模糊软集合理论在税收组合预测中的应用

Application of fuzzy soft sets in tax combined forecasting

结合模糊软集合理论建立税收收入的组合预测模型,根据税收收入的特点,代表性地选择了Elman回归神经网络模型、含政策虚拟变量的自回归模型、ARIMA(1,1,1)的时间序列模型、多因素SVM回归模型这四种模型作为组合预测中的单一模型,并以1980年到2008年的税收收入等相关数据为背景进行了说明和分析.结果表明该组合预测模型能有效减小预测误差,为税收工作实践提供了一个应用研究工具,并推广和丰富了软集合理论在税收经济模型研制中的实际应用.

Fuzzy soft set theory to establish a combination of tax revenue forecasting model, based on the characteristics of tax revenue, representative selection of the Elman recurrent neural network model, with the policy dummy variables from the regression model, ARIMA（1,1,1） time series models, multi-factor regression model of these four SVM model as a combination of a single model for forecasting. Also, this paper described and analyzed the selected tax revenue and other relevant data of our country from 1980 to 2008. The results show that the combination forecasting model can effectively reduce the forecast error and provide an applied research tool for the tax work practice. Also the combination forecasting model can promotes and enriches the soft set theory in the practical application of tax economic model developing.