Estimation and short-term prediction of link traffic volume based on data from vehicle detectors

This paper describes an application of time-series model and Kalman filter theory to estimation and short-term prediction of traffic volume on specified link in the road network. First, AR modle, ARMA model and ARIMA model are applied to the estimation of link traffic volume based on traffic data collected from vehicle detectors. Second, time-series models and Kalman filter theory are applied to the prediction of link traffic volume. Effectiveness is examined and compared by evaluation estimation errors and prediction errors.