基于神经网络的客户流失预警研究

发布时间：2019-05-15 01:05

【摘要】：面对不断变化的市场需求和竞争激烈的市场环境，将客户流失率降至最低，是企业赢得市场、取得成功的根本。自从我国加入WTO后，各个市场对外开放，，我国各行各业都面临来自外国产品的竞争，大大加重了企业对客户资源的争夺，与此同时，信息技术的快速发展推动了电子商务时代的到来，网络营销以其独特的优势运营而生，使得供应商也成为企业的竞争对手，这对零售企业来说无疑是雪上加霜。因此，对零售行业的进行客户关系管理显得迫切而重要。客户关系管理的重点在于降低客户流失率，而降低客户流失率的关键在于客户流失预警，进行客户流失预警的常用技术是数据挖掘。本论文在客户关系管理的大知识背景框架下，梳理分析客户流失管理相关理论的基础上，利用数据挖掘技术方法对零售业客户流失预测问题进行了研究。本文首先回顾了客户关系管理的相关理论知识，客户流失的定义、原因及客户流失管理过程，探讨了客户价值的概念以及客户价值评估的几类算法。然后提出了基于RFM客户价值和IG-NN属性选择的客户流失预警模型，利用RFM模型计算出客户价值，用信息增益选择主要属性，再用神经网络分析每个主要属性对客户流失率的影响程度并结合二八法则判断导致客户流失的关键属性，并以客户价值、关键属性作为神经网络的输入，客户流失概率作为网络输出，构建基于RFM客户价值和IG-NN属性选择的客户流失预警模型。然后将本文所得结果与单一神经网络和基于IG-NN属性选择的客户流失预警模型进行对比，发现本文的客户流失预警模型在准确率、命中率、覆盖率以及提升度方面均优于另外两个模型。最后，对本文的研究结论进行了总结，并对未来研究提出展望。
[Abstract]:Facing the changing market demand and the fierce competition market environment, it is the foundation for the enterprise to win the market and achieve success by reducing the customer wastage rate to the lowest. Since China's entry into WTO, various markets have been opened to the outside world, and all kinds of industries in our country are facing competition from foreign products, which greatly aggravates the competition for customer resources by enterprises. At the same time, The rapid development of information technology has promoted the arrival of the era of electronic commerce. Network marketing is born with its unique advantages, which makes suppliers become competitors of enterprises, which is undoubtedly even worse for retail enterprises. Therefore, it is urgent and important to carry out customer relationship management in retail industry. The focus of customer relationship management is to reduce the customer turnover rate, and the key to reduce the customer turnover rate lies in customer loss early warning. Data mining is the common technology to carry out customer turnover early warning. Under the framework of customer relationship management (CRM), this paper combs and analyzes the related theories of customer turnover management, and studies the prediction of retail customer turnover by using data mining technology. This paper first reviews the relevant theoretical knowledge of customer relationship management, the definition of customer loss, the causes and the process of customer loss management, and discusses the concept of customer value and several kinds of algorithms of customer value evaluation. Then a customer loss early warning model based on RFM customer value and IG-NN attribute selection is proposed. RFM model is used to calculate the customer value, and the information gain is used to select the main attributes. Then the influence degree of each main attribute on the customer turnover rate is analyzed by neural network, and the key attributes leading to customer loss are judged by the 28 ~ (th) rule, and the customer value and key attributes are taken as the input of the neural network. As the network output, the customer loss probability is used as the network output, and the customer loss early warning model based on RFM customer value and IG-NN attribute selection is constructed. Then the results obtained in this paper are compared with the single neural network and the customer turnover early warning model based on IG-NN attribute selection, and it is found that the customer turnover early warning model in this paper has the accuracy and hit rate. Coverage and improvement are superior to the other two models. Finally, the research conclusions of this paper are summarized, and the prospect of future research is put forward.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F274;F724.2;TP183

【参考文献】