Network flow prediction based on grey-support vector regression technology

In order to solve the shortcoming of BP neural network, a novel prediction method is presented to predict network flow. The combination of grey prediction model and support vector regression is applied to predict network flow in the paper. We employ collected network flow experimental data to test the performance of the combination model of grey prediction model and support vector regression. The mean relative error of grey prediction model and support vector regression is 1.94, while the mean relative error of BP neural network is 3.43. It can be seen that the combination model of grey prediction model and support vector regression is superior to BP neural network.