DEMAND FORECASTING MODEL FOR RAILROAD USING ROUTE SEARCH HISTORY DATA

Abstract

Optimal operation of demand-responsive transport and mobility services requires the accurate prediction of variable transportation demands, rather than the average transportation demand. This study focuses route search data for public transportation. The data on routes searched by prospective travelers on public transportation route search systems while planning trips can be indicative of intended travel routes in the near future, which may be used to predict fluctuations in transportation demands. In this context, this study proposes a model to forecast daily fluctuations in transportation demands based on route search data and verifies its effectiveness. The Kagawa Prefecture and the “Kotoden” rail system are selected as the target region and target transportation system, respectively. In this study, in addition to route search data, traffic IC card data is used to estimate the number of public transportation users during empirical analysis. Specifically, 1) a bivariate state-space model is used to identify fluctuations in the number of route searches and the number of uses of traffic IC cards, and 2) a weighted regression model is used to quantify the relationships between the fluctuations. The prediction accuracy of the proposed model is compared with those of existing methods, revealing an improvement. In addition, the high accuracy of traffic demand prediction achieved using the proposed method demonstrates the effectiveness of the proposed model and shows the useful of route search data as an indicator of variation in traffic demand.