ABSTRACT
Accurate and timely forecasting of traffic flow is of paramount importance for effective management of traffic congestion in intelligent transportation system (ITS). Ross (1982)
and Okutani and Stephanedes (1984) apply a filtering technique for forecasting short term traffic volume based on the past traffic flow. Davis and Nihan (1991) present a nonparametric forecasting model based on the ^-nearest-neighbor approach for
forecasting short-term freeway traffic flow. However, both the filtering technique and the nonparametric model can forecast the short-time traffic flow with reasonable accuracy only when the traffic flows are relatively constant. More recently, the linear statistical time series models, such as the autoregressive integrated moving average (ARIMA) model (Williams et al. 1998) and seasonal ARIMA model (Lee and Fambro 1999) have
been used to improve the accuracy of short-time freeway traffic flow forecasting.
