基于规则强度与证据推理的电信客户状态识别  

Reasoning of Telecom Consumers' Behavior Based on Rule Strength and Evidential Theory

作　　者：王锐[1] 王虎[1] 熊亚洲[1] 李荻[1] 

机构地区：[1]武汉理工大学管理学院,湖北武汉430070

出　　处：《武汉理工大学学报（信息与管理工程版）》2015年第2期212-214,219,共4页Journal of Wuhan University of Technology：Information & Management Engineering

基　　金：国家自然科学基金资助项目(71071122)

摘　　要：针对利用移动电话用户的行为数据来对客户的流失进行预警这一电信客户流失管理的难点,首先利用相关分析、粗糙集及信息熵理论从众多的用户消费行为属性中提取出对用户状态有显著影响的客户特征行为属性作为推理证据,然后给出证据对应的BPA的客观确定方法,并在此基础上提出了基于规则强度集与证据推理的电信客户行为知识推理算法,最后通过实证验证了算法的有效性。How to use mobile-phone consumer＇behavior data predicting the customers＇churn is always important and diffi-cult in telecommunications churn management.Firstly, correlation analysis, rough set and entropy theory were employed to ex-tract the characteristic attributes from a large number of consumers behavior attributes, which have effected on the state of the customers＇status significantly.The reasoning evidences were constituted by these characteristic attributes.Secondly, the method to calculate the BPA corresponding to every evidence was given out.Meanwhile, customers＇behavior knowledge reasoning algo-rithm based on rule strength and evidential theory was also proposed.At last the valid of algorithm was verified by empirical anal-ysis.

关 键 词：规则强度 客户行为 证据理论 

分 类 号：F272[经济管理—企业管理] TP301[经济管理—国民经济]