摘要
提出了基于有序分类和支持向量机方法的企业信用评级模型。因企业的信用等级是有序的,使用一般分类方法或回归方法建立的模型不能完全反映数据的内在结构。内置空间法将分类变量之间的有序信息嵌入输入向量的扩展位,使得有序分类问题转化为标准的二类分类问题;然后对此二类分类问题采用核分类器———支持向量机进行求解得到有序分类模型。实证结果表明,基于该方法建立的企业信用评级模型显示出较强的泛化能力。
In this paper, a new model on credit rank assessment of enterprises is presented, by using support vector machine（SVM） based on the ordinal classification scheme. The general classification or regression approaches can＇t capture the inherent structure in the sample - data, because the credit ranks of enterprises are ordinal category. The embedded space approach maps an ordinal classification problem to a standard binary classification, exploring the inherent structure within the pattern and without largely increasing the size of the dataset ; then based on this mapping, we can use the kernel classifie, SVM to solve this classification problem. The results of the experiments show that the generalization ability of the SVM model using embedded space approach is better than that of combined SVM model and neural network model.
出处
《杭州电子科技大学学报（自然科学版）》
2005年第6期44-47,共4页
Journal of Hangzhou Dianzi University：Natural Sciences
关键词
有序分类
内置空间法
支持向量机
企业信用评级
ordinal classification
embedded space approach
support vector machine
credit rank assessment
