Continuous adaptive prediction of highway traffic volumes using artificial neural networks

This report details the investigation into the effect of several artificial neural network (ANN) architectures on the ability of these networks to predict highway traffic volumes (HTVs). The scope of this project is intended as a basis for further research by determining favourable choices for an underlying network architecture which may be expanded to incorporate recurrent temporal sensitive features. This will be determined on the basis of the errors that the predicted traffic volumes deviate from the target values in the training set. The training phase of the models will exclude data for up to a year to allow for the performance evaluation of the trained networks.