Customer Churn Prediction Using Machine Learning Approaches

Abstract

Customer Churn (CC) is a major issue and important concerns for large organizations and businesses alike. Telecom industries are attempting to improve methods to predict possible customer churn due to the immediate impact on revenue, particularly in the telecom sector. This paper discusses the various ML algorithms used to construct the churn model that helps telecom operators to predict customers who are likely to churn. The experimental results are compared to predict the best model among various techniques. As a result, the use of the Random Forest combined with SMOTE-ENN outperforms best result than other in terms of Fl-score. According to our analysis, the maximum prediction is 95 percent based on Fl-score.