Short-term Traffic Flow Forecasting Based on a Three-regime SETAR Model

Most existing literature about short-term traffic flow forecasting only consider traffic flows under normal or non-incident conditions,so a three-regime self-exciting threshold autoregressive(SETAR) model was employed to forecast short-term traffic flow on urban road because the structure of the model is suitable to explore the dynamic characteristics of short-term traffic flow under different traffic conditions.A three-regime SETAR model and an autoregressive integrated moving-average(ARIMA) model used for comparison purpose were adopted to forecast 5 min traffic flow data of urban roads in an empirical study,and the latter was used as a benchmark model for comparison purpose.The result shows that the three-regime SETAR model can not only reasonably explain the behavior of these traffic flows but also has better one-step-ahead out-of-sample forecast performance than the ARIMA model does from aspects of the change magnitude and change directions of traffic flow.