A Comparative Study of Four Smooth Support Vector Regressions

Researching smooth support vector machine for regression (SVR) is an active field in data mining. This paper presents a comparison among four smooth SVRs , epsilon-SSVR, 1st-order polynomial smooth SVR (1PSSVR), 2nd-order polynomial smooth SVR (2PSSVR) and third-order spline smooth SVR (TSSSVR). Accuracy, convergence speed and computational complexity of these regressions are compared.