{"cells": [{"metadata": {"collapsed": true, "id": "06c88850-7b8e-42d1-a6ad-ae2360698688"}, "cell_type": "markdown", "source": "\n\n

IBM-AWS Immersion Day Lab 4

Notebook 2 : Predict future COVID-19 cases for Wallonia region with Long Short-Term Memory (LSTM) Model

"}, {"metadata": {"id": "499cfa88fe8141488b0726ac89c783c7"}, "cell_type": "markdown", "source": "In this lab exercise, you will learn a popular opensource machine learning algorithm, Long Short-Term Memory (LSTM). You will use this time-series algorithm to build a model from historical data of total COVID-19 cases. Then you use the trained model to predict the future COVID-19 cases."}, {"metadata": {"id": "b63f82d05b984223974a40be6249f092"}, "cell_type": "markdown", "source": "### Import required libraries"}, {"metadata": {"id": "13ad35bac1f84a0785d69d1a7753b4fb"}, "cell_type": "code", "source": "import boto3\nimport numpy as np \nimport pandas as pd \nfrom keras.layers.core import Dense, Dropout\nfrom keras.layers.recurrent import LSTM\nfrom keras.models import Sequential\nfrom tensorflow.keras.optimizers import Adam\nimport math, time\nfrom sklearn.preprocessing import MinMaxScaler\nimport matplotlib.pyplot as plt\nfrom numpy import newaxis\nfrom keras.callbacks import EarlyStopping\nimport tensorflow\nfrom io import StringIO\nimport datetime\nimport io\nimport itertools\nfrom project_lib import Project\nproject = Project.access()\n%matplotlib inline", "execution_count": 1, "outputs": []}, {"metadata": {"id": "54ef6c9c375945638d46540986ef68a5"}, "cell_type": "markdown", "source": "### Load the dataset from Amazon S3 into pandas dataframe\n>Note: you can add the comment `# @hidden_cell` in the below code cell. Cloud Pak for Data will automatically hide the cell before sharing it."}, {"metadata": {"id": "10f3b48312fd423b804adb94be85c6f0"}, "cell_type": "code", "source": "", "execution_count": 2, "outputs": [{"data": {"text/plain": " DATE REGION Total_cases\n0 15/03/20 Wallonia 383\n1 16/03/20 Wallonia 568\n2 17/03/20 Wallonia 654\n3 18/03/20 Wallonia 925\n4 19/03/20 Wallonia 1245\n5 20/03/20 Wallonia 1468\n6 21/03/20 Wallonia 1377\n7 22/03/20 Wallonia 1464\n8 23/03/20 Wallonia 1758\n9 24/03/20 Wallonia 1777", "text/html": "
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DATEREGIONTotal_cases
015/03/20Wallonia383
116/03/20Wallonia568
217/03/20Wallonia654
318/03/20Wallonia925
419/03/20Wallonia1245
520/03/20Wallonia1468
621/03/20Wallonia1377
722/03/20Wallonia1464
823/03/20Wallonia1758
924/03/20Wallonia1777
\n
"}, "metadata": {}, "execution_count": 2, "output_type": "execute_result"}]}, {"metadata": {"id": "d6175c18974a42f58023affb240c02da"}, "cell_type": "code", "source": "regionData = data_df_1", "execution_count": 3, "outputs": []}, {"metadata": {"id": "64796fd153dc41c9899feec32458e8ec"}, "cell_type": "markdown", "source": "#### Drop REGION column"}, {"metadata": {"id": "d20ddda88d6e413b8fa734a7fcd64e5c"}, "cell_type": "code", "source": "data_df_1 = data_df_1.drop('REGION', axis=1)", "execution_count": 4, "outputs": []}, {"metadata": {"id": "8cc3027adbf047858764df215dfc1ca1"}, "cell_type": "markdown", "source": "#### Drop the index column and set DATE column as index column"}, {"metadata": {"id": "882d547aee92491288ce7b0fbe1361c6"}, "cell_type": "code", "source": "data_df_1.set_index('DATE', inplace=True)", "execution_count": 5, "outputs": []}, {"metadata": {"id": "26aa6afc1d044b7c8d9517a591abef3a"}, "cell_type": "markdown", "source": "### Fix random seed for reproducibility"}, {"metadata": {"id": "58584bc38f44413c8759f6adb67b1951"}, "cell_type": "code", "source": "tensorflow.random.set_seed(1309)", "execution_count": 6, "outputs": []}, {"metadata": {"id": "d58d669e3b0e4f9b9799a3754df9937f"}, "cell_type": "markdown", "source": "### Rename the dataframe and convert the datatype"}, {"metadata": {"id": "d823cfe088ed436494add605f7f9ffef"}, "cell_type": "code", "source": "series = data_df_1\nseries = series.astype(float)", "execution_count": 7, "outputs": []}, {"metadata": {"id": "36c46721995546e69cb0097defe800d2"}, "cell_type": "markdown", "source": "### Plot the data"}, {"metadata": {"id": "986f5539ec4641e0af2abd0b9971c60b"}, "cell_type": "code", "source": "plt.figure(figsize=(20,6))\nplt.plot(series.values)\nplt.show()", "execution_count": 8, "outputs": [{"data": {"text/plain": "
