【摘要】 结合粗糙集理论的属性约简和支持向量机(SVM)的分类机理,提出一种数据分类的混合算法;建立了基于此算法的商业银行信用风险评估模型。模型以粗糙集属性约简作为预处理器,删除冗余属性和冲突对象,但不损失有效信息;然后基于SVM进行分类建模和预测。实证表明,创建的模型分类性能良好,降低SVM分类过程的复杂度,一定程度上避免了训练模型的过拟合现象。通过与SVM和神经网络模型的比较,证实该方法用于信用风险评估的有效性。更多还原

【Abstract】 Based on the attribute reduction of the Rough Set (RS) and classification principles of the Support Vector Machine (SVM), a new hybrid algorithm, RS_SVM model is introduced in this paper. Using the new algorithm, a model of credit risk assessment in commercial banks is built. Firstly, the attribute reduction of RS has been applied as preprocessor to delete redundant attributes and conflicting objects without losing efficient information. Then, an SVM classification model is built to make a forecast. Empirical results show that RS_SVM model obtains good classification performance, and it highly decreases the complexity in the process of SVM classification and prevents the over-fit of training model in a certain extent. Compared with the SVM model and neural network model, the results show the effectiveness of the new proposed model.更多还原