Intelligent Prediction for Time Series Using Smooth Support Vector Regression

A new smooth support vector regression (SSVR) was introduced to solve the prediction problem of complicated time series. The basic idea is replacing the constrained quadratic optimization problem of standard SVR with an unconstrained convex quadratic optimization problem, which effectively reduces training complexity and enhances the speed of regression. In this experiments, SSVR algorithm was tested on Mackey-Glass time series to compare the performances of standard SVR algorithms. According to the experiment results, SSVR has faster speed of convergence and higher fitting precision, which achieves a high-quality prediction about time series.