", "image/png": "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\n"}, "metadata": {"needs_background": "light"}, "output_type": "display_data"}]}, {"metadata": {"id": "d5e5c9cd3c9b44f09e375748cb9023d7"}, "cell_type": "markdown", "source": "### Normalize the data"}, {"metadata": {"id": "828692bc0a9e436184e07d24f1f898bb"}, "cell_type": "code", "source": "series = series.values\nscaler = MinMaxScaler(feature_range=(0, 1))\nseries = scaler.fit_transform(series)", "execution_count": 9, "outputs": []}, {"metadata": {"id": "19c195736c9a414e87966beba99c0388"}, "cell_type": "markdown", "source": "### Train Test Split 70:30 Ratio"}, {"metadata": {"id": "c971f04cf8d6460c893af685cb13fed5"}, "cell_type": "code", "source": "train_size = int(len(series) * 0.70)\ntest_size = len(series) - train_size\ntrain, test = series[0:train_size,:], series[train_size:len(series),:]\nprint(len(train), len(test))", "execution_count": 10, "outputs": [{"name": "stdout", "text": "518 223\n", "output_type": "stream"}]}, {"metadata": {"id": "e9eb1ee7f638470ba9b0419baab37e0f"}, "cell_type": "markdown", "source": "### Helper function to generate the dataset with input(X) & output(Y) variables"}, {"metadata": {"id": "01f605f6c04644dfa6d75ffe782b4191"}, "cell_type": "code", "source": "def create_dataset(dataset, look_back=1):\n dataX, dataY = [], []\n for i in range(len(dataset)-look_back-1):\n a = dataset[i:(i+look_back), 0]\n dataX.append(a)\n dataY.append(dataset[i + look_back, 0])\n return np.array(dataX), np.array(dataY)", "execution_count": 11, "outputs": []}, {"metadata": {"id": "35016902524f47c4a3901c50d95ee145"}, "cell_type": "markdown", "source": "### Create a dataset with a look back period of 15 observations\nThis is where we convert the time series problem into a regression problem"}, {"metadata": {"id": "61dc59629f84427d8c90af7a08de916f"}, "cell_type": "code", "source": "look_back = 30\ntrainX, trainY = create_dataset(train, look_back)\ntestX, testY = create_dataset(test, look_back)", "execution_count": 12, "outputs": []}, {"metadata": {"id": "0a93a0bd8a0f40d880362267ba44effe"}, "cell_type": "markdown", "source": "### Review the shape of datasets"}, {"metadata": {"id": "45b7737b846c41488d7e4d03598aea94"}, "cell_type": "code", "source": "trainX.shape", "execution_count": 13, "outputs": [{"data": {"text/plain": "(487, 30)"}, "metadata": {}, "execution_count": 13, "output_type": "execute_result"}]}, {"metadata": {"id": "1a8f882a4dc64cac8193c8a4b6a9be94"}, "cell_type": "code", "source": "testX.shape", "execution_count": 14, "outputs": [{"data": {"text/plain": "(192, 30)"}, "metadata": {}, "execution_count": 14, "output_type": "execute_result"}]}, {"metadata": {"id": "8c9eeedcbe79482eb01d2ab17dccd676"}, "cell_type": "markdown", "source": "### Reshape the data to 3D\nThe LSTM model requires the input data to be three dimensional"}, {"metadata": {"id": "d1e126a09526463e868011c9b2547ff6"}, "cell_type": "code", "source": "trainX = np.reshape(trainX, (trainX.shape[0], trainX.shape[1], 1))\ntestX = np.reshape(testX, (testX.shape[0], testX.shape[1], 1))", "execution_count": 15, "outputs": []}, {"metadata": {"id": "2413b15abb4941df8d2c1e48529055b1"}, "cell_type": "code", "source": "trainX.shape", "execution_count": 16, "outputs": [{"data": {"text/plain": "(487, 30, 1)"}, "metadata": {}, "execution_count": 16, "output_type": "execute_result"}]}, {"metadata": {"id": "23c134fbb71544579451910acb28af8f"}, "cell_type": "markdown", "source": "### Define the LSTM model\nActivation function will activate the neurons for the learning. Rectified linear unit (ReLu) is one of the most popular activations because the output does not go beyond 0.\n\nUnits will be the number of neurons in the input & hidden layers.\n\nStateful is where we define whether the previous information has to be remembered or not.\n\nDropout is where we omit random neurons for each layer as per the value (0 to 1). In this case we omit 20% of the neurons.\n\nOptimiser is where the weights are back propagated through the network to enhance the learnings closer to the desired outcome. Adam optimiser is an efficient method for enhanced accuracy."