Dynamic churn prediction framework with more effective use of rare event data: The case of private banking
作者：
Highlights：
• Dynamic churn prediction framework for creating training data from customer records.
• Improves accuracy significantly even vs. balanced data, across prediction horizons.
• Independently trained binary classifiers approach outperforms survival analysis.
• Horizon specific ranking allows targeting retention efforts across time and customers.
• Allows capturing the effect of the environmental conditions on churn probability.
摘要
•Dynamic churn prediction framework for creating training data from customer records.•Improves accuracy significantly even vs. balanced data, across prediction horizons.•Independently trained binary classifiers approach outperforms survival analysis.•Horizon specific ranking allows targeting retention efforts across time and customers.•Allows capturing the effect of the environmental conditions on churn probability.
论文关键词：Dynamic churn prediction,Data mining,Customer retention,Private banking,Customer relationship management,Rare event,Sampling,Training data generation
论文评审过程：Available online 19 June 2014.
论文官网地址：https://doi.org/10.1016/j.eswa.2014.06.018
