Dynamic User Modeling in the Context of Customer Churn Prediction in Loyalty Program Marketing

Abstract

Daily human activities are dictated by real-life motivators such as schedules, locations, tastes and expectations. The similarity in the motivators can give rise to similarity in behavior. Thus, similarity in motivators can create a latent group among restaurant goers. In this paper such latent groups are identified and profiles for every latent group of restaurant goers are discovered/generated. User is modeled dynamically and HMM is used for user modeling based of the individual user’s weekly restaurant visit history. The weekly restaurant visit data used in this paper are real industrial data, collected from a collaborating restaurant marketing company which promotes restaurants through offering redeemable gifts. Group profiling of the users can help predicting the possibility of churn of a particular user group or segment of users for user retention purposes.