Estimation of annual average daily traffic volumes using Neural Networks

This study compared the estimations of annual average daily traffic (AADT) volumes using the conventional method(Factors), multiple regression analysis, and the neural network approach. All three approaches were compared using three different classification schemes as well as different duration of traffic counts. The neural network and multiple regression approaches consistently performed better than the conventional approach, and the neural network approach in many cases slightly outperformed the multiple regression approach. Apart from providing a good modeling tool for estimating AADT, the results also provide useful insight into the duration of the short term traffic counts and the classification schemes for the highway sites.