Predicting Customer Churn in the Telecom Industry Using Data Analytics

Abstract

Around the world, the telecommunications industry is rapidly expanding, as we are entering into a smartphone dominated era. Internet availability is now cited as a basic necessity and requirement for the generation today. With this, comes a competition amongst service providers to provide the best services to customers, along with the best prices to retain the already existing ones. Customers may choose to leave, for reasons known or unknown due to their experiences with a certain provider. Churn, simply put, is the process where a customer suspends or cancels his/her service with a provider. This paper presents a solution to this problem by recognizing those who may sway towards leaving, providing a vital solution to companies as retentive of existing customer is much easier than securing a new customer. Predictive, unsupervised models can organize and prevent such situations and can tell us what to expect in the near future. The research done here is an application of Logistic regression, Random Forests and K - Means clustering with the help of R - to predict churn. The data set consists of 3400 instances were considered in the dataset and 19 out of 22 attributes being decisive in the process of prediction.