Sparse support vector regression algorithm with piecewise loss function

Applying sparse algorithm can improve the prediction speed of support vector regression effectively. This paper solves sparse support vector regression with piecewise loss function based on iterative reweight method. By reducing support vector number, the length of regression function expansion and the prediction time of regression function for new samples are decreased. Comparing with sparse LS-SVR, the method has the advantages of suiting different noise data-, steady prediction performance and good generalization performance.