2

np.sqrt(((dataY_plot[-testX.size()[0]:] - data_predict[-testX.size()[0]:] ) ** 2).mean())

3

MAE = mean_absolute_error(LSTM_test_outputs, predictions)

4

def model_score(model, X_train, y_train, X_test, y_test):
    trainScore = model.evaluate(X_train, y_train, verbose=0)
    print('Train Score: %.5f MSE (%.2f RMSE)' % (trainScore[0], math.sqrt(trainScore[0])))

    testScore = model.evaluate(X_test, y_test, verbose=0)
    print('Test Score: %.5f MSE (%.2f RMSE)' % (testScore[0], math.sqrt(testScore[0])))
    return trainScore[0], testScore[0]


model_score(model, X_train, y_train, X_test, y_test)

rmse

def return_rmse(test,predicted):
    rmse = math.sqrt(mean_squared_error(test, predicted))
    print("The root mean squared error is {}.".format(rmse))