Estimation of the arrival distribution of the railway commuters by applying mixture distributions

Abstract

Recently, the delay of railway operation chronically occurs in Tokyo metropolitan area and many studies focused on the travel time reliability of railway operation from various points of view. However, there are few studies analyzing the actual arrival situation. Therefore, this study focused on arrival situation of railway commuters. An internet survey was conducted in order to obtain the data regarding not only departure situation but also arrival situation. At first, properties of departure and arrival situations were analyzed and it became clear that plural distributions are mixing. Then, this study tried to estimate arrival distribution model by applying mixture distribution model. It makes the arrival situation of the railway commuters more useful using both normal distribution and the exponential distribution than it uses the only one distribution. As the results of parameter estimation, mixture model containing two normal distributions and one exponential distribution have highest reproducibility among the estimated mixture distribution models.