基于聚类分析方法的客户分类研究

发布时间：2018-01-06 09:05

本文关键词：基于聚类分析方法的客户分类研究　出处：《北京理工大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着互联网产业的迅猛发展,互联网企业间竞争逐渐加剧;企业竞争的实质就是对客户资源的竞争,这意味着能够有效开发与保持客户的企业往往更加具有竞争优势。然而,企业资源是有限的,难以同时满足所有客户的需求;为保障收益最大化,企业需要将资源尽可能多地分配到对企业贡献最大的客户群体上。因此,企业需要对客户进行分类,识别不同细分类别的客户特征与价值;根据不同群体的客户的需求与价值,合理分配资源及采取合适的运营策略,才能够有效提高企业的行业竞争力以及创造出更多的利润。本文在对客户分类理论与方法的研究基础上,结合A公司的实际问题,选取了客户的消费数据,分别用K-means聚类、Two-step聚类和Kohonen神经聚类等三种方法对A公司的客户进行了分类,同时运用了图示法及内部检验法对三种方法下的分类效果进行了评价,发现基于Two-step算法的分类模型相较于基于其他两种方法的模型具有最佳的分类效果。基于Two-step聚类法的分类结果,将客户划分成五个群体,对各个细分群的客户的特征与价值进行了分析与总结,针对各细分群体的客户特征差异提出了研究建议,以便于公司制定运营策略。
[Abstract]:With the rapid development of Internet industry, the competition between Internet enterprises is becoming more and more serious. The essence of enterprise competition is the competition of customer resources, which means that enterprises that can effectively develop and maintain customers often have more competitive advantage. However, enterprise resources are limited. Difficult to meet the needs of all customers at the same time; In order to ensure the maximization of income, enterprises need to allocate as much resources as possible to the customer groups that contribute the most to the enterprise. Therefore, enterprises need to classify customers and identify the customer characteristics and value of different subdivision categories. According to the needs and value of different groups of customers, reasonable allocation of resources and adoption of appropriate operational strategies. In order to effectively improve the competitiveness of the industry and create more profits. Based on the study of customer classification theory and method, combined with the actual problems of company A, this paper selects customer consumption data. K-means clustering two-step clustering and Kohonen neural clustering are used to classify the customers of A company. At the same time, the classification effect of the three methods is evaluated by using the graphical method and the internal test method. It is found that the classification model based on Two-step algorithm has the best classification effect compared with the model based on the other two methods. The classification result based on Two-step clustering method. The customer is divided into five groups, the characteristics and value of each subdivision group are analyzed and summarized, and the research suggestions on the customer characteristics difference of each subdivision group are put forward in order to facilitate the company to formulate the operation strategy.
【学位授予单位】：北京理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F274;F49

【相似文献】