Gyeongju-si: Advanced Institute of Convergence I T
Global competition has raised the cost of acquiring new customers, making it imperative to determine which customers are likely to churn. Although many researchers proposed many different techniques to construct good churn models, none of them focus on evaluating models effectiveness for real marketing activities and the verification process in real case. In this paper, we proposed three measurements to evaluate churn models in telecom industry. We propose modification measure, effectiveness measure, and cost benefits analysis to aid in evaluating the execution effectiveness for adjusting the data mining models or redesign marketing activities. An experimental architecture on telecom churn management application in real world is designed to evaluate the effectiveness of the discovered churn models. Not only the data mining models but also the retention activities influence the results of the churn management. The proposed experimental architecture can be a pilot project for real application in any CRM system. For those models which need be adjusted, the solution of models or marketing activity according to the effectiveness measures. In the future, we will focus integrating more domain knowledge, business considerations, and mining models into this system and solve the critical problems is customer relationship management in real business applications.
Journal of Convergence Information Technology 6(11), pp.192-200