Integrated Generalized Cost Model Considering the Whole Trip for Urban Rail Transit Station Choices

Abstract

Convenient transfer between urban rail transit (URT) and feeder modes could be fundamental for improving the URT service. In the field of URT, most studies have attempted to estimate the station choice of sub-trips like access trips and egress trips rather than the whole trip, which may be locally optimal and inconsistent with the traveler’s behavior of maximizing the utility of the whole trip. In this study, we build an integrated generalized cost model for URT to estimate the access and egress station choices of URT travelers from the perspective of the whole trip and apply a genetic algorithm to estimate the model parameters. Data used in this study include a revealed preference (RP) mode chain survey conducted in Nanjing, China, and the travel cost data obtained from an online navigation application program interface (API), thus reducing the transit networking modeling work as the conventional traffic assignment process. The proposed model has promising station choice prediction accuracies for whole trips. The results show that: (a) travelers do not always choose the nearest station, and the whole trip’s generalized cost would affect their choices; (b) travelers make heterogeneous station choices. They have a varied perception of time and fare and are more sensitive to access trips; and (c) in forecasting-oriented applications, the model parameters can be set flexibly, and the model can maintain acceptable forecasting accuracy. The findings contribute to understanding station choice mechanism, transfer optimization, and feeder demand forecast for stations.