}, {"metadata": {"id": "e4abd31fb46b44c1a8c634699e148022"}, "cell_type": "markdown", "source": "Hyperparameters for the current model:\n- **train_test_split:** 0.70\n- **lookback:** 30\n- **hidden_layers:** 2\n- **units:** 60, 100\n- **dropouts:** 0.15, 0.15\n- **optimizer:** adam\n- **learning_rate:** 0.001 (default)\n- **epochs:** 25\n- **batch_size:** 32"}, {"metadata": {"id": "71067fb37f6641268bf84af22eeac41d"}, "cell_type": "code", "source": "print('LSTM Model Summary')\nmodel = Sequential()\nmodel.add(LSTM(input_shape=(trainX.shape[1], trainX.shape[2]), kernel_initializer=\"uniform\", return_sequences=True, stateful=False, units=60))\nmodel.add(Dropout(0.15))\nmodel.add(LSTM(100, kernel_initializer=\"uniform\", activation='relu',return_sequences=False))\nmodel.add(Dropout(0.15))\nmodel.add(Dense(32,kernel_initializer=\"uniform\",activation='relu'))\nmodel.add(Dense(1, activation='linear'))\n# optimizer = Adam(learning_rate=0.0006)\n# model.compile(loss=\"mean_squared_error\", optimizer=optimizer)\nmodel.compile(loss=\"mean_squared_error\", optimizer='adam')\nmodel.summary()", "execution_count": 17, "outputs": [{"name": "stdout", "text": "LSTM Model Summary\nModel: \"sequential\"\n_________________________________________________________________\n Layer (type) Output Shape Param # \n=================================================================\n lstm (LSTM) (None, 30, 60) 14880 \n \n dropout (Dropout) (None, 30, 60) 0 \n \n lstm_1 (LSTM) (None, 100) 64400 \n \n dropout_1 (Dropout) (None, 100) 0 \n \n dense (Dense) (None, 32) 3232 \n \n dense_1 (Dense) (None, 1) 33 \n \n=================================================================\nTotal params: 82,545\nTrainable params: 82,545\nNon-trainable params: 0\n_________________________________________________________________\n", "output_type": "stream"}]}, {"metadata": {"id": "051715d0c1e84ab2ae582f5905298219"}, "cell_type": "markdown", "source": "### Parameter Calculation\nparams = 4 * (size_of_input + 1 * size_of_output) + 4 * size_of_output^2\n\n"}, {"metadata": {"id": "c815bcf26f494d769654005be9006edc"}, "cell_type": "markdown", "source": "### Optimize computation time using early stopping\n\nWe monitor the accuracy of validation loss ('val_loss') and end the training if there's no improvement in the accuracy after five iterations ('patience=5')."}, {"metadata": {"id": "d90bff143b644a4e93d8e899d1c5988d"}, "cell_type": "code", "source": "early_stopping=EarlyStopping(monitor='val_loss', patience=5, verbose=1, mode='auto')", "execution_count": 18, "outputs": []}, {"metadata": {"id": "a04dd4e4a37140d78e8439207b9ce463"}, "cell_type": "markdown", "source": "### Fitting the model for training data"}, {"metadata": {"id": "47db82e646c54e86ad97475695c55622"}, "cell_type": "code", "source": "start = time.time()\nhistory = model.fit(trainX, trainY, batch_size=32, epochs=25, verbose=1, shuffle=False, validation_split=0.10, callbacks=[early_stopping])\nprint(\"> Compilation Time : \", time.time() - start)", "execution_count": 19, "outputs": [{"name": "stdout", "text": "Epoch 1/25\n14/14 [==============================] - 6s 138ms/step - loss: 0.0142 - val_loss: 0.0010\nEpoch 2/25\n14/14 [==============================] - 1s 97ms/step - loss: 0.0111 - val_loss: 0.0045\nEpoch 3/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0110 - val_loss: 0.0035\nEpoch 4/25\n14/14 [==============================] - 1s 101ms/step - loss: 0.0103 - val_loss: 0.0031\nEpoch 5/25\n14/14 [==============================] - 1s 102ms/step - loss: 0.0089 - val_loss: 0.0016\nEpoch 6/25\n14/14 [==============================] - 1s 101ms/step - loss: 0.0073 - val_loss: 7.0599e-04\nEpoch 7/25\n14/14 [==============================] - 1s 97ms/step - loss: 0.0081 - val_loss: 0.0029\nEpoch 8/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0061 - val_loss: 4.8584e-04\nEpoch 9/25\n14/14 [==============================] - 1s 99ms/step - loss: 0.0052 - val_loss: 2.3361e-04\nEpoch 10/25\n14/14 [==============================] - 1s 101ms/step - loss: 0.0056 - val_loss: 0.0028\nEpoch 11/25\n14/14 [==============================] - 1s 106ms/step - loss: 0.0042 - val_loss: 1.3609e-04\nEpoch 12/25\n14/14 [==============================] - 1s 100ms/step - loss: 0.0039 - val_loss: 1.1586e-04\nEpoch 13/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0044 - val_loss: 0.0018\nEpoch 14/25\n14/14 [==============================] - 1s 