Modeling dynamic travel mode choices using cumulative prospect theory

Travelers’ mode choice behavior is jointly influenced by their attributes, such as personal income, vehicle ownership, and travel purpose, and external factors, such as the built environment and traffic state. The mode chosen by a traveler before a trip may change during the trip due to uncertainties associated with transportation systems. A cumulative prospect theory-based dynamic mode choice model that describes the changes in decisions at different travel stages is proposed to investigate the within-trip mode-shifting behavior during daily travel. The journey duration or travel time of a trip is divided into a series of stages, which includes the access or connection time (i.e., the time required for the traveler to reach the station or vehicle), waiting time, in-vehicle time, and parking time. The generalized travel cost includes the expense cost, time cost, and penalty cost. The penalty cost is expressed as a negative exponential function with respect to the punctuality satisfaction rating. A novel method is proposed for calculating the mode choice possibility in terms of the changes in travel cost during a trip, in which the psychological weights at different stages are reflected by the ratio of the actual travel time to the expected travel time. Data from a journey survey involving 637 respondents in Beijing were used to test the model. The results of a random parameter logit model reveal the main influencing factors of each travel mode, and show that economic strategies can influence the mode choice of 77.94% of travelers. The dynamic mode choice model predicted the pre-trip mode and mode shift with accuracies of 90.05% and 67.86%, respectively, both of which are higher than those of the expected utility theory-based model. Moreover, the results show that mode-shifting behaviors during a trip generally occur in public transit mode and/or active travel mode. The study results demonstrate the dynamic characteristic and quantitative change in mode choice behavior in multi-mode daily travel and are beneficial for the study of travel behavior in relation to the traffic environment.