Abstract: A traffic forecasting model based on the support vector machine(SVM) and chaos was developed to improve the accuracy of the traffic prediction. Based on the phase space reconstruction,it calculates the real-time traffic's delay time, embedded dimension and I_yapunov exponent, and proves that the traffic chaos phenomena exists. hhat a chaos-SVM model was constructed and pairs of training samples was determined to forecast the real network traffic. The resups show that the chaos-SVM model is able to predict network traffic effectively. In comparison with the BP neutral network, it has higher accuracy of prediction.

Key words: Support vector machine(SVM) , Traffic prediction, Regression, Chaos