摘要
将二次Renyi熵应用于企业财务困境预测,提出了一种基于二次Renyi熵的最小二乘支持向量机(LS-SVM)模型。通过将该模型与传统的LS-SVM模型、标准SVM模型以及与二项Logistic回归模型、BP神经网络(BP-ANN)的分析比较,表明了该模型无论是训练样本的数量还是运算时间,都显著优于其他模型,且有较好的稳定性。实证分析表明,将二次Renyi熵引入企业财务困境预测领域是成功的,同时,通过对原始输入变量进行显著性检验、因子分析处理,减少了输入变量个数,预测正确率达到了88%,说明因子分析法是有效的。
A learning algorithm of noniterative Least Squares Support Vector Machine （LS-SVM） based on quadratic Renyi-entropy was propused in the article by using quadratic Renyi-entropy in financial distress prediction. By comparing the model of LS-SVM based on quadratic Renyi-entropy with traditional LS-SVM, standard SVM, binomial Logistic regression model and Back Propagation Artificial Neural Network （BP-ANN）, this paper concluded that either the number of training samples or the computing time, the model of noniterative I,S-SVM based on quadratic Renyi-entropy is remarkably better than the others, as well as the stability. Indicated by demonstration analysis, the model of noniterative LS-SVM based on quadratic Renyi-entropy is successful in financial distress prediction. Meanwhile, although the number of input variable has been reduced by conspicuity test and gene analysis, the accuracy rate of the prognosis still reached 88%. In a word, the factor analysis method has been successfully proved in the article.
出处
《计算机应用》
CSCD
北大核心
2009年第10期2751-2754,2757,共5页
journal of Computer Applications
基金
国家自然科学基金资助项目(70840018)
山东省科技攻关计划资助项目(2008GG30009005)
山东省软科学研究计划资助项目(2008RKA223)
关键词
二次Renyi熵
最小二乘支持向量机
标准支持向量机
非迭代
因子分析
财务困境预测
quadratic Renyi-entropy
Least Squares Support Vector Machine （LS-SVM）
standard SVM
noniterative
factor analysis
financial distress prediction
