nonlinear_regression_keras_createModel.py

# Create model
model = Sequential()
model.add(Dense(128, activation="relu", input_dim=6))
model.add(Dense(32, activation="relu"))
model.add(Dense(8, activation="relu"))
# Since the regression is performed, a Dense layer containing a single neuron with a linear activation function.
# Typically ReLu-based activation are used but since it is performed regression, it is needed a linear activation.
model.add(Dense(1, activation="linear"))

# Compile model: The model is initialized with the Adam optimizer and then it is compiled.
model.compile(loss='mean_squared_error', optimizer=Adam(lr=1e-3, decay=1e-3 / 200))

# Patient early stopping
es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=200)

# Fit the model
history = model.fit(X1, Y1, validation_data=(X2, Y2), epochs=10000000, batch_size=100, verbose=2, callbacks=[es])

# Calculate predictions
PredTestSet = model.predict(X1)
PredValSet = model.predict(X2)

# Save predictions
numpy.savetxt("trainresults.csv", PredTestSet, delimiter=",")
numpy.savetxt("valresults.csv", PredValSet, delimiter=",")