关键词: 销售时序, 高斯核, 支持向量机, 神经网络, 预测

Abstract: Aiming at the characteristics of multi-dimension and nonlinearity existing in the product sale series,an expanded Radial Basis Function (RBF) kernel was designed and applied to Support Vector Machine (SVM) to get an Expanded RBF kernel Support Vector Machine (ERBF-SVM),and an improved immune optimization algorithm was designed to optimize the parameters of ERBF-SVM.The proposed method was applied to the automobile sales forecasting in contrast with Back Propagation Neural Network (BPNN),SVM of RBF and Multi-Scale Support Vector Machine (MS-SVM),The experiment results indicated that ERBF-SVM was effective and feasible,and the prediction accuracy was better than other three methods.

Key words: sale series, Gaussian kernel, support vector machine, neural network, forecasting