Self-adapt evolution SVR in a traffic flow forecasting

This paper proposes a self-adapt evolution support vector regression (SaDE-SVR) in order to improve the performance of traffic flow forecasting. By incorporating the Self-adapt differential evolution algorithm, the parameters of SVR are optimized during the training phase. Additionally, a numerical example of traffic flow data from Xi´an is used to evaluate the performance of the proposed method. The experiment has shown that the proposed SaDE-SVR can achieve the better accuracy without any manually choosing generation and control parameters. It provides an alternative method for traffic flow forecasting.