基于数据挖掘的保险业客户识别与开发研究

发布时间：2018-04-25 20:05

  本文选题：客户识别 + 客户开发　； 参考：《河南工业大学》2013年硕士论文

【摘要】：近年来，中国保险市场快速扩张的同时，客户的识别成本和开发成本也在大幅提升。从节约成本的角度出发，如何准确地识别目标客户和最大限度的开发现有客户的潜在价值已成为保险企业的难题，而数据挖掘技术的出现为这一问题的解决提供了更好的途径。 在这样的背景下，本文对保险行业在客户识别和客户开发活动中应用数据挖掘技术的相关理论和实践进行探讨。首先，阐述了CRM中的客户识别与客户开发理论，逐一对数据挖掘的概念、分类、主要算法和流程进行了介绍，设计了保险业的数据挖掘主题，实现了数据挖掘与CRM的结合。接着，结合XX人寿保险公司存在的问题，从客户和产品两个角度出发，运用数据挖掘软件Clementine对提取的保险公司客户购买信息数据进行以下三方面的实证分析，完成了客户识别和客户开发的任务： 第一，构建了基于C5.0算法的目标客户分析模型，归纳出了购买和不购买意外保险的客户特征，利用这些客户特征预测潜在客户购买和不购买意外保险的概率，以此来完成潜在客户识别任务。 第二，构建了基于Apriori算法的市场购物篮分析模型，挖掘出哪些险种会被客户同时购买，为企业制定合理的险种组合策略提供借鉴，用于支持客户开发工作。 第三，提出了基于K-means细分的交叉销售模型，该模型的生成分为两步：首先，根据年缴保费和年收入这两个维度预设将总体客户划分为Ⅰ类客户、Ⅱ类客户、Ⅲ类客户和Ⅳ类客户，然后分别对这四类客户进行K-means聚类，实现更加具体的客户细分，并对运行出来的聚类结果进行了分析；然后，利用客户细分模型中的聚类结果，找出各聚类组的特征险种，旨在寻找向现有客户销售新险种或服务的机会，实现了客户识别和客户开发两项任务的结合。 文章最后，结合本文得出的主要结论和在研究过程中遇到的问题，指出了在保险业中实施数据挖掘技术的未来研究方向和不足之处。
[Abstract]:In recent years, the rapid expansion of the insurance market in China, customer identification costs and development costs are also rising significantly. From the perspective of cost saving, how to accurately identify the target customers and maximize the potential value of existing customers has become a difficult problem for insurance enterprises, and the emergence of data mining technology provides a better way to solve this problem. In this context, this paper discusses the theory and practice of applying data mining technology to customer identification and customer development in insurance industry. Firstly, this paper introduces the theory of customer identification and customer development in CRM, introduces the concept, classification, algorithm and flow of data mining one by one, designs the topic of data mining in insurance industry, and realizes the combination of data mining and CRM. Then, combined with the problems of XX life insurance company, from the customer and product point of view, using the data mining software Clementine to extract the insurance company customer purchase information data from the following three aspects of empirical analysis. Completed the tasks of customer identification and customer development: First, the target customer analysis model based on C5.0 algorithm is constructed, and the customer characteristics of buying and not buying accident insurance are summed up, and the probability of potential customers buying and not buying accident insurance is predicted by these customer characteristics. In order to complete the potential customer identification task. Secondly, the market shopping basket analysis model based on Apriori algorithm is constructed to find out which kinds of insurance will be purchased by customers at the same time. Thirdly, a cross-selling model based on K-means subdivision is proposed. The model is divided into two steps: first, according to the two dimensions of annual premium and annual income, the total customers are divided into class I customers, class II customers. The third class customer and the 鈪，

本文编号：1802756