Customer retention prediction with CNN

Yen Huei Ko, Ping Yu Hsu, Ming Shien Cheng, Yang Ruei Jheng, Zhi Chao Luo

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations


The prediction of customer retention provides competitive advantage to enterprises. When customer purchases more with satisfaction, it will increase customer retention. Customer repurchase behavior represents customer with satisfaction on enterprise to buy again resulting in customer retention. In e-commerce and telemarketing, trust and loyalty are key factors influencing customer repurchase behavior. In the past, most researches were relied on questionnaire survey to collect data. The drawbacks of such approach are those participants may not be willing to fill the lengthy questions which caused low data collection rate and even low quality data being collected. This study is to apply data driven techniques to extract information from transaction logs in ERP system utilized to compute trust and loyalty based on the verified formulation. The values of defined trust and loyalty are treated as independent variables to predict customer repurchase behavior by Convolutional Neural Networks (CNNs). The prediction accuracy reaches 84%.

Original languageEnglish
Title of host publicationData Mining and Big Data - 4th International Conference, DMBD 2019, Proceedings
EditorsYing Tan, Yuhui Shi
PublisherSpringer Verlag
Number of pages10
ISBN (Print)9789813295629
StatePublished - 2019
Event4th International Conference on Data Mining and Big Data, DMBD 2019 - Chiang Mai, Thailand
Duration: 26 Jul 201930 Jul 2019

Publication series

NameCommunications in Computer and Information Science
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937


Conference4th International Conference on Data Mining and Big Data, DMBD 2019
CityChiang Mai


  • CNN
  • Customer repurchase behavior
  • Customer retention
  • Data driven technique
  • Loyalty
  • Trust


Dive into the research topics of 'Customer retention prediction with CNN'. Together they form a unique fingerprint.

Cite this