Network Flow Forecasting by a Novel Twin Support Vector Regression Algorithm

It is well-known that accurate prediction for network flow is very important to meet the communication requirement of internet network. This study is to propose a novel twin support vector regression algorithm for network flow forecasting. The twin support vector regression algorithm is comprised of a pair of the standard SVR. In order to show the excellent performance of twin support vector regression algorithm compared with other prediction methods. Then, we perform the comparison of the prediction error of network flow between twin support vector regression algorithm and support vector regression algorithm, and the experimental results show that network flow prediction ability of the twin support vector regression algorithm is better than that of support vector regression algorithm.