Multi-scenario Route Choice Modeling Based on Random Parameters Multinomial Logit Model with Heterogeneity in Means and Variance

Travel demand management (TDM) strategies have been widely adopted to tackle the persisting traffic congestion in metropolitan areas [1]. To evaluate the effects of TDM strategies, it is imperative to disentangle the dynamics of individual travel behaviors, usually through developing route choice models. Based on long-term trajectory data from passenger vehicles, this study investigates individual route characteristics, travel features, and traveler attributes, to explore route choice behaviors for respectively morning and evening travel during weekdays. By categorizing the choice set of each trip into five alternatives based on different routing strategies, we propose a new route choice model that accounts for route overlap and incorporates random parameter heterogeneity in means and variances. This model demonstrates superior fitting performance compared to Path-Size Multinomial Logit and Mixed Logit models. Furthermore, our findings reveal temporal disparities in route choice behavior, providing valuable insights for the implementation of personalized measures that aim at guiding travelers to change their departure time or modify their route choices.