Efficient planning of long-distance rail services requires appropriate forecasting of passengers' demand fluctuation based on historical ridership records. However, this forecast is difficult, because the records consist of a mixture of passengers' demand variations. An effective approach for achieving an appropriate forecast is to decompose it into several independent demand variation components and forecast each of them. This study applies the independent component analysis to decompose the fluctuations into several additive variation components. Then, a forecasting method for passengers' demand fluctuation is developed using the calendar structure and event venue capacity.
J-STAGEがリニューアルされました！https://www.jstage.jst.go.jp/browse/-char/ja/
