DEVELOPMENT OF A POLICY SENSITIVE MODEL OF TRANSIT ACCESS USING ARTIFICIAL NEURAL NETWORKS

Abstract

Transportation professionals face challenges of increasing complexity to meet the goals of providing safe, efficient, reliable, and environment friendly transportation. Travel demand forecasts, incident management programs, traffic volume forecasts, traffic-flow forecasts, traffic control systems, intersection capacity, safety, freight transportation, pavement maintenance, metro operations, etc., are some of the key elements of transportation system. The efficiency of transportation system directly depends on the reliability of the methods adopted to analyze and predict these key elements. In recent years, there has been increased interest among both transportation researchers and practitioners in exploring the feasibility of applying artificial neural networks (ANNs) to improve analyzing and predicting methodologies. In the presented work, ANNs were applied for access mode choice modeling. Three different topologies of ANNs were built; MLP (Classifier), ANNs resembling the logit analysis, and ANN resembling the nested logit analysis. The results obtained for these networks were compared with the results of logit and nested logit model. From the comparison it was found that the prediction success rate of ANN is relatively high.