A fuzzy rough support vector regression machine

A fuzzy rough support vector regression (FRSVM) is proposed to deal with the overfitting problem caused by outliers in v - support vector regression (v - SVR). Based on rough set theory, the training data points are divided into three regions, i.e., positive region, boundary region and negative region. A fuzzy membership function is also applied to the training data points. Experimental results on benchmark datasets confirm the validity and feasibility of our proposed algorithm.