Churn Prediction using Neural Network based Individual and Ensemble Models

Churn prediction is still a challenging problem in telecom industry. Many data mining techniques have been employed to predict customer churn and