Regression analysis is often formulated as an optimization problem with squared loss functions. Facing the challenge of the selection of the proper function class with polynomial smooth techniques applied to Support Vector Regression models, this study takes three interplation points spline interpolation technology and modification interpolation value to generate a new polynomial smooth function in -insensitive support vector regression. The experimental analysis shows that -function is better than -function and -function in properties, and the approximation accuracy of the proposed smooth function is three order of higher than that of classical -function.