Short-term forecasting of transit route OD matrix with smart card data

This paper presents a short-term forecasting approach on transit route OD (Origin-Destination) matrix using smart card data of buses. Temporal characteristics of stop-level boardings/alightings and route-level number of passengers are analyzed. A three-step methodology that provides short-term forecasts of transit route OD matrices is proposed. The proposed method is applied to bus route 966 in Beijing as a case study. The forecasted results are compared with the observed data. The proposed short-term forecasting approach has a mean absolute percentage error no more than 13.05% or lower and a maximum absolute error no more than 17 suggesting the usefulness of the proposed method to support the deployment of dynamic urban transportation management systems.