CHURN PREDICTION FOR TELECOMMUNICATION SERVICE

Abstract

This paper evaluates different Machine Learning models which the aim to find out the good algorithm for predicting churners in prepaid telecom industry. Several models have been used and compared on the basis of different Data Mining methods and algorithm (Naïve Bayes, KNearest Neighbours, Decision Tree, Random Forest, and Support Vector Machine). To handle the imbalance in the data, we use Receiver Operating Characteristics (ROC) curves to evaluate the result beside the accuracy and churn rate. For the modeling examples we used RStudio analysis tool and introduced the R language.