摘要
客户信用度是评价客户欠费风险的指标.目前国内对电信客户信用度评估采用线性公式,其对相关客户属性的权值进行简单的相加,但是实际客户信用度与各相关属性之间并不是简单的线性关系,而是非线性关系.因此,针对电信客户信用度评分问题,提出了一种应用马尔科夫毯贝叶斯网络分类器建立模型的方法.实验结果显示,该算法建立的客户信用度评分模型简洁、易懂和准确率高.
A client＇s credit is a guide line for valuing the risk of one＇s owing. So far, the credit score is computed in telecom industry by using linear formula to add up the values of correlative client attributes. In practical it is not linear relationship but nonlinear between clients＇ credit and correlative client attributes. So, for evaluating telecom clients＇ credit, an approach based on Markov Blanket Bayesian Classifier is proposed in this paper. Experimental results show that the model of credit scoring built by the algorithm is compact, comprehensible and accurate.
出处
《北京工商大学学报（自然科学版）》
CAS
2006年第3期44-46,50,共4页
Journal of Beijing Technology and Business University:Natural Science Edition
基金
北京市教委重点学科共建项目资助