99ms/step - loss: 0.0033 - val_loss: 4.8268e-05\nEpoch 15/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0027 - val_loss: 4.9645e-05\nEpoch 16/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0030 - val_loss: 7.5402e-04\nEpoch 17/25\n14/14 [==============================] - 1s 104ms/step - loss: 0.0022 - val_loss: 4.5085e-05\nEpoch 18/25\n14/14 [==============================] - 2s 110ms/step - loss: 0.0021 - val_loss: 4.2691e-05\nEpoch 19/25\n14/14 [==============================] - 2s 109ms/step - loss: 0.0022 - val_loss: 1.8926e-04\nEpoch 20/25\n14/14 [==============================] - 1s 95ms/step - loss: 0.0019 - val_loss: 4.6648e-05\nEpoch 21/25\n14/14 [==============================] - 1s 101ms/step - loss: 0.0020 - val_loss: 4.1863e-05\nEpoch 22/25\n14/14 [==============================] - 1s 96ms/step - loss: 0.0018 - val_loss: 4.3175e-05\nEpoch 23/25\n14/14 [==============================] - 1s 95ms/step - loss: 0.0015 - val_loss: 4.0031e-05\nEpoch 24/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0019 - val_loss: 8.4009e-05\nEpoch 25/25\n14/14 [==============================] - 1s 98ms/step - loss: 0.0017 - val_loss: 4.2816e-05\n> Compilation Time : 39.40610432624817\n", "output_type": "stream"}]}, {"metadata": {"id": "cf16354ada204b128efabce30fab4f68"}, "cell_type": "markdown", "source": "Model run time is ~ 300 seconds and has produced almost similar values for training & validation loss which is great."}, {"metadata": {"id": "6850758269a34f74a5e5439d7dd72764"}, "cell_type": "markdown", "source": "### Create a function to calculate accuracy\nWe will be using 'Mean Squared Error' & 'Root Mean Squared Error' functions to calculate accuracy"}, {"metadata": {"id": "3c45d320fdee4f4aafe109028749310c"}, "cell_type": "code", "source": "def model_score(model, trainX, trainY, testX, testY):\n trainScore = model.evaluate(trainX, trainY, batch_size=32, verbose=0)\n print('Train Score: %.5f MSE (%.2f RMSE)' % (trainScore, math.sqrt(trainScore)))\n print('Train Accuracy: %.2f %%' % (100 - math.sqrt(trainScore)*100))\n\n testScore = model.evaluate(testX, testY, batch_size=32, verbose=0)\n print('Test Score: %.5f MSE (%.2f RMSE)' % (testScore, math.sqrt(testScore)))\n print('Test Accuracy: %.2f %%' % (100 - math.sqrt(testScore)*100))\n return trainScore, testScore", "execution_count": 20, "outputs": []}, {"metadata": {"id": "e3213a83f67b4377976a9af76e99cba9"}, "cell_type": "markdown", "source": "### Check the Accuracy of the model"}, {"metadata": {"id": "9253ca4e259542f98cd9fd0c0b581589"}, "cell_type": "code", "source": "model_score(model, trainX, trainY, testX, testY)", "execution_count": 21, "outputs": [{"name": "stdout", "text": "Train Score: 0.00134 MSE (0.04 RMSE)\nTrain Accuracy: 96.34 %\nTest Score: 0.01155 MSE (0.11 RMSE)\nTest Accuracy: 89.25 %\n", "output_type": "stream"}, {"data": {"text/plain": "(0.0013373465044423938, 0.011548556387424469)"}, "metadata": {}, "execution_count": 21, "output_type": "execute_result"}]}, {"metadata": {"id": "1c629badaaee4d6c94ea839691ed90aa"}, "cell_type": "markdown", "source": "We can observe that the Root Mean Squared Error (RMSE) values are almost similar for training & test data which confirms the accuracy of the model without overfitting or underfitting.\n\nThe model accuracy is > 94% as per the values of Mean Squared Error (MSE)"}, {"metadata": {"id": "eccf42416a8a45968fc738bfbd08932e"}, "cell_type": "markdown", "source": "### Review the learning of training & validation loss (error evaluation)"}, {"metadata": {"id": "c1a7b07f166d432f897df3f032868a03"}, "cell_type": "code", "source": "'''Review the learning'''\n\nplt.plot(history.history['loss']) # Train\nplt.plot(history.history['val_loss']) # Test\nplt.show()", "execution_count": 22, "outputs": [{"data": {"text/plain": "
