A Forecasting Model Based Support Vector Machine and Particle Swarm Optimization

In view of the bad forecasting results of the standard epsilon-support vector machine (SVM) for product sale series with the normal distribution noise, a SVM based on the Gaussian loss function named by g - SVM is proposed. And then, a hybrid forecasting model for product sales and its parameter-choosing algorithm are presented. The results of its application to car sale forecasting indicate that the short-term forecasting method based on g - SVM is effective and feasible.