def evaluate_model(Model, X_Train_REFINED, Y_Train, X_Val_REFINED,
Y_Validation, X_Test_REFINED, Y_Test):
ES = EarlyStopping(monitor='val_loss', mode='min', verbose=0, patience=30)
History = Model.fit(X_Train_REFINED,
Y_Train,
batch_size=128,
epochs=250,
verbose=0,
validation_data=(X_Val_REFINED, Y_Validation),
callbacks=[ES])
y_pred = Model.predict(X_Test_REFINED)
CNN_NRMSE, CNN_R2 = NRMSE(Y_Test, y_pred)
print('NRMSE > %.3f' % (CNN_NRMSE))
return CNN_NRMSE, History
callbacks=[ES])
Y_Val_Pred_CNN = model.predict(CNN_Val, batch_size=128, verbose=0)
Y_Pred_CNN = model.predict(CNN_Test, batch_size=128, verbose=0)
Y_Val_Save[:, cnt + 1] = Y_Val_Pred_CNN.reshape(-1)
Y_Test_Save[:, cnt + 1] = Y_Pred_CNN.reshape(-1)
#print(model.summary())
# Plot the Model
# plt.plot(CNN_History.history['loss'], label='train')
# plt.plot(CNN_History.history['val_loss'], label='Validation')
# plt.legend()
# plt.show()
# Measuring the REFINED-CNN performance (NRMSE, R2, PCC, Bias)
CNN_NRMSE, CNN_R2 = NRMSE(Y_Test, Y_Pred_CNN)
MAE = mean_absolute_error(Y_Test, Y_Pred_CNN)
print(CNN_NRMSE, "NRMSE of " + modell + SEL_CEL)
print(CNN_R2, "R2 of " + modell + SEL_CEL)
Y_Test = np.reshape(Y_Test, (Y_Pred_CNN.shape))
CNN_ER = Y_Test - Y_Pred_CNN
CNN_PCC, p_value = pearsonr(Y_Test, Y_Pred_CNN)
print(CNN_PCC, "PCC of " + modell + SEL_CEL)
Y_Validation = Y_Validation.reshape(len(Y_Validation), 1)
Y_Test = Y_Test.reshape(len(Y_Test), 1)
Bias = Bias_Calc(Y_Test, Y_Pred_CNN)
Results_Data[0, :] = [CNN_NRMSE, MAE, CNN_PCC, CNN_R2, Bias]
cnt += 1
Results = pd.DataFrame(data=Results_Data,