Modeling Day-to-Day Modes Choices with Heterogeneous Travelers under MaaS Scenarios

Abstract

MaaS (Mobility as a Service) platforms allow travelers to choose from various mode combinations, altering their behaviors and decision-making. This paper introduces a dynamic mode choice evolution model for heterogeneous users, focusing on long-term learning. The model adjusts choices based on personal experience and historical data, incorporating a value of time (VOT) function to distinguish between car and non-car owners. Based on Beijing's MaaS scenario, results indicate that learning from past experiences accelerates convergence, with the learning parameter affecting the rate but not the final equilibrium. The impact of transfer times for combined modes on mode choices for both car and non-car owners is analyzed. Non-car owners are less sensitive to VOT in their choices, while car owners show more heterogeneity in their behaviors. Therefore, providing uniform incentives for non-car owners can promote public transport, while differentiated VOT-based pricing for car owners can reduce private car use.