A Bayesian spatial–temporal model for predicting passengers occupancy at Beijing Metro

The growing population density in cities requires urban transportation to meet the travel needs of citizens fast and accurately. Therefore, the correct prediction of daily passenger flow in urban subway transportation is of great practical importance for rationalizing the traffic arrangement and safely responding to unexpected passenger flow. This work builds a Bayesian spatial–temporal model for predicting station occupancy. The proposed one provides point estimations of daily passenger flow, a reliable assessment of their uncertainty, and the possibility of understanding traffic features. It also provides a prediction accuracy that meets the standards of the Beijing Metro enterprise. The model discussed in this paper is in force at Beijing Metro Group Ltd to programme daily train schedules.