", "image/png": "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\n"}, "metadata": {"needs_background": "light"}, "output_type": "display_data"}]}, {"metadata": {"id": "d3efe96f75d14d4c8957fbe61d5dbeaf"}, "cell_type": "markdown", "source": "There's no vanishing gradient descent as the LSTM model with optimal configueration has taken care of the gradient descent problem."}, {"metadata": {"id": "e6f251c98b594f90b233091181d38be9"}, "cell_type": "markdown", "source": "### Get the configuration of the model\nThis will give us an idea about all the parameters available and which ones have been choosen."}, {"metadata": {"id": "7ce92d3acb1c4a018fd9a7fbd3eee2f2"}, "cell_type": "code", "source": "model.get_config()", "execution_count": 23, "outputs": [{"data": {"text/plain": "{'name': 'sequential',\n 'layers': [{'class_name': 'InputLayer',\n 'config': {'batch_input_shape': (None, 30, 1),\n 'dtype': 'float32',\n 'sparse': False,\n 'ragged': False,\n 'name': 'lstm_input'}},\n {'class_name': 'LSTM',\n 'config': {'name': 'lstm',\n 'trainable': True,\n 'batch_input_shape': (None, 30, 1),\n 'dtype': 'float32',\n 'return_sequences': True,\n 'return_state': False,\n 'go_backwards': False,\n 'stateful': False,\n 'unroll': False,\n 'time_major': False,\n 'units': 60,\n 'activation': 'tanh',\n 'recurrent_activation': 'hard_sigmoid',\n 'use_bias': True,\n 'kernel_initializer': {'class_name': 'RandomUniform',\n 'config': {'minval': -0.05, 'maxval': 0.05, 'seed': None}},\n 'recurrent_initializer': {'class_name': 'Orthogonal',\n 'config': {'gain': 1.0, 'seed': None}},\n 'bias_initializer': {'class_name': 'Zeros', 'config': {}},\n 'unit_forget_bias': True,\n 'kernel_regularizer': None,\n 'recurrent_regularizer': None,\n 'bias_regularizer': None,\n 'activity_regularizer': None,\n 'kernel_constraint': None,\n 'recurrent_constraint': None,\n 'bias_constraint': None,\n 'dropout': 0.0,\n 'recurrent_dropout': 0.0,\n 'implementation': 1}},\n {'class_name': 'Dropout',\n 'config': {'name': 'dropout',\n 'trainable': True,\n 'dtype': 'float32',\n 'rate': 0.15,\n 'noise_shape': None,\n 'seed': None}},\n {'class_name': 'LSTM',\n 'config': {'name': 'lstm_1',\n 'trainable': True,\n 'dtype': 'float32',\n 'return_sequences': False,\n 'return_state': False,\n 'go_backwards': False,\n 'stateful': False,\n 'unroll': False,\n 'time_major': False,\n 'units': 100,\n 'activation': 'relu',\n 'recurrent_activation': 'hard_sigmoid',\n 'use_bias': True,\n 'kernel_initializer': {'class_name': 'RandomUniform',\n 'config': {'minval': -0.05, 'maxval': 0.05, 'seed': None}},\n 'recurrent_initializer': {'class_name': 'Orthogonal',\n 'config': {'gain': 1.0, 'seed': None}},\n 'bias_initializer': {'class_name': 'Zeros', 'config': {}},\n 'unit_forget_bias': True,\n 'kernel_regularizer': None,\n 'recurrent_regularizer': None,\n 'bias_regularizer': None,\n 'activity_regularizer': None,\n 'kernel_constraint': None,\n 'recurrent_constraint': None,\n 'bias_constraint': None,\n 'dropout': 0.0,\n 'recurrent_dropout': 0.0,\n 'implementation': 1}},\n {'class_name': 'Dropout',\n 'config': {'name': 'dropout_1',\n 'trainable': True,\n 'dtype': 'float32',\n 'rate': 0.15,\n 'noise_shape': None,\n 'seed': None}},\n {'class_name': 'Dense',\n 'config': {'name': 'dense',\n 'trainable': True,\n 'dtype': 'float32',\n 'units': 32,\n 'activation': 'relu',\n 'use_bias': True,\n 'kernel_initializer': {'class_name': 'RandomUniform',\n 'config': {'minval': -0.05, 'maxval': 0.05, 'seed': None}},\n 'bias_initializer': {'class_name': 'Zeros', 'config': {}},\n 'kernel_regularizer': None,\n 'bias_regularizer': None,\n 'activity_regularizer': None,\n 'kernel_constraint': None,\n 'bias_constraint': None}},\n {'class_name': 'Dense',\n 'config': {'name': 'dense_1',\n 'trainable': True,\n 'dtype': 'float32',\n 'units': 1,\n 'activation': 'linear',\n 'use_bias': True,\n 'kernel_initializer': {'class_name': 'GlorotUniform',\n 'config': {'seed': None}},\n 'bias_initializer': {'class_name': 'Zeros', 'config': {}},\n 'kernel_regularizer': None,\n 'bias_regularizer': None,\n 'activity_regularizer': None,\n 'kernel_constraint': None,\n 'bias_constraint': None}}]}"}, "metadata": {}, "execution_count": 23, "output_type": "execute_result"}]}, {"metadata": {"id": "90c16b9e423c47ce964c97671ad947c6"}, "cell_type": "markdown", "source": "### Create a function to plot predicted vs actual values "}, {"metadata": {"id": "23c7df9f-6951-46c1-b5fc-1ce017d616a0"}, "cell_type": "code", "source": "def plot_the_results(predicted_data, true_data, prediction_len):\n fig = plt.figure(facecolor='white', figsize=(16,8))\n ax = fig.add_subplot(111)\n ax.plot(true_data, label='True Data')\n for i, data in enumerate(predicted_data):\n padding = [None for p in range(i * prediction_len)]\n plt.plot(padding + data, label='Prediction')\n