Combination Method and Multiple Interval Freeway Volume Forecasting

With the deployment of Intelligent Transportation Systems (ITS) infrastructure, the need for long-range traffic flow forecasting is demanding. This paper investigated the performance of a combination method for long-range freeway traffic forecasting, and it was compared to other statistical forecasting methods using two statistical error measures. Results showed that the performance of each model was subject to the variability of traffic flow itself. In general, exponential smoothing and autoregressive method gave better performance for short-range traffic forecasting, and the combination method produced the good results for long-range traffic forecasting regardless of levels of traffic flow fluctuation.