Factors Affecting Performance of Parametric and Non-parametric Models for Daily Traffic Forecasting

Traffic forecasts are used for a wide variety of purposes from the planning to the design and operational stages of the highway network. The forecasting models need the historical traffic data and some supporting variables that are relevant to the traffic demand. Apart from that, choice of an appropriate model or technique is also an important consideration. This paper gives an overview of the traffic forecasting process and the models that are used for this purpose with emphasis on the use of different types of Artificial Neural Networks (ANNs). In this research, two types of Artificial Neural Networks (ANNs) are being compared with the traditional parametric technique of linear regression analysis for daily traffic forecasting on King Fahd causeway, which provides a link between Saudi Arabia and Bahrain. It was observed from the estimated error values that ANNs have better accuracy than linear regression technique for predicting daily traffic. However; increasing the size of dataset and restructuring of the dataset showed the greatest effect on predictive accuracy of the models instead of the configuration of the model or type of technique. In fact, it was also observed that providing large number of classified data samples can make the accuracy of the regression analysis comparable to ANNs