Predicting customer churn in mobile networks through analysis of social groups

Abstract

Churn prediction aims to identify subscribers who are about to transfer their business to a competitor. Since the cost associated with customer acquisition is much greater than the cost of customer retention, churn prediction has emerged as a crucial Business Intelligence (BI) application for modern telecommunication operators. The dominant approach to churn prediction is to model individual customers and derive their likelihood of churn using a predictive model. Recent work has shown that analyzing customers' interactions by assessing the social vicinity of recent churners can improve the accuracy of churn prediction. We propose a novel framework, termed Group-First Churn Prediction, which eliminates the a priori requirement of knowing who recently churned. Specifically, our approach exploits the structure of customer interactions to predict which groups of subscribers are most prone to churn, before even a single member in the group has churned. Our method works by identifying closely-knit groups of subscribers using second order social metrics derived from information theoretic principles. The interactions within each group are then analyzed to identify social leaders. Based on Key Performance Indicators that are derived from these groups, a novel statistical model is used to predict the churn of the groups and their members. Our experimental results, which are based on data from a telecommunication operator with approximately 16 million subscribers, demonstrate the unique advantages of the proposed method. We further provide empirical evidence that our method captures social phenomena in a highly significant manner.

Impact of Social Network to Churn in Mobile Network

“…The model proposed by Richter et al [17] is similar to our research from the context standpoint and the observed attributes. However, the quantification of social relatedness between two subscribers is significantly more computationally expensive; specifically, the relatedness of two users is calculated by examining the number of calls to common neighbors.…”

“…Richter et al [17] proposed an approach for churn prediction for groups of interconnected users by only analyzing call data without demographic information. The authors first employed a novel, information theoretic based measure to quantify the connectivity between each pair of subscribers in the network.…”

A diffusion model for churn prediction based on sociometric theory

“…Some works include different centralities (e.g., betweenness and closeness) as additional features, some add features such as the number of neighbouring churners and the number of calls to neighbouring churners (Dierkes et al 2011), and others use calculated network attributes, such as the neighbour composition, tie strength, similarity, and homophily (Zhang et al 2010(Zhang et al , 2012. Richter et al (2010) proposed a systematic study that evaluates the relevance of group-related features to churn.…”

Applications of Data Mining Techniques in Telecom Churn Prediction

“…From a machine learning perspective, churn prediction is a supervised (i.e. labeled) problem defined as follows: Given a predefined forecast horizon, the goal is to predict the future churners over that horizon, given the data associated with each subscriber in the network [7]. The churn prediction problem represented here involves 3 phases, namely, i) training phase, ii) test phase, iii) prediction phase.…”