Table 6.

Pseudocode of the ANN metamodel developed and investigated in this study.

PROCEDURE	ARTIFICIAL NEURAL NETWORK
BEGIN
	Split dataset into training set and validation set
	Scale input values for the training set
	Scale the output values for ‘Vent_XY’ dataset
	Initialise parameter W and b
	Define the epoch, learning rate, and number of hidden neurons
REPEAT
	Forward propagation
	Compute the loss using Root Mean Square Error
	Calculate the gradient of loss
	Calculate new W and b using the gradient and update the parameters
UNTIL	Max epoch has reached
RETURN	Trained model with updated parameters (W, b)
	Trained model is used to predict the test dataset
	RMSE of the model prediction and real value is presented
END