RECENT ADVANCES IN TRAVEL CHOICE BEHAVIOR MODELING

Abstract

This special issue “Travel Choice Behavior Modeling” is organized based on selected papers presented at the International Association of Travel Behavior Research (IATBR) in Kyoto, Japan, 2006. Great strides have been made recently in travel choice behavior modeling (Bovy and Fiorenzo-Catalano, 2007; Beale and Bonsall, 2007; Chorus et al., 2007; Han et al., 2007; Kim et al., 2007; Tsirimpa et al., 2007). New approaches have been developed to augment and improve existing ones or to seek new modeling paradigms (Bonsall, 2004; Senbil and Kitamura, 2004; Jou et al., 2008; Arentze and Timmermans, 2007). It is timely to take inventory of what we have so far and provide a preview of what lies ahead. Travel involves choices of destination, departure time, mode, and route. Typically, travelers face a large set of alternatives, often with imperfect information about them. How to generate reasonable alternatives and therefore avoid burdening the modeling effort with unnecessary ones; how to develop and include new explanatory variables beyond the traditional socio-economic ones, such as the influence of travel information, marketing, etc; how to take advantages of revealed preference (RP) and stated preference (SP) surveys to improve the accuracy of travel choice models; and how to consider correlated choices in the presence of taste heterogeneity, etc are just some examples of recent advancements. This special issue is aimed to cover some recent methodological advances in travel choice behavior modeling. Five papers are collected in this special issue, as summarized below. Defining the choice sets appropriately is important for analyzing choice behavior, estimating parameters of choice models, or predicting choice probabilities. This, however, is not a simple task. The paper by Van Nes, Hoogendoorn-Lanser and Koppelman investigates the relationship between the type of choice set and the purpose of the analysis, for estimation or prediction. The authors differentiate choice sets as being “objective” (generated) or “subjective” (observed). The study concludes that for estimation purposes, observed subjective choice sets are preferable; on the other hand, generated objective choice sets appear to be most suitable for prediction purposes. The paper by Sakano and Benjamin incorporates both RP and SP data in the structural equations model to analyze activity-based travel behavior. The model is then applied to examine the relationship between activities and mode choice for the Puget Sound Regional Council in Seattle, USA. The study finds that the attitude toward comfort of travel, car requirement for job, and travel time are significant predictors for the mode choice decision. The authors conclude that the RP-SP model can help improve the fit of the model over the RP-alone model and that the model developed can be extended to the analysis of more complex decision-making processes involving more detailed information on modes, activities, and times. Despite substantial recent efforts in improving route choice models, few of them are integrated into traffic assignment studies. The paper by Bekhor, Toledo, and Prashker