基于改进的模糊决策树ETC系统客户欺诈分析研究  

Fraud Analysis of ETC System Customer Data Based on Modified Fuzzy Decision Tree

作　　者：刘勇[1] 刁洪祥[2] 刘伟铭[1] 

机构地区：[1]华南理工大学,广州510641 [2]湖南农业大学,长沙410128

出　　处：《交通与计算机》2006年第2期1-4,共4页Computer and Communications

基　　金：国家十五重大科技攻关资助项目(批准号:2002BA-404A19)

摘　　要：随着我国ETC系统的普及,对其进行管理成了急需解决的问题,因利用电子标签和智能卡进行欺诈的行为将给高速公路管理部门带来了极大的损失。介绍了一种改进的模糊决策二叉树模型,即采用最优分割熵方法确定最优割点来对数据模糊化,引入模糊数学中的不确定性理论,依据分类的不确定性建造树,生成规则,在此基础上利用规则匹配的方法检测任意的样本数据是否为欺诈数据。With the increase of ETC system and saved card customers in China, the management of the customer in the expressway is very necessary. The fraud action by use of electronic tag and saved card will bring great loss to the expressway management department. This paper introduces a modified fuzzy decision tree model. Firstly,the data are fuzzed by the use of optimal division point with optimal division entropy. Secondly ,uncertainty theory of fuzzy math is introduced and a tree is built according to uncertainty of classification and creation rule. Finally, each unit can he detected about whether or not it is a fraud according to rule match method.

关 键 词：决策树 模糊数学 欺诈分析 电子收费 客户关系管理 

分 类 号：U491[交通运输工程—交通运输规划与管理]