基于客户洞察的电信数据挖掘研究

发布时间：2018-05-03 03:28

本文选题：客户洞察 + 数据挖掘　；参考：《山东大学》2017年硕士论文

【摘要】：随着现代科技的飞速发展,我国电信市场呈现日益激烈的竞争格局,客户需求的电信业务和服务呈现多样化和差异化趋势,这要求运营商必须对客户消费行为进行深入分析,通过精确识别和细分客户并提供相应服务,才能有效提升客户满意度与忠诚度,增强企业的竞争力。在当前的运营活动中,电信运营商获取和保存了海量的客户数据,但往往对电信客户数据的挖掘和应用不够深入,难以实现从客户数据到企业收益的转变。基于此,本文以实际电信客户数据为基础,联系企业经营活动,提出基于客户洞察的电信数据挖掘模及其新型两阶段聚类分析方法,实现对电信客户的深入分析和挖掘,为电信企业的科学决策提供依据。本文主要针对电信客户数据的相关特征,以客户洞察、顾客关系管理、数据挖掘等相关理论和研究方法为基础,构建相应的数据网络分析模型,并运用新型两阶段聚类分析方法进行处理和分析,主要内容如下:综合客户洞察相关理论和数据挖掘技术,针对电信客户数据的特点构建合理的数据网络模型:(1)收集中国移动公司电信客户数据,充分理解和分析电信客户数据特征,并通过对客户数据进行数据清洗、数据转化和属性选择等相关工作,构建更加准确且便于理解的数据网络模型。(2)结合一种新型图论聚类方法QCM算法在聚类过程中可生成较小层次树、聚类结果允许模糊聚类的优点,以及Ward方法在确定聚类数目上的相关特性,提出一种新型两阶段聚类分析方法,运用该方法开展基于客户洞察的电信数据挖掘研究。第一阶段利用SPSS软件工具运用Ward法进行统计分析,根据不同聚类数目下集聚系数的增量顺序来确定最合理的聚类数目;第二阶段选用第一阶段统计分析得到的聚类数目,并运用新型图论聚类方法QCM算法进行更深层次的聚类分析和挖掘,得到更为合理的电信客户聚类结果。(3)运用新型两阶段聚类算法对电信客户数据进行聚类处理和分析,实现对电信客户的科学合理聚类,针对不同电信客户群体提出相关策略和建议,有利于开展充分而精确的客户洞察工作,也为电信企业提供科学有效的策略指导和决策支持。
[Abstract]:With the rapid development of modern science and technology, the telecom market of our country presents the increasingly fierce competition pattern, the telecommunication service and the service of the customer demand present the diversification and the differentiation tendency, which requires the operator to carry on the thorough analysis to the customer consumption behavior. By accurately identifying and subdividing customers and providing corresponding services, we can effectively enhance customer satisfaction and loyalty and enhance the competitiveness of enterprises. In the current operation activities, telecom operators obtain and preserve a large amount of customer data, but often the mining and application of telecom customer data is not deep enough, it is difficult to realize the transformation from customer data to enterprise income. Based on this, based on the actual telecom customer data, this paper puts forward the telecom data mining module based on customer insight and its new two-stage clustering analysis method, which can realize the in-depth analysis and mining of telecom customers. It provides the basis for the scientific decision of telecom enterprises. Based on the related theories and methods of customer insight, customer relationship management and data mining, this paper constructs the corresponding data network analysis model, which is based on the related characteristics of telecom customer data, such as customer insight, customer relationship management, data mining and so on. A new two-stage clustering analysis method is used to deal with and analyze, the main contents are as follows: comprehensive customer insight related theory and data mining technology, According to the characteristics of telecom customer data, we construct a reasonable data network model: 1) collect telecom customer data from China Mobile Company, fully understand and analyze the characteristics of telecom customer data, and through the data cleaning of customer data, Based on data transformation and attribute selection, a more accurate and easy to understand data network model is constructed, which combines a new graph theory clustering method, QCM algorithm, to generate a smaller tree in the process of clustering. The clustering results allow the advantages of fuzzy clustering and the correlation characteristics of Ward method in determining the number of clusters. A new two-stage clustering analysis method is proposed, which is used to carry out the research of telecom data mining based on customer insight. The first stage uses the SPSS software tool to carry on the statistical analysis using the Ward method, according to the different cluster number increment order of the agglomeration coefficient to determine the most reasonable cluster number, the second stage selects the first stage statistical analysis to obtain the clustering number, A new graph theory clustering method, QCM algorithm, is used for further clustering analysis and mining, and a more reasonable result of telecom customer clustering is obtained. The new two-stage clustering algorithm is used to cluster and analyze the telecom customer data. To realize the scientific and reasonable clustering of telecom customers and to put forward relevant strategies and suggestions for different telecom customer groups is conducive to the development of full and accurate customer insight and provides scientific and effective strategic guidance and decision support for telecom enterprises.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;F626;F274

【相似文献】