plt.plot(padding + data, 'b^')\n plt.show()", "execution_count": 24, "outputs": []}, {"metadata": {"id": "3d9a954976f742d18b5579a009a72d92"}, "cell_type": "markdown", "source": "### Create a function to predict future values"}, {"metadata": {"id": "e69c634bc5764c5d836cb69e84e194d3"}, "cell_type": "code", "source": "def predict_the_sequences(model, data, window_size, prediction_len):\n prediction_seqs = []\n for i in range(int(len(data)/prediction_len)):\n curr_frame = data[i*prediction_len]\n predicted = []\n for j in range(prediction_len):\n predicted.append(model.predict(curr_frame[newaxis,:,:])[0,0])\n curr_frame = curr_frame[1:]\n curr_frame = np.insert(curr_frame, [window_size-1], predicted[-1], axis=0)\n prediction_seqs.append(predicted)\n return prediction_seqs", "execution_count": 25, "outputs": []}, {"metadata": {"id": "3f69fe49e53a4d828ea6a23ed34c417d"}, "cell_type": "markdown", "source": "### Predict future values & plot the results\n In this case, we are predicting the current values.\n If we need to predict t+1 then the prediction_len parameter has to be changed to 2\n and if we need t+2 then prediction_len would be 3"}, {"metadata": {"id": "b94874c3be1241ed8ee0743f30dc53f9"}, "cell_type": "code", "source": "predictions = predict_the_sequences(model, testX, look_back, 1)\n\nplot_the_results(predictions, testY, 1)", "execution_count": 26, "outputs": [{"data": {"text/plain": "
", "image/png": "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\n"}, "metadata": {}, "output_type": "display_data"}]}, {"metadata": {"id": "f20574d659d248acb8e22d7c48aa4766"}, "cell_type": "markdown", "source": "We can observe that the model is able to catch the pattern in the data. This can be further improved by changing the hyper parameters however we are demonstrating the methodology."}, {"metadata": {"id": "9b33aead24da44808d524565de55eaee"}, "cell_type": "markdown", "source": "### Predictions for next 7 days"}, {"metadata": {"id": "63c0f4630da94826bd8a12670d7e3462"}, "cell_type": "code", "source": "predictions7 = predict_the_sequences(model, testX, look_back, 7)", "execution_count": 27, "outputs": []}, {"metadata": {"id": "9bec4249-6b91-4c19-a840-f11f80fdf20f"}, "cell_type": "markdown", "source": "### Denormalize the predicted values and save the data\nDenormalize & Convert the predicted output to a dataframe & print the results"}, {"metadata": {"id": "9b5f9ee0e5d74bd09bc47644a247ddfc"}, "cell_type": "code", "source": "predictionValues = scaler.inverse_transform(predictions7)\nresults = pd.DataFrame(np.round(predictionValues[-1:]))", "execution_count": 28, "outputs": []}, {"metadata": {"id": "a759ad93ab5845de928683260dd1152a"}, "cell_type": "code", "source": "results_list = results.values.tolist()", "execution_count": 29, "outputs": []}, {"metadata": {"id": "e87293121b334aa2918637bdb3ca572c"}, "cell_type": "code", "source": "results_list = [int(i) for i in results_list[0]]", "execution_count": 30, "outputs": []}, {"metadata": {"id": "28b2a5d9bef341f6b2a01380e1ac3f7b"}, "cell_type": "code", "source": "last_date = data_df_1.iloc[-1].name\nlastDate = datetime.datetime(int(last_date.split('/')[2]), int(last_date.split('/')[1]), int(last_date.split('/')[0]))\nnext_date = lastDate + datetime.timedelta(days=1)\nnext_date = next_date.strftime('%d/%m/%y')", "execution_count": 31, "outputs": []}, {"metadata": {"id": "3ec0523b96834fdf8a6e7eaef77a1e5f"}, "cell_type": "code", "source": "next_7_days = pd.date_range(start=next_date, periods=7, freq='D')\nnext_7_days_df = pd.DataFrame({'DATE': next_7_days.tolist(), 'Prediction': results_list})", "execution_count": 32, "outputs": []}, {"metadata": {"id": "a2314aa4a9cb47428fd391ecfa6206cc"}, "cell_type": "code", "source": "next_7_days_df['DATE'] = next_7_days_df['DATE'].dt.strftime('%d/%m/%y')", "execution_count": 33, "outputs": []}, {"metadata": {"id": "270406a57eba418d87283de44d9ca160"}, "cell_type": "code", "source": "next_7_days_df.set_index('DATE', inplace=True)", "execution_count": 34, "outputs": []}, {"metadata": {"id": "c4ff1899b636481e9753ae3d7569c58c"}, "cell_type": "code", "source": "finalDf = pd.DataFrame(data_df_1.tail(7))", "execution_count": 35, "outputs": []}, {"metadata": {"id": "144c727f1c8f4d5683f719895cd3e8a4", "scrolled": true}, "cell_type": "code", "source": "next_7_days_df.reset_index()", "execution_count": 36, "outputs": [{"data": {"text/plain": " DATE Prediction\n0 26/03/22 1859\n1 27/03/22 1964\n2 28/03/22 1974\n3 29/03/22 1994\n4 30/03/22 2035\n5 31/03/22 2099\n6 01/04/22 2176", "text/html": "
