Short-Term Traffic Flow Combined Forecasting Based on Nonparametric Regression

To improve the accuracy of short-term traffic flow forecasting, a combined forecasting method based on nonparametric regression is proposed in this paper. In this method, two independent forecasting models are introduced and the matching parameters of them are selected from the prediction and neighbor node, respectively. Grey correlation degree and correlation coefficient are used to determinate the input variables of pattern matching for the two forecasting model, respectively. Finally, the two forecasting model are combined together and entropy theory is utilized to determine the weight of each single forecasting model. With simulation using highway traffic data, it is demonstrated that the proposed combined forecasting method can effectively improve the forecasting accuracy.