A MANOVA of Historical and Unhistorical Factors for Traffic Volume Forecasting

摘要

Traffic congestion has become a serious social problem. One way to ease traffic congestion is traffic volume forecasting, which is helpful to support decision-making for management or provide better route selection for passengers. Nowadays, the historical traffic flow data are used as the original input data for predictor training in forecasting and proves helpful for forecasting. However, other unhistorical factors can also influence traffic volume, such as weather condition and holidays, are often ignored in recent forecasting researches. This paper applied a Multivariate analysis of variance (MANOVA) for analyzing the relation between traffic volume and the unhistorical factors. The experiments are carried out on the data collected from a section of a national road, which show that weather conditions and holidays are helpful for traffic forecasting.