ABSTRACT
This chapter presents a brief introduction to the problem of trip production, and explains the database used for modelling. It presents a study on the effect of varying the Learning Rate (LR) and the Momentum Constant (MC) values on the network training time and the prediction accuracy of the trained models. The chapter applies Backpropagation artificial neural networks (ANNs) to modelling trip production, a part of the four step travel demand forecasting phase in the urban transportation planning procedure. The objective of the study is to demonstrate the importance of the choice of the LR and the MC toward developing effective ANN models. The study also demonstrates a procedure for selecting the most appropriate values of these two parameters. The results from the analysis of the training time with respect to the MC values indicate that the networks would train fastest for MC values between 0.2 and 0.4.
