基于ODR-BADASYN-SVM的中小企业信用风险评估

Credit Risk Assessment of SMES Based on ODR-BADASYN-SVM

本文在文献调研和问卷调查的基础上,构建中小企业信用风险评价指标体系,并以A股非金融上市公司中的中小企业为研究对象,通过SVM构建信用风险评估模型。将逐级优化递减欠采样方法 (ODR)和边界自适应合成抽样方法 (BADASYN)与SVM相结合,构建ODR-BADASYN-SVM模型评估中小企业信用风险,以克服样本不平衡性对模型性能的影响。研究结果表明,改进后的SVM模型在中小企业信用风险评估中具有较高的稳定性及预测能力。

Based on literature review and questionnaire survey,this paper constructs the credit risk evaluation index system for SMEs. Taking SMEs of non-financial A-share listed companies as the research object,the paper builds the credit risk assessment model through SVM.Moreover,in order to reduce the negative impacts of imbalance sample on model performance,optimization of decreasing reduction(ODR) and border adaptive synthetic sampling approach(BADASYN) are combined with SVM to construct ODR-BADASYN-SVM Model to evaluate SME Credit Risk. The result shows that the improved SVM has high stability and prediction ability in credit risk assessment of SMEs.