An SVM based churn detector in prepaid mobile telephony

The context of prepaid mobile telephony is specific in the way that customers are not contractually linked to their operator and thus can cease their activity without notice. In order to estimate the retention efforts which can be engaged towards each individual customer, the operator must distinguish the customers presenting a strong churn risk from the other. This work presents a data mining application leading to a churn detector. We compare artificial neural networks (ANN) which have been historically applied to this problem, to support vectors machines (SVM) which are particularly effective in classification and adapted to noisy data. Thus, the objective of this article is to compare the application of SVM and ANN to churn detection in prepaid cellular telephony. We show that SVM gives better results than ANN on this specific problem.