基于粒子群优化偏最小二乘支持向量机的我国税收收入预测研究

Particle Swarm Optimized Partial Least Square Support Vector Machine Model for Tax Revenue Prediction

由于我国税收收入存在高度的非线性、耦合性和多因素影响的复杂性,故而对其进行预测是传统的预测方法难以胜任的。首先,本文对当前税收预测方法存在的不足进行了阐述,在此基础上,提出了偏最小二乘支持向量回归法(PLS-SVR)对我国税收收入进行预测的思路,并建立了相应的数学模型。其次,由于参数集(C,σ~2)直接影响支持向量技术的预测优劣,故笔者采用改进的粒子群算法对参数集进行全局寻优,这样既保证了处理非线性和多因素复杂性的优势,又确保了支持向量回归模型的稳定性与精确性。最后以我国近30年的税收收入为研究对象,经参数集寻优后的支持向量回归法的预测精度较其他预测方法的预测精度有着显著提高,说明了该模型的有效性与实用性。

Chinese tax revenue is non-linear and coupled,and it is influenced by many factors.Therefore,traditional forecasting methods are not sufficient to predict the value of it.Firstly,disadvantages of the existing forecasting methods are analyzed in this paper.Then partial least square support vector machine(PLS-SVR) is used to construct a tax revenue prediction model.An improved particle swarm algorithm is used to optimize the parameter set of(C,cr2),which influences the performance of this model directly.By doing so,this model can deal with the nonlinearity and multi-factors of tax revenue,and stability and accuracy of support vector machine based regression can be guaranteed.Case study on Chinese tax revenue during the last 30 years demonstrates that the optimized PLSSVR model is much more accurate than other prediction methods.