Estimation of Short-Term Traffic Volume Under Special Events: Comparison Between Seasonal ARIMA and Seasonal VARMA Models

A special event may create a surge in travel demand and result in congestion and temporary road closures, which will adversely impact the normal social and economic activities of local residents and businesses. There is an urgent need for quick-response techniques to analyze and forecast traffic volume changes during an event. In this paper, seasonal ARIMA and seasonal vector ARMA (VARMA) with intervention models are adopted for estimations of short-term traffic volume under special events. The case study is carried out by analyzing traffic data at 12 locations on the Second and Third Ring Roads around Beijing Worker Stadium in Beijing, China. The forecasting performances of these two kind models are discussed and compared.