Railway Passenger Flow Forecast Based on Hybrid PVAR-NN Model

Abstract

Rail transportation is the backbone of modern transportation. Accurate railway passenger flow forecasting can be applied to support transportation system management such as operation plan and route selection design. This paper proposes a hybrid linear + nonlinear time series analysis model, which uses the panel vector autoregression (PVAR) and neural network (NN) hybrid PVAR-NN prediction methods to predict passenger flow in the railway system. The proposed model combines the pros of both linear and non-linear model with easy-to-interpretation for stakeholders. The empirical analysis results further indicate that the proposed hybrid PVAR-NN approach performs with improved accuracy in forecasting the railway passenger flow.