EMPIRICAL ANALYSIS OF TRAVEL DESTINATION CHOICE WITH BAYESIAN METHODS, A CASE STUDY OF JILIN,CHINA

摘要

This paper presents a Bayesian method for travel destination choice of urban residents. We describe a Bayesian approach for learning Bayesian networks from a combination of prior knowledge and statistical data, which is derived from an inhabitant trip survey in Jilin, China. First, a methodology for assessing informative priors needed for Bayesian network learning is expounded. Second, we illustrate discrete choice model of predicting travel destination choice. Third, under the help of the urban travel data from an urban traffic area in Jilin, China, we do a case study on inhabitant destination choice with Bayesian methods based on discrete decision model. A simulation model is established to explain the many factors that affect the destination choice of the residents. We also can use Bayesian networks to analyse how many factors can affect the destination choice, and the relationship between the factors. Finally, we describe a methodology for evaluating Bayesian network learning algorithms, and apply this approach to a comparison of various approaches. We analyse the prediction results which have a higher prediction accuracy from the disaggregate level.