Modeling Commuters’ Travel Behavior by Bayesian Networks

Abstract Previous studies indicate that residential location and commute distance may influence individual's travel behavior, but most models are limited to capture the internal relationship. In this paper, the methodology of Bayesian networks is introduced. With the combination of network structure and conditional probability table, Bayesian networks are capable of capturing the uncertainty nature. Moreover, the data is based on Fuyang Resident Travel Survey in 2012. The outputs illustrate that commute distance influences commuters’ travel mode choice directly while the residential location doesn’t. The experimental results show that different groups of people have different reactions to distance on mode.