THE IMPACT OF DATA QUANTITY ON THE PERFORMANCE OF NEURAL NETWORK FREEWAY INCIDENT DETECTION MODELS. IN: NEURAL NETWORKS IN TRANSPORT APPLICATIONS

This paper discusses one of the difficulties in the development of artificial neural network (ANN) models: the inability to determine in advance the number of samples or observations needed for training ANN models. A further complication recognized is that when dealing with 'real world', data is not easily available or is difficult and time consuming to collect. The paper reports on a neural network freeway incident detection model that was developed using 'real world' incident and traffic data. The results reported in this paper can be used to make decisions about the sample size required for retraining the ANN incident detection model once it becomes out of date due to changed traffic conditions and/or upgrading the facility.