Mixture distribution model of departure time decision for railway commuters

Kota Miyauchi, Kazuyuki Takada

Last modified: 2019-03-04

Abstract


In recent years, many studies of travel time reliability energetically are performed. However, these studies didn’t focus on considering heterogeneity between individuals. In this study, the research of considering heterogeneity between individuals which focused on railway commuters in Tokyo metropolitan was performed.
First, it was performed online survey and collected data about the commuter situation in the railway. As a result of the online survey, the railway commuters recognize train travel time as uncertain and tend to decide the departure time considering buffer time. As a result, the arrival situation of railway commuters became variety, it is difficult to arrive on their ideal arrival time. To define arrival situation, the arrival distribution was constructed to their arrival situation. This distribution consists of three types of considering the delay recognition. In this study, the arrival distribution for railway commuters assumed that mixture distribution. The divided ratio of mixture distribution is different every railway commuter.
Next, using this result, the factors of buffer time of railway commuters estimated by MCMC method. MCMC method could estimate the heterogeneity between individuals. As a result of estimation, the factor of increase buffer time was travel distance(km), number of transfers(times), and the experience time encountering train delay (min/week). The application of mixture distribution model was verified by this study.

Keywords


travel time reliability, mixture distribution, optimal departure time

Full Text: PDF