计算机工程与应用 ›› 2022, Vol. 58 ›› Issue (11): 84-92.DOI: 10.3778/j.issn.1002-8331.2104-0221

• 大数据与云计算 • 上一篇    下一篇

电子商务客户流失的DBN预测模型研究

周婉婷,赵志杰,刘阳,王加迎,韩小为   

  1. 1.哈尔滨商业大学 计算机与信息工程学院,哈尔滨 150028
    2.哈尔滨商业大学 黑龙江省电子商务与信息处理重点实验室,哈尔滨 150028
  • 出版日期:2022-06-01 发布日期:2022-06-01

Research on DBN Prediction Model of E-Commerce Customer Churn

ZHOU Wanting, ZHAO Zhijie, LIU Yang, WANG Jiaying, HAN Xiaowei   

  1. 1.School of Computer and Information Engineering, Harbin University of Commerce, Harbin 150028, China
    2.Heilongjiang Key Laboratory of Electronic Commerce and Information Processing, Harbin University of Commerce, Harbin 150028, China
  • Online:2022-06-01 Published:2022-06-01

摘要: 在电子商务迅速发展,企业快速抢占市场的背景下,客户成为企业竞争的核心因素。现有相关研究多致力于采用全数据输入模式解析客户流失现象,不同类型客户造成的差异性还有待进一步探讨。鉴于传统RFM模型不能精确解释电子商务客户流失原因,该研究将客户分为活跃与非活跃两个集群,提出一种优化的RFM理论模型与深度信念网络实证模型对电子商务客户流失进行预测。结果表明,不同类型客户流失因素的影响强度不同。对活跃用户而言,客户购买总金额是影响客户流失的主要因素;对非活跃用户而言,客户进入店铺的时间越长越可能留住客户。通过剖析非活跃用户不流失和活跃用户流失的原因,可帮助企业制定有效的客户管理策略,以最大程度地吸引潜在客户及保留现有客户,获取最多的市场利益。

关键词: 电子商务, 客户流失预测, 活跃客户细分, RFM模型, 深度信念网络

Abstract: With the rapid development of e-commerce and the rapid market share of enterprises, customers have become the core factor of competition among enterprises. Existing related studies are mostly devoted to analyzing the phenomenon of customer churn using full data input mode, and the differences caused by different types of customers need to be further explored. In view of the traditional RFM model can not accurately explain the reasons for e-commerce customer churn, this study divides customers into active and inactive clusters, and proposes an optimized RFM theoretical model and an empirical model of deep belief network to predict e-commerce customer churn. The results show that the influence intensity of different types of customer churn factors is different. For active users, the total amount of customer purchases is the main factor affecting customer churn; for inactive users, the longer the customer enters the store, the more likely it is to retain the customer. By analyzing the reasons why inactive users do not churn and active users churn, it can help enterprises formulate effective customer management strategies to attract potential customers and retain existing customers to the greatest extent, so as to obtain the most market benefits.

Key words: e-commerce, customer churn prediction, active customer segmentation, RFM model, deep belief network