Tax Forecasting Based on Linear Discriminant Analysis-wavelet Support Vector Regression Algorithm

Abstract The accurate prediction for future tax values is the key to make the reasonable tax collection policy. In order to improve the prediction accuracy of wavelet support vector regression, linear discriminant analysis-wavelet support vector regression algorithm is proposed to predict future tax values in the paper. In the linear discriminant analysis-wavelet support vector regression algorithm,the linear discriminant analysis(LDA) is used to optimize wavelet support vector regression algorithm. Tax data of China from 1978 to 2001 are used as our experimental data. The comparison of tax prediction values between the LDA-WSVR model and traditional WSVR model with the 3~7 input nodes respectively shows that the prediction results for future tax of LDA-WSVR are better than those of traditional WSVR.