Forecasting Traffic Volume with Space-Time ARIMA Model

The paper proposes a space–time autoregressive integrated moving average (STARIMA) model to predict the traffic volume in urban areas. The methodological framework incorporates the historical traffic data and the spatial features of a road network. Moreover, the spatial characteristics in a way that reflects not only the distance but also the average travel speed on the links. In order to response the time-varying speed, six traffic modes are classified by level of service (LOS) which is updated in 5 minute interval. In the end, with the real traffic data in Beijing for experiments, the model achieves a very good accuracy on the 5 minute interval forecasting, it also provides a sufficient accuracy of 30 minute interval forecasting compared with ARIMA model.