Analysis of Usage Statistics to predict High Value Customer churn in Telecom Industry

Abstract

It is a well-known fact that the retention of customers is essential for any operator in the telecom industry. What is more important is the retention of customers who are of high value. The proposed predictive model can identify the customer indicators at an increased risk to churn by using machine learning algorithms such as Logistic Regression and dimensionality reduction techniques such as Recursive Feature Elimination and Principal Component Analysis. A model that provides higher accuracy in predicting customer churn and tells us about the important variables of usage behavior has been proposed.  Few suggestions from the management perspective have also been given to reduce the customer churn rate.