Churn Prediction in the Telecommunications Sector Using Support Vector Machines

Abstract—In these days, due to challenges resulted from global competition, customer churn represents one of the significant concerns for companies in different industries. With a churn rate of 30%, the telecommunication sector takes the first place on the list. In order to solve this problem, predictive models need to be implemented to identify customers who are at risk of churning. In this paper, an advanced methodology for predicting customers churn in mobile telecommunications industry is presented. The dataset used, contains call details records and has 21 attributes for each of its 3333 records. We use a Support Vector Machines algorithm with four kernel functions to implement the predictive models. The performance of the models is evaluated and compared using gain measure. Keywords—Call details records, churn prediction, support vector machines, telecommunications. C I.