A prediction model for telecom customer lifecycle based-on CART tree

Abstract

Customer life-cycle prediction is the key to achieve long-term customer value. Current methods of customer lifecycle prediction are built on complete database. While the transaction volume, number of transactions and other numerical information are missing, those methods do not work. Here we try to predict telecom customer lifecycle based on the database of off-net customer which has only the information of demographics, then extract the demographics of the determinants of the classification of customer lifecycle by CART tree, finally gain the lifecycle of in-net customer. This model is applied for telecom operator in some province in China and good accuracy is achieved.