An approach for real-time estimation of railway passenger flow

The ability to obtain passenger flow information in real time is expected to prove useful in application to railway traffic operations. In this report, forecasting of the number of gate passages according to the time of day was attempted using past data from automatic ticket-checking gates, and the number of passages by each origin station was estimated using the number of gate passages. As attributes of daily passage data were apparent, two prediction approaches were found to be applicable to stable data and irregular data. Consequently, the applicability criteria for each approach were clarified. Moreover, accurate forecasting of the number of OD (combination of Origin and Destination) passages was performed using the forecasting model developed.