Abstract
In recent years, travel time reliability of urban railway in Tokyo metropolitan area has decreased and it becomes a social problem in transportation system in this area. There were many studies investigating this problem from various aspects. This study investigated a statistical distribution of the gap between desired arrival time and actual arrival time of railway commuters. Online survey was conducted to obtain the information on the arrival situation such as arrival early, arrival on time, and arrival late. The distribution of arrival situation was identified by applying Kernel density estimation method. Moreover, expected utility regarding the departure time decision was formulated and it was determined. Meanwhile, the departure time was dependent on the amount of buffer time. Therefore, statistical distribution of the buffer time was estimated by applying the Markov chain Monte Carlo simulation method.
