基于Clementine的广告客户数据挖掘模型设计与实现

发布时间：2018-05-23 19:31

本文选题：数据挖掘 + 客户细分模型　；参考：《北京邮电大学》2010年硕士论文

【摘要】：企业建立自己的数据库系统,由计算机管理代替手工操作,以此来收集、存贮、管理业务操作数据,改善办公环境,提高操作人员的工：作效率。企业需要从业务数据中提取有用的信息,帮助他们在业务管理和发展上作出即时、正确的判断。当一个企业明确了自己的客户之后,紧接着就应该做客户细分。不同的客户对同一产品的需要存在着明显的差别,客户对产品的要求越来越理性和严格,对企业服务的整体质量也提出了更高要求。不同类型的客户选择的往往不仅是产品的单一特性,还可能是产品特性的某种组合。对于企业来说,同一产品不可能满足市场上所有的客户的需要,只能面向某一种类型的客户。另一方面,某一一特定的产品要不仅满足单一类型的客户,还要满足多范围、多层次、有着不同需要的客户群。客户细分的目的,就是要更精确地回答谁是我们的客户,客户到底有哪些实际需要,企业应该去吸引哪些客户,应该重点保持哪些客户,应该如何迎合重点客户的需求等重要问题。企业可以通过响应率分析能够有效的降低市场推广的费用,同时能够更加有针对性的面对目标市场,达到以最小的投入获得最佳效果的目的。需要构建预测模型,找到最合适的响应客户,预测哪些客户能够响应,以及响应的可能性是多少。针对广告营销市场的不断发展,企业收集了大量的客户资料。数据挖掘需求来自新雅迪传媒。为了便于广告中心制定较为合理的营销策略,将用SPSS Clementine建立模型,以提升新客户开发的成功率,降低长单客户的流失率。本文根据业务部门需求,经过与业务人员的不断沟通,将营销过程、客户信息与数据挖掘技术相结合,经过数据理解、数据清洗、模型训练等数据挖掘过程,设计并实现了客户细分模型和客户响应预测模型。再对模型进行评估和部署,有效的从杂乱无章的客户数据中发现具有商业价值的信息。
[Abstract]:In order to collect, store and manage the operation data, improve the office environment and improve the working efficiency, the enterprise establishes its own database system and uses computer management instead of manual operation to collect, store and manage the operation data. Enterprises need to extract useful information from business data to help them make immediate and correct judgments on business management and development. When an enterprise has identified its own customers, the next step should be customer segmentation. Different customers have different needs for the same product. The requirements of customers for the same product are more and more rational and strict, and the overall quality of the enterprise service is also put forward higher requirements. Different types of customers often choose not only the single feature of the product, but also some combination of the product characteristics. For enterprises, the same product can not meet the needs of all customers in the market, only for one type of customers. On the other hand, a particular product should not only meet a single type of customer, but also meet a multi-scope, multi-level, with different needs of the customer base. The purpose of customer segmentation is to answer more precisely who our customers are, what the actual needs of our customers are, what customers enterprises should attract, and which customers should be kept. How to meet the needs of key customers and other important issues. Through the response rate analysis, enterprises can effectively reduce the cost of market promotion, at the same time, they can face the target market more pertinently, and achieve the goal of getting the best effect with the minimum investment. A predictive model is needed to find the most appropriate response customer, which customers can respond, and what is the likelihood of the response. In view of the continuous development of advertising marketing, enterprises have collected a large number of customer information. The demand for data mining comes from New Yadi Media. In order to facilitate the advertising center to formulate more reasonable marketing strategy, SPSS Clementine will be used to establish a model to improve the success rate of new customer development and reduce the loss rate of long single customer. According to the demand of business department, this paper combines marketing process, customer information with data mining technology, data understanding, data cleaning, model training and other data mining processes through continuous communication with business personnel. The customer subdivision model and customer response prediction model are designed and implemented. Then the model is evaluated and deployed, and the information of commercial value is found from the chaotic customer data.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2010
【分类号】：TP311.13

【参考文献】