基于PSO_SVM的企业债务违约损失率判别模型  

The Discrimination Model via the PSO_SVM on LGD of Corporate Debt

作　　者：罗晓光[1] 孔慧[1] 张志超[1] 

机构地区：[1]哈尔滨理工大学经济学院

出　　处：《投资研究》2015年第1期148-159,共12页Review of Investment Studies

基　　金：黑龙江省自然科学基金(G201302)(黑龙江省社会诚信建设运行状态数据挖掘模型研究)支持

摘　　要：针对企业债务违约损失率判别问题中属性变量居多这一特点,选择支持向量机模型进行判别,并从贷款回收的角度将以往简单的两类有无回收模式(有回收和无回收)扩展为三类。为提高模型效率,将逐步判别分析法应用到模型变量的选择上,同时为了避免人为选择参数的随意性,采用粒子群算法优化支持向量机的参数,将建立的PSO-SVM多分类判别模型对500笔银行贷款进行实证研究。结果表明,该模型不仅提高了分类准确率,而且具有良好的稳健性。To overcome the difficult of too many indicating variables in the analysis of corporate debt Loss Given Default,this paper choose Support Vector Machine discrimination model to discriminate .From the perspective of loan recovery, the previous simple recovery model of two types（The recovery and without recovery）was extended to three categories. In order to improve the efficiency of the model,the stepwise discriminant analysis method is employed to select the model variables, while particle swarm optimization is used to optimize the parameters of the SVM model,to avoid the arbitrariness of artificial selection param- eters.The PSO-SVM multi-classification of discriminant model is established for the 500 bank loan in empirical research. The results show that the model not only improves the classification accuracy, but also has a high stable performance.

关 键 词：违约损失率 支持向量机 粒子群算法 逐步判别法 

分 类 号：F275[经济管理—企业管理] F832.4[经济管理—国民经济]