Short-term Prediction of Passenger Flow Based on Self-organising Data Mining

Abstract

Accurate passenger flow prediction is critical to managing the use of transportation infrastructures and preventing emergencies resulting from large crowds. Consequently, short-term flow forecasting technology has become increasingly significant in the field of intelligent transportation systems. Passenger flow forecasting is based on historical data that provide access to personalised travel information, from which a corresponding prediction model can be constructed to improve prediction accuracy. A group method of data handling (GMDH)-based short-term passenger flow forecast model that predicts the inbound passenger volume of urban rail transit stations based on self-organising data mining is proposed in this study. In the proposed model, passenger flow is divided into different categories according to the regularity of metro passenger travel. To evaluate the performance of the proposed model, it was applied to predict inbound passenger flow of Beijing metro system based on consecutive 35-day historical automatic fare collection (AFC) data. The results obtained indicate that the proposed model is more accurate than existing prediction models Back Propagation Neural Network (BPNN) and Elman Neural Network (ENN) as it has lower mean absolute percentage error (MAPE).