【作者】 于莉；

【导师】 孙学军；

【作者基本信息】 天津大学 ， 计算机应用技术， 2006， 硕士

【摘要】 随着移动通信业务领域竞争的日趋激烈,电信运营商要想在竞争中立于不败之地,就必须向客户提供满足其需要的高质量的服务,以留住现有客户,同时争取新客户。但在当前激烈的竞争环境中,发展新客户越来越困难,于是运营商将注意力集中在向现有的客户提供价值更高的增值服务,以从中获得更多的收入和利润。数据挖掘是对大量数据进行分析,从而揭示数据背后隐藏的知识的技术。它能帮助运营商从大量的客户数据中分析客户的消费行为,识别出可能使用增值服务的用户特征,确定使用某种增值服务可能性较高的目标客户,帮助客服和营销人员更有针对性的进行业务推广,从而以较低的销售成本,获得较高的价值。本文运用数据挖掘中的决策树方法建立提升销售模型来预测客户使用炫铃服务的可能性。论文首先对数据挖掘的理论进行了概述,包括数据挖掘的基本概念、数据挖掘的过程、数据挖掘的方法以及数据挖掘与数据仓库和OLAP的关系等;然后介绍了分类的概念及其常用的基本方法,并对本文所采用的决策树分类方法进行了较全面的论述,包括决策树的生成算法、剪枝方法以及决策树的评价方法等;本文的重点是以某移动通信运营商的客户数据为基础,在SAS统计分析软件中建立炫铃提升销售模型。论文对整个数据挖掘过程进行了较深入的论述和分析,包括模型建立前的数据准备、数据采样、变量选择等预处理工作;在SAS中通过参数设置近似实现C4.5和CART算法建立模型的方法以及模型结果的评价。最后,根据模型的评价结果,通过综合比较,选定C4.5算法建立的模型作为最终的分析模型,并通过实际检验取得了良好的应用效果。更多还原

【Abstract】 With the fierce competition in the field of mobile communication,the telegraphic traders,in order to be in incincible position,have to offer high quality service to satisfy old customers and attract new customers.However,under the current circumstances,it is more and more difficult to develop new clients,so the traders pay more attention to pushing the added-value service with higher prices,aiming at more income and profit.Data mining,on the basis of the analysis of considerable data,reveals the know-how behind the data.It will help the traders sum up customers’consuming activities,recognize the features of the customers to use the added-value service and pick up the target customer,that will make the customer service and the marketing staff push the added service more effectively in order to get higher profit.The essay,throgh the decision tree in data mining,establishes the lift sales model to predict the possibility for the customer to use dazzling ring service.Beginning with summary of the data mining theory,including the basic conception,the process,the method and the relationship among data mining,data warehouse and OLAP,the essay introduces the idea of classification and discusses the method to categorize the decision tree,such as the generating algorithm and the way to prue and evalue the decision tree.The essay,based on the data collected from customers in a mobile communication trader,attaches importance to establishing dazzling ring service lift sales model by SAS statistice analysis program.Through the deep discussion and analysis of the data mining process,inclusive of data preparation,collection and variable choosing and so on,the essay established model approximately fulfiling the C4.5 and CART by parametric configuration and evaluates the result.At last,according to the evaluation result,by comparasion,C4.5 if proved to be the suitable analysis mode that obtains a good application effect from practical testing.更多还原