基于混合双正交小波核支持向量机的企业财务危机预警模型及应用研究

项目名称： 基于混合双正交小波核支持向量机的企业财务危机预警模型及应用研究

项目编号： No.71201024

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 管理科学与工程

项目作者： 黄超

作者单位： 东南大学

项目金额： 18万元

中文摘要： 本项目基于混合双正交小波核支持向量机，对企业财务危机预警模型及应用进行研究。项目首先提出一类新的双正交小波核函数的构造方法，并进一步提出双正交小波核函数之间以及双正交小波核函数与其他核函数之间的组合方法、基本核函数的选择以及权重系数的优化方法，在此基础上构造混合双正交小波核函数。随后，从盈利能力、偿债能力、成长能力和营运能力等四个维度上建立企业财务危机预警的指标体系，基于对不同规模、类型和行业中企业财务数据特征的分析，提出基于混合双正交小波核的核主成份分析方法，进行企业财务危机预警指标优化。最后分别建立基于混合小波核的两值分类、多值分类和单值分类支持向量机模型，以近年来我国上市公司的财务数据和商业银行部分企业财务数据为样本，进行企业财务预警、分级评价以及财务状况异常检测实证研究。

中文关键词： 双正交小波；核函数；支持向量机；核主成份分析；财务危机预警

英文摘要： The research of corporate financial crisis prediction model and its application based on mixed biorthogonal wavelet kernel support vector machine (SVM) is conducted in this project. The constructor method of a new class of biorthogonal wavelet kernel function is prosposed firstly. Further, Combination methods between different biorthogonal wavelet kernels or biorthogonal wavelet kernel and other traditional kernels, and the choice methods of base kernels and Weight optimization methods are also proposed. Afterwards, new mixed biorthogonal wavelet kernel functions are constructed. Then the financial evaluation index system are established from the four dimensions which contain profitability, solvency, growth capacity and operating capacity, and kernel principal component analysis method based on mixed biorthogonal wavelet kernel is presented to feature selection according to the analysis of data collected from corporations of different sizes, types and fields. At last, two-class, multi-class and single-class SVM modles based on mixed biorthogonal wavelet kernel are respectively structured and used to conduct empirical experiments of corporate financial crisis prediction on financial data from listed companies and commercial bank in China in recent years.

英文关键词： Biorthogonal Wavelet；Kernel Function；Support Vector Machine；Kernel Principal Component Analysis；Financial Distress Prediction