A Parameter Choosing Method of SVR for Time Series Prediction

It is important to choose good parameters in support vector regression (SVR) modeling. Choosing different parameters will influence the accuracy of SVR models. This paper proposes a parameter choosing method of SVR models for time series prediction. In the light of data features of time series, the paper improves the traditional cross-validation method, and combines the improved cross-validation with epsilon-weighed SVR in order to get good parameters of models. The experiments show that the method is effective for time series prediction.

1. Introduction

Support Vector Machine(SVM) is a new learning method, proposed by Vapnik according to Statistics Learning Theory^[1]. It follows the rule of Structural Risk Minimization (SRM) with the characteristics of structure-simple, global optimization, good generalization, and has become new research hotspot in recent years. It was used to solve problems of Pattern Recognition at first. With the introduction of -non-sensitive loss function, its use has extended to regression function estimation, nonlinear system discrimination, prediction, and so on, showing better learning performance.