A Multi-Class Traffic Assignment Model for Predicting Transit Passenger Flows - A Case Study of Beijing Subway Network

This paper describes a case study comparing a multi-class transit assignment model with its single class counterpart for estimating the passenger flows of the Beijing subway network – one of the largest railway transit networks in the world. Multi-class traffic assignment has been widely considered as a theoretically sound approach to capturing the inherent variation in users' route choice behavior. However, few empirical studies have been devoted to showing the effectiveness of this approach in improving the accuracy of the underlying passenger flow estimation process. In this research, a passenger classification scheme is proposed on the basis of a data set from a large stated preference survey conducted in the City of Beijing, China. Separate generalized cost functions are calibrated for different classes of subway users in Beijing and applied in a multi-class transit assignment model for estimating passenger flows over a subway network. The case study has shown that the proposed multi-class approach resulted in significantly improved estimation results with an average estimation error of less than 15% on the transfer flows as compared to 30% for the single class model.