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DATEPrediction
026/03/221859
127/03/221964
228/03/221974
329/03/221994
430/03/222035
531/03/222099
601/04/222176
\n
"}, "metadata": {}, "execution_count": 36, "output_type": "execute_result"}]}, {"metadata": {"id": "61c48b0533764d0d9772e25f133da03f", "scrolled": true}, "cell_type": "code", "source": "finalDf.reset_index()", "execution_count": 37, "outputs": [{"data": {"text/plain": " DATE Total_cases\n0 19/03/22 1222\n1 20/03/22 727\n2 21/03/22 2948\n3 22/03/22 2782\n4 23/03/22 2750\n5 24/03/22 1371\n6 25/03/22 33", "text/html": "
\n\n\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n
DATETotal_cases
019/03/221222
120/03/22727
221/03/222948
322/03/222782
423/03/222750
524/03/221371
625/03/2233
\n
"}, "metadata": {}, "execution_count": 37, "output_type": "execute_result"}]}, {"metadata": {"id": "d034f2a42e6a4619a4988ac18f38efaf"}, "cell_type": "code", "source": "next_7_days_df = next_7_days_df.astype({'Prediction': int})", "execution_count": 38, "outputs": []}, {"metadata": {"id": "f1b700b4cb4e49c3a4a09799d550c9c7"}, "cell_type": "code", "source": "finalDf = finalDf.astype({'Total_cases': int})", "execution_count": 39, "outputs": []}, {"metadata": {"id": "b9ddab06655d448681a3378f8ef30f1e"}, "cell_type": "code", "source": "next_7_days_df.to_csv('_buffer_next7.csv')\nfinalDf.to_csv('_buffer_actual7.csv')\nregionData.to_csv('_buffer_region.csv')", "execution_count": 40, "outputs": []}, {"metadata": {"id": "26a0f751e89c43b69c39802f8ecef8d0"}, "cell_type": "code", "source": "next_7_days_df = pd.read_csv('_buffer_next7.csv')\nfinalDf = pd.read_csv('_buffer_actual7.csv')\nregionData = pd.read_csv('_buffer_region.csv')", "execution_count": 41, "outputs": []}, {"metadata": {"id": "0c8d1f4b9ee1489189460e8111734a6b"}, "cell_type": "code", "source": "next_7_days_df['DATE'] = pd.to_datetime(pd.Series(next_7_days_df['DATE']), format=\"%d/%m/%y\")\nfinalDf['DATE'] = pd.to_datetime(pd.Series(finalDf['DATE']), format=\"%d/%m/%y\")\nregionData['DATE'] = pd.to_datetime(pd.Series(regionData['DATE']), format=\"%d/%m/%y\")", "execution_count": 42, "outputs": []}, {"metadata": {"id": "e15614f0129b430986e20b87d43d68f7"}, "cell_type": "code", "source": "finalDf = pd.concat([finalDf,next_7_days_df], axis=1)", "execution_count": 43, "outputs": []}, {"metadata": {"id": "22dff2d1fa67457591d140ddb2fae6f2"}, "cell_type": "markdown", "source": "### Store the prediction data to Amazon S3 and Cloud Pak for Data Project Assets"}, {"metadata": {"id": "3dc78ebed0d54c5c8e0ee87969638e7c"}, "cell_type": "code", "source": "filename = 'wallonia-next7Prediction.csv'", "execution_count": 44, "outputs": []}, {"metadata": {"id": "eb3abdcef497430e86715377d483d949"}, "cell_type": "code", "source": "originalData = \"Wallonia.csv\"", "execution_count": 45, "outputs": []}, {"metadata": {"id": "9893203e8878410d8c3d4a6c3e2754ed"}, "cell_type": "code", "source": "from ibm_watson_studio_lib import access_project_or_space\nwslib = access_project_or_space()\nAWS_S3_credentials = wslib.get_connection(AWS_S3_data_request['connection_name'])", "execution_count": 46, "outputs": []}, {"metadata": {"id": "f366caf305e240e98729f38e606a2989"}, "cell_type": "code", "source": "s3 = boto3.resource(\n service_name = \"s3\",\n region_name = AWS_S3_credentials['region'],\n aws_access_key_id = AWS_S3_credentials['access_key'],\n aws_secret_access_key = AWS_S3_credentials['secret_key']\n)", "execution_count": 47, "outputs": []}, {"metadata": {"id": "6cd04b37703348ae8514ee4352327cfd"}, "cell_type": "code", "source": "bucket = AWS_S3_credentials['bucket']\ncsv_buffer = StringIO()\nnext_7_days_df.to_csv(csv_buffer, index=False)\ns3.Object(bucket, \"model_output/\"+filename).put(Body=csv_buffer.getvalue())\ncsv_buffer = StringIO()", "execution_count": 48, "outputs": []}, {"metadata": {"id": "653bef89e5944897a8e89670734a03b5"}, "cell_type": "code", "source": "next_7_days_df.to_csv(filename, index=False)\n\nwith open(filename, 'rb') as z:\n data = io.BytesIO(z.read())\n