移动通信中客户信誉评价方法的研究与实践

发布时间：2018-03-02 00:16

本文关键词： 移动客户信誉评价体系贝叶斯理论信息增益法　出处：《东华大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着移动通信业务多样化的不断发展,个人信用问题越来越受到移动运营商的关注。目前大多省市的移动运营商很少有客户的信誉记录,只要用户欠费,系统就会停止向客户提供服务。这种直接停机的做法会导致用户投诉率增长、企业服务水平能力下降等问题,甚至导致运营商面临客户流失的问题,不利于移动行业多元化的发展和运营商竞争力的提高。因此,运营商需要找到一种合理、准确的客户信誉评价方法,通过用户的基本属性和行为属性评价和预测用户的信誉等级,从而根据用户的信誉等级为其提供不同的服务。本文主要的研究工作和创新点概述如下： 1、调查研究了当今造成移动运营商话费流失现象的原因,分析和讨论了移动客户信用管理的现状； 2、根据现有移动通信的业务规则和场景,探讨分析了影响移动用户信誉的因素,并就信誉评价方法进行了研究； 3、针对移动通信行业的客户具有数据量大、客户信息变更频繁等特点,提出了一种基于信息增益和贝叶斯分类法的客户信誉评价分析模型。该模型需要对移动运营商数据库中未经处理的、全面的数据进行清洗、集成和变换等预处理后,然后采用信息增益法计算每个属性的信息熵来确定该属性能为分类系统带来多少信息贡献量,带来的信息贡献量越多,该属性对客户信誉分类越重要； 4、采用贝叶斯分类模型方法进行移动客户信誉分类预测,以达到预测评价用户信誉等级的目的。本文采用某省移动客户的基本和通话交易数据对本文提出的客户信誉分析模型进行验证分析。验证结果表明,该模型预测偏离度低、具有较好的预测效果。本文的研究具有可行性和可借鉴性。
[Abstract]:With the diversified development of mobile communication business, personal credit problems more and more attention of the operators. At present most of the mobile operators have few records of customer's credit, as long as the user arrears, the system will stop providing services to customers. The direct shutdown would result in customer complaint rate of growth, decline in business services the level of ability, and even lead to operators face the problem of customer churn, is not conducive to the diversified development of mobile industry and improve the competitiveness of operators. Therefore, operators need to find a reasonable and accurate method of customer credit evaluation, the basic attributes and behavior evaluation of users and predict the user's credit rating, according to the user the credit rating to provide different services for them.
The main research work and innovation point of this paper are summarized as follows:
1, it investigates and studies the reasons for the loss of mobile operators' charges, and analyzes and discusses the current status of mobile customer credit management.
2, according to the business rules and scenes of existing mobile communications, the factors affecting the reputation of mobile users are discussed and analyzed, and the methods of reputation evaluation are studied.
3, the mobile communication industry customers with a large amount of data, customer information changes frequently, puts forward a model for analysis of customer credit evaluation information gain and Bias classification. Based on the model of the untreated mobile operators in the database, comprehensive data cleaning, integration and transformation processing after the information entropy is then calculated for each attribute using the information gain method to determine which attribute can bring much information contribution to the classification system, the more information contribution caused by the property more important for classification of customer credit;
4, the Bias classification model is used to predict the credit classification of mobile customers, so as to achieve the purpose of predicting the credit rating of the users.
Based on the proposed a mobile client basic transaction data and call customer credit analysis model to validate the results. The verification results show that the prediction deviation is low, has good prediction effect. This research is feasible and referential.

【学位授予单位】：东华大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F626

【参考文献】