CAJ | 학술논문

银行贷款风险评估一直是金融界高度关注的主要问题，现有方法主要包括K-means聚类、BP神经网络、简单决策树、VAR方法等多种风险评估算法。但对于客户属性值缺失的案例，上述方法就很难达到良好的效果。为了解决属性值缺失的风险评估问题。提出了一种基于贝叶斯决策树算法的贷款风险评估算法（DBT ），实验结果证明了该算法的有效性。
은행대관풍험평고일직시금융계고도관주적주요문제，현유방법주요포괄K-means취류、BP신경망락、간단결책수、VAR방법등다충풍험평고산법。단대우객호속성치결실적안례，상술방법취흔난체도량호적효과。위료해결속성치결실적풍험평고문제。제출료일충기우패협사결책수산법적대관풍험평고산법（DBT ），실험결과증명료해산법적유효성。
The risk assessment of bank loan has been valued highly by financial world .Various risk evaluation methods for risk classification ,such as K means clustering algorithm ,BP neural net-work ,simple decision tree and VAR ,have been produced ,but they all suffer the problem of attribute values missing .To solve this problem ,we propose a novel Bayesian decision tree algorithm (DBT ) , and the experimental results show its effectiveness .