Mining social networks for customer churn prediction
Authors: Wouter Verbeke, Karel Dejaeger, Thomas Verbraken, David Martens, and Bart Baesens
Presented at: Interdisciplinary Workshop on Information and Decision in Social Networks, Cambridge, 31 May - 1 June 2011
Keywords: Classification, customer churn prediction, social networks
Customer churn prediction models aim to detect customers with a high propensity to attrite. This study investigates the applicability of relational learning techniques to predict customer churn using social network information. A range of existing, extended, and novel relational classifiers and collective inference procedures have been (re-) implemented and applied on two large-scale real life data sets obtained from international telco operators, containing both networked (call detail record data) and non-networked (usage statistics, socio-demographic, marketing related) information about millions of customers. The results of the experiments indicate the existence of a limited but relevant impact of network effects on customer churn behavior. Combining a relational and a local classifier therefore improves the predictive power of a customer churn prediction model compared to a stand-alone local or relational classifier. Collective inference procedures however are shown to have a negative impact on the classification performance.