Customer churn is an important and challenging problem that faces telecommunication companies worldwide. Recently, companies have been investing more in developing accurate prediction models which ...can forecast which customers are about ending their subscriptions or switching to another competitor service provider. These models can help Customer Relationship Management (CRM) in designing effective strategies to retain current customers. In this article, a Support Vector Machine (SVM) model is developed, for predicting customer churn in a local telecommunication company. In general, SVM models are parametric and the initial values of its parameters have great influence of its accuracy and performance. Therefore, a new combined evaluation metric is applied, while tuning the parameters of SVM, in order to maximize its effectiveness for churn management.
PDF
