基于稀疏贝叶斯学习的个人信用评估

发布时间：2018-01-06 03:06

本文关键词：基于稀疏贝叶斯学习的个人信用评估　出处：《计算机应用》2013年11期 　论文类型：期刊论文

【摘要】：针对传统信用评估方法分类精度低、特征可解释性差等问题,提出了一种使用稀疏贝叶斯学习方法来进行个人信用评估的模型(SBLCredit)。SBLCredit充分利用稀疏贝叶斯学习的优势,在添加的特征权重的先验知识的情况下进行求解,使得特征权重尽量稀疏,以此实现个人信用评估和特征选择。在德国和澳大利亚真实信用数据集上,SBLCredit方法的分类精度比传统的K近邻、朴素贝叶斯、决策树和支持向量机平均提高了4.52%,6.40%,6.26%和2.27%。实验结果表明,SBLCredit分类精度高,选择的特征少,是一种有效的个人信用评估方法。
[Abstract]:Aiming at the problems of low classification accuracy and poor interpretability of the traditional credit evaluation methods. A model of personal credit evaluation using sparse Bayesian learning method is proposed. SBLCreditit.SBLCredit makes full use of the advantage of sparse Bayesian learning. In the case of the priori knowledge of the added feature weights, the feature weights are sparse as far as possible, so as to achieve personal credit evaluation and feature selection. In Germany and Australia real credit data sets. The classification accuracy of SBLCredit method is 4.52% higher than that of traditional K-nearest neighbor, naive Bayes, decision tree and support vector machine. The experimental results show that the classification accuracy of SBL Credit is high and the selection of features is few. It is an effective personal credit evaluation method.
【作者单位】：西南财经大学经济信息工程学院;西南财经大学
【基金】：教育部人文社会科学研究青年基金资助项目(11YJCZH084) 中央高校基本科研业务专项资金资助项目(JBK130142,JBK130503) 西南财经大学科研基金资助项目(2011XG130)
【分类号】：F830.5;F224
【正文快照】： 0引言信用评估是商业银行控制风险的关键技术,发生在美国的“次贷危机”就是信用风险的大爆发,因此信用评估方法的研究具有非常重要的现实意义。信用评估实质上是数据挖掘中的分类问题———将贷款者根据其属性分成能够按期还本付息的可信的“好”客户(正类)和违约的“坏”客

【参考文献】