project.save_data(filename, data, set_project_asset=True, overwrite=True)", "execution_count": 49, "outputs": []}, {"metadata": {"id": "50bb81b39b1b4010b42a7c04a69c1100"}, "cell_type": "code", "source": "regionData.to_csv(originalData, index=False)\n\nwith open(originalData, 'rb') as z:\n data = io.BytesIO(z.read())\n project.save_data(originalData, data, set_project_asset=True, overwrite=True)", "execution_count": 50, "outputs": []}, {"metadata": {"id": "e72ae7b8003b4e0e97cc7012c7971fdf"}, "cell_type": "markdown", "source": "### Store Model Loss Data to S3 and Assets"}, {"metadata": {"id": "43e0f53d3d0f47bda24e85d3b09207e0"}, "cell_type": "code", "source": "errorevaluation = pd.DataFrame({'index': [int(i) for i in range(len(history.history['loss']))], 'loss': history.history['loss'], 'val_loss': history.history['val_loss']})", "execution_count": 51, "outputs": []}, {"metadata": {"id": "4dfc1090967942df901ecced937dcc20"}, "cell_type": "code", "source": "errorFilename = 'wallonia-errorEvaluation.csv'", "execution_count": 52, "outputs": []}, {"metadata": {"id": "0bc8127ca5e244608d0aa1f4a8ccb8b5"}, "cell_type": "code", "source": "csv_buffer = StringIO()\nerrorevaluation.to_csv(csv_buffer, index=False)\ns3.Object(bucket, \"model_output/\"+errorFilename).put(Body=csv_buffer.getvalue())", "execution_count": null, "outputs": []}, {"metadata": {"id": "fb1c15e1c56d479ab91adba5d6b54e3d"}, "cell_type": "code", "source": "errorevaluation.to_csv(errorFilename, index=False)\n\nwith open(errorFilename, 'rb') as z:\n data = io.BytesIO(z.read())\n project.save_data(errorFilename, data, set_project_asset=True, overwrite=True)", "execution_count": 54, "outputs": []}, {"metadata": {"id": "1afdebecb5ef46b28d545e4979eb472c"}, "cell_type": "markdown", "source": "### Store Actual vs Predicted Data to S3 and Assets"}, {"metadata": {"id": "54206734ea0c44ada75ee4d7694be9d5"}, "cell_type": "code", "source": "actualVsPredicted = pd.DataFrame({'index': [int(i) for i in range(len(predictions))], 'actual': testY, 'prediction': list(itertools.chain.from_iterable(predictions))})", "execution_count": 55, "outputs": []}, {"metadata": {"id": "08c49c7512e74069879864a83d9aeb01"}, "cell_type": "code", "source": "actualVsPredictedFilename = 'wallonia-actualVsPredicted.csv'", "execution_count": 56, "outputs": []}, {"metadata": {"id": "a2315f06a55f4fd3846f43240a8e995b"}, "cell_type": "code", "source": "csv_buffer = StringIO()\nactualVsPredicted.to_csv(csv_buffer, index=False)\ns3.Object(bucket, \"model_output/\"+actualVsPredictedFilename).put(Body=csv_buffer.getvalue())", "execution_count": null, "outputs": []}, {"metadata": {"id": "df4e1ddeff384f498e1de9330d239ae4"}, "cell_type": "code", "source": "actualVsPredicted.to_csv(actualVsPredictedFilename, index=False)\n\nwith open(actualVsPredictedFilename, 'rb') as z:\n data = io.BytesIO(z.read())\n project.save_data(actualVsPredictedFilename, data, set_project_asset=True, overwrite=True)", "execution_count": 58, "outputs": []}, {"metadata": {"id": "ca8848ca75004e108771d8c617a73a7f"}, "cell_type": "markdown", "source": "

Want to learn more?

\n\nThe AutoAI graphical tool in Watson Studio analyzes your data and discovers data transformations, algorithms, and parameter settings that work best for your predictive modeling problem. AutoAI displays the results as model candidate pipelines ranked on a leaderboard for you to choose from.\n\nAlso, you can use Watson Studio to run these notebooks faster with bigger datasets. Watson Studio is IBM's leading cloud solution for data scientists, built by data scientists. With Jupyter notebooks, RStudio, Apache Spark and popular libraries pre-packaged in the cloud, Watson Studio enables data scientists to collaborate on their projects without having to install anything. Join the fast-growing community of Watson Studio users today with a free account at Watson Studio\n\n

Thanks for completing this Lab!

\n\n

Author: Manoj Jahgirdar & Sharath Kumar RK

\n\n"}], "metadata": {"kernelspec": {"name": "python3", "display_name": "Python 3.9", "language": "python"}, "language_info": {"name": "python", "version": "3.9.7", "mimetype": "text/x-python", "codemirror_mode": {"name": "ipython", "version": 3}, "pygments_lexer": "ipython3", "nbconvert_exporter": "python", "file_extension": ".py"}}, "nbformat": 4, "nbformat_minor": 1}