Customer retention is very important issue. In this reasons, many studies have been conducted for customer retention to demonstrate the potential of data mining through experiments and case studies. But current customer retention procedures only focus on detection of potential defectors based on likelihood of defection overlooking cost. In this paper, we propose an economic procedure for profit maximizing based on the customers' LTV and behavior state transition cost. For the aim of this paper, we use past and current customer behavior by integrating data mining techniques. In this procedure, SOM is used to determine the possible states of customer behavior from past behavior data. Based on this state representation, a Markov chain is applied to generate likelihood of state transition.
Chae, Kyunghee; Song, Heeseok; and Choi, Juchoel, "A Defection Detection Procedure Using SOM and Markov Chain : A Case of On-Line Game Provider" (2004). ICEB 2004 Proceedings (Beijing, China). 48.