Improvement of traffic flow combination prediction base on the optimal weighting method
Manchun Tan · Journal of Jinan University · 2010
The linear and non-linear components of traffic flow time series are analyzed.ARIMA model,exponential smoothing model and the gray model are used to predict the linear component of traffic flow time series.The optimal weighting coefficients of the three model predictions are determined by the criterion of the minimum square sum of the forecasting,then the optimal combination of the three models are generated.Finally,the SVM model is applied to the nonlinear residual component prediction.An example is given to show that the combination model can produce more accurate predictions of traffic flow than that of single model.