Aoi Watanabe, Toyota Labs

"Perturbed Utility Models: Representing Zero Flow in Stochastic User Equilibrium and Estimating Choice Sets"


In the evolving field of travel behavior analysis, the perturbed utility model, especially with its integration of the additive random utility model, is gaining attention. This study introduces the αPURC (Perturbed Utility Route Choice) model, employing α-Tsallis entropy for its perturbation function, paired with the αSUE (stochastic user equilibrium). Differing from traditional logit-based SUEs that assign non-zero flows even to inefficient routes, leading to impractical traffic distributions and potential biases in traffic assignment, the αPURC model, encapsulating the Multinomial Logit (MNL) model, represents zero probabilities in choice sets. This capability of αSUE allows for unique and simultaneous determination of choice sets and equilibrium flows, thus mitigating existing biases. The αPURC model further enhances the understanding of travel behavior by enabling direct estimation of travelers' route choice sets from their selection outcomes, paving the way for more effective traffic management and urban planning strategies.

Contact person: Mogens Fosterau