Some Findings on the Application of Disaggregate Nested Logit Model to Railway Station and Access Mode Choice

This paper constructed nested logit (NL) models with three level structure and test prediction accuracy as well as temporal stability of the models. The authors present three main findings in the paper. Using the before-and-after data on the new station prepared by the same method on the same sample, the authors found the same choice structure is to hold the same set variables and their coefficients are temporarily stable. The estimation flow of the station usage containing a lot of simplification proved to be practical and useful in terms of the reproduction and prediction accuracy. The predictive error on the usage of the new station is proved to be within 20 percent. As a result, the authors concluded that the NL models developed in the paper can be useful with reasonable accuracy to analyze the station and access mode choice behavior as well as main mode choice behavior.