Incorporating travel behavior regularity into passenger flow forecasting

Accurate forecasting of passenger flow (i.e., ridership) is critical to the operation urban metro systems. Previous studies mainly model as time series by aggregating individual trips and then perform based on values in past several steps. However, this approach essentially overlooks fact that consists from each traveler. For example, a traveler’s work trip morning can help predict his/her home evening, while causal structure cannot be explicitly encoded standard models. In paper, we propose new framework for boarding incorporating generative mechanism into models leveraging strong regularity rooted travel behavior. doing so, introduce returning previous alighting covariate, which captures long-range dependencies data We develop return probability parallelogram (RPP) summarize relationships estimate flow. The proposed evaluated using real-world data, results confirm flow—a single covariate—can substantially consistently improve various tasks, including one-step ahead forecasting, multi-step under special events. And method more effective business-type stations with most passengers come within same day. This study extended other modes transport, it also sheds light general demand problems, are generated user