlJaRIoD7YEvHMtSjTG+O9/h8d2w6D2o3x8qtrA7jVL2qN8fyjezrkWcP213GuUGniwQ5YG9WX5OdDyW1VqgH4CItAAqAxWAqkAS8JWIrBGRz0WkqLONiMgAEVkpIiuTkpLc/Tu4Zubz1pC/Li/as32lfEFAAPR8Hc4ctEY1qXzPkwXCWV/r7PfkjwZKiEgc8AiwBkgHCgFNgI+NMY2Bs4DTbnfGmE+NMc2MMc2io6PdFt5luxx3krZ9AsKz1z+lCpgKzaDBTdZ9Ecd22p1GXSZPFohEoGKWnysA/xoobYw5ZYy52xjTCOsaRDSw0/HcRGPMMseqE7EKhm/JzICpQyC8IrR+xO40SvmGLi9AQBDMeM7uJOoyebJArABiRaSKiAQDNwF/ZF3BMVIp2PHjfcACR9E4COwVkZqOZZ2BTR7Mmjerv4VD661x4EFF7E6jlG8oXg7aPWl1M94x3+406jJ4rEAYY9KBh4HpWCOQJhhjNorIQBEZ6FitNrBRRBKwRjs9luUlHgF+EJF1QCPgVU9lzZNzJ2DOy1C5DdS5xu40SvmWVg9DRCWr22tGut1pVB55tM2oMWYKMCXbY2OzfL8EiM3huXFAM0/muyzz34DkY9BjtE4jqlR2QSHQbRRMuB1Wf23NqKjyHR2wnxdJW2D5J9DkDijbwO40Svmm2ldBTDuYM8r6MKXyHS0QeTFjOASFQqfn7U6ilO8Ssbq9ppyA+a/bnUblgRaI3NoyA7bOgCsHQzEbhtUqlZ+UqW/NG7H8MzicYHcalUtaIHIjPRWmD4PI6tacvEqpS+s43GrFMV2nJ81vtEDkxorP4OhW6P4aFAq+9PpKKSgaBR2GwvY5sGW63WlULmiBcNXZIzDvdajeFWp0szuNUvlL8/sgqoZ1FJGeanca5SItEK6a8wqknYXuvnU7hlL5QmCQdeR9bAcsG3vp9ZVP0ALhigPrYNXX1nWH6Bp2p1Eqf4rtArHdYcGbcOaw3WmUC7RAXIox1t2goSXhymftTqNU/tb9VUhLtroQKJ+nBeJSNv0Ou/+GTs9BkRJ2p1Eqf4uqDlcMhNXfwf44u9OoS9ACcTFp52DG81C6HjS50+40SvmHK5+F0EiYNkSHvfo4LRAXs+QDOLnH6rcUEGh3GqX8Q0g4dH4e9iyBjb/anUZdhBaInJzab82KVftqqNLO7jRK+ZfGt1t3Wc8YAanJdqdROdACkZNZL1oTAnXTi2lKuV1AIPR4HU4lwuL37U6jcqAFwpm9y2HdeGuWuBIxdqdRyj/FtIG618Lf78DJRLvTKCe0QGSXmQlTB0NYWWueaaWU53R9CTAw8wW7kygntEBkt+4n2L8aurxoNRhTSnlORCVo/ShsmAi7l9idRmWjBSKr86dh1kgo3wzq97c7jVIFQ9vHIaycNew1M9PuNCoLLRBZLRwDZw5Cz9chQHeNUl4RXNQ61XQgDtb+aHcalYW+C15wbKd130PDm6GC706FrZRfqn89VLzCOoJPOWV3GuWgBeKCGc9BQBB01otlSnmdiHVD6tnDsPAtu9MoBy0QADvmQ8JkaP8UFC9rdxqlCqbyTaDRrbDkIzi63e40CihkdwDbZaRb3VojKkPLh+xOo1TB1nmE1SDzy+5WvyblmiIl4Z6pbn9ZLRDp56B8Y6tPfVCI3WmUKtjCysB1X+jF6twKCffIy4rxo26KzZo1MytXrrQ7hlJK5RsissoY43Rkjl6DUEop5ZQWCKWUUk5pgVBKKeWUFgillFJOaYFQSinllBYIpZRSTmmBUEop5ZQWCKWUUk751Y1yIpIE7M7j06OAI26M4y6aK3c0V+5ortzxx1yVjTHRzhb4VYG4HCKyMqe7Ce2kuXJHc+WO5sqdgpZLTzEppZRySguEUkopp7RA/L9P7Q6QA82VO5ordzRX7hSoXHoNQimllFN6BKGUUsopLRBKKaWcKlAFQkR6iMhmEdkmIkOcLBcRec+xfJ2INPGRXB1E5KSIxDm+Rngp15ciclhENuSw3K79dalcdu2viiIyV0TiRWSjiDzmZB2v7zMXc3l9n4lIiIgsF5G1jlwjnaxjx/5yJZctf2OObQeKyBoRmexkmXv3lzGmQHwBgcB2oCoQDKwF6mRbpxcwFRCgJbDMR3J1ACbbsM/aA02ADTks9/r+cjGXXfurLNDE8X0YsMVH/sZcyeX1febYB8Uc3wcBy4CWPrC/XMlly9+YY9tPAj86276791dBOoJoAWwzxuwwxqQCPwF9s63TF/jWWJYCESJS1gdy2cIYswA4dpFV7NhfruSyhTHmgDFmteP700A8UD7bal7fZy7m8jrHPjjj+DHI8ZV91Iwd+8uVXLYQkQpAb+DzHFZx6/4qSAWiPLA3y8+J/PcfiSvr2JELoJXjkHeqiNT1cCZX2bG/XGXr/hKRGKAx1qfPrGzdZxfJBTbsM8fpkjjgMDDTGOMT+8uFXGDP39i7wLNAZg7L3bq/ClKBECePZf9U4Mo67ubKNldj9UtpCLwP/ObhTK6yY3+5wtb9JSLFgF+Ax40xp7IvdvIUr+yzS+SyZZ8ZYzKMMY2ACkALEamXbRVb9pcLuby+v0SkD3DYGLPqYqs5eSzP+6sgFYhEoGKWnysA+/OwjtdzGWNOXTjkNcZMAYJEJMrDuVxhx/66JDv3l4gEYb0J/2CM+dXJKrbss0vlsvtvzBhzApgH9Mi2yNa/sZxy2bS/2gBXi8gurFPRnUTk+2zruHV/FaQCsQKIFZEqIhIM3AT8kW2dP4A7HCMBWgInjTEH7M4lImVERBzft8D6/3bUw7lcYcf+uiS79pdjm18A8caYMTms5vV95kouO/aZiESLSITj+yJAFyAh22p27K9L5rJjfxljhhpjKhhjYrDeJ+YYY27Ltppb91ehvMfNX4wx6SLyMDAda+TQl8aYjSIy0LF8LDAFaxTANiAZuNtHcl0PPCgi6cA54CbjGLLgSSIyDmu0RpSIJAIvYF2ws21/uZjLlv2F9QnvdmC94/w1wDCgUpZsduwzV3LZsc/KAt+ISCDWG+wEY8xku/9NupjLrr+x//Dk/tJWG0oppZwqSKeYlFJK5YIWCKWUUk5pgVBKKeWUFgillFJOaYFQSinllBYIpXyAWN1B/9OdUyk7aYFQSinllBYIpXJBRG4Ta66AOBH5xNHU7YyIvC0iq0VktohEO9ZtJCJLxerLP0lESjgery4isxyN3laLSDXHyxcTkYkikiAiP1y4U1cpu2iBUMpFIlIbuBFo42jklgHcChQFVhtjmgDzse7sBvgWGGyMaQCsz/L4D8CHjkZvrYELrRAaA48DdbDmB2nj8V9KqYsoMK02lHKDzkBTYIXjw30RrHbQmcB4xzrfA7+KSDgQYYyZ73j8G+BnEQkDyhtjJgEYY1IAHK+33BiT6Pg5DogB/vb8r6WUc1oglHKdAN8YY4b+60GR57Otd7H+NRc7bXQ+y/cZ6L9PZTM9xaSU62YD14tIKQARKSkilbH+HV3vWOcW4G9jzEnguIi0czx+OzDfMQ9Doohc43iNwiIS6tXfQikX6ScUpVxkjNkkIs8BM0QkAEgDHgLOAnVFZBVwEus6BcCdwFhHAdjB/3fWvB34RERecrzGDV78NZRymXZzVeoyicgZY0wxu3Mo5W56ikkppZRTegShlFLKKT2CUEop5ZQWCKWUUk5pgVBKKeWUFgillFJOaYFQSinl1P8BBNNQpVRoHMMAAAAASUVORK5CYII=\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = load_model('model_adv.h5')","execution_count":64,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import os","execution_count":65,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"train_gen.class_indices","execution_count":66,"outputs":[{"output_type":"execute_result","execution_count":66,"data":{"text/plain":"{'Covid': 0, 'Normal': 1}"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual = []\ny_test = []","execution_count":67,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for i in os.listdir('../input/ieeeimagedataset/datasetcovid19/Normal'):\n  img = image.load_img('../input/ieeeimagedataset/datasetcovid19/Normal/'+i, target_size=(224,224))\n  img = image.img_to_array(img)\n  img = np.expand_dims(img, axis=0)\n  p = model.predict_classes(img)\n  y_test.append(p[0,0])\n  y_actual.append(1)","execution_count":69,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for i in os.listdir('../input/ieeeimagedataset/datasetcovid19/Covid'):\n  img = image.load_img('../input/ieeeimagedataset/datasetcovid19/Covid/'+i, target_size=(224,224))\n  img = image.img_to_array(img)\n  img = np.expand_dims(img, axis=0)\n  p = model.predict_classes(img)\n  y_test.append(p[0,0])\n  y_actual.append(0)","execution_count":71,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual = np.array(y_actual)\ny_test = np.array(y_test)","execution_count":72,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"y_actual, y_test","execution_count":73,"outputs":[{"output_type":"execute_result","execution_count":73,"data":{"text/plain":"(array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]),\n array([1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1,\n        1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n        1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,\n        0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], dtype=int32))"},"metadata":{}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"from sklearn.metrics import confusion_matrix","execution_count":74,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"cm = confusion_matrix(y_actual, y_test)","execution_count":75,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"import seaborn as sns","execution_count":76,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"sns.heatmap(cm, cmap=\"plasma\", annot=True)","execution_count":77,"outputs":[{"output_type":"execute_result","execution_count":77,"data":{"text/plain":"<matplotlib.axes._subplots.AxesSubplot at 0x7f5e89018350>"},"metadata":{}},{"output_type":"display_data","data":{"text/plain":"<Figure size 432x288 with 2 Axes>","image/png":"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\n"},"metadata":{"needs_background":"light"}}]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}