Performance Analysis of a Novel Robust Support Vector Regression Algorithm

Abstract

The experimental data with polyhedral perturbation is a huge challenge for regression model and data forecasting. This paper aims to study on regression performance of the robust support vector regression for processing input data with polyhedral perturbation which was proposed by the author of this paper. First, the mathematical model of the robust support vector regression method was introduced and the feature of the polyhedral perturbation was analyzed. Second, the robust regression algorithm flow was listed and the most widely used Gaussian kernel function was present. Third, two numerical experiments including linear regression and nonlinear regression were given to prove effectiveness of the robust regression method for processing input data with polyhedral perturbation.