基于贝叶斯网络的大客户离网预警模型研究

Research on Key Client Leaving-Network Alarming Model Based on Bayes Network

大客户是通信运营商关注的焦点,如何及时发现可能离网的大客户对运营商是极其关键的。本文基于贝叶斯网络提出了一个能及时发现大客户具有离网倾向的预警模型。模型首先分析了大客户离网的相关条件,包括:话费开销、通话次数、通话时长、数据业务等;进而,统计分析条件间的相关情况;最后,基于贝叶斯概率理论分析大客户是否具有离网倾向。将模型应用于湖南某地区移动系统的结果表明,模型能及时预警大客户的离网倾向,具有较好的商业价值。

Key clients are the focus of the communication company,so it is very important to find the possible key clients to leave the network as soon as possible.In this paper,an alarming model,which can find the key clients ＇ leaving-network action,is proposed based on Bayes network.First,the related conditions of the key clients leaving-network are analyzed.The related conditions include telephone fare,the number of call,the call time and the data business.Second,the relationship among the conditions is calculated.Finally,the possibility of the key clients＇ leaving-network is evaluated based on Bayes network.The model is applied to the managing system of some branch in Hunan Mobile Company.The results show that the proposed model can predict the key clients＇ action of leaving-network and has good business value.