The Research on Forecasting Model Based on Support Vector Machine and Discrete Grey System

Abstract

In order to improve the prediction accuracy, this paper proposes a combined forecasting model based on the residual correction of DGM (1,1). After introducing the basic DGM(1,1) model and LSSVM regression model, DGM-LSSVM combination model is established. In this model, DGM(1, 1) is used to predict the original data sequence and obtain the predictive value and residual value. Then LSSVM model is used to correct the residual errors. Finally, the predicted results of DGM (1, 1) and residual correction of LSSVM model are combined, and the final prediction result is obtained by the combination forecasting model. Experimental results demonstrate that the proposed model has the advantages of effectiveness and feasibility.