摘要
本文提出了一种以遗传算法和多维判别模型为核心的移动用户信用度预测评估方法并给出了试应用的结果。对以广东移动珠海分公司全球通用户为例进行的信用度预测评估 ,准确率接近90% ,误判率和漏判率均低于15%。这种方法对于各级移动通信公司都是极有意义的 ,如推广应用 。
An approach to the evaluation of the credits for mobile clients is proposed in this paper and a result of a testing application in Zhuhai Branch, GMCC. The essential method for the credit_evaluation used in this study is an organic coalescence of client behavior analysis and genetic algorithm. The precision of the evaluation and forecasting of the client's credit is near 90%. At the same time, the judge_by_mistake rate and the leaving_out rate are both under 15%. This study is very significant to the mobile communication corporation at all levels. The popularization of the techniques and the result will produce great benefits of both society and economy.
出处
《电信科学》
北大核心
2002年第8期17-20,共4页
Telecommunications Science
基金
广东移动通信公司2001年度重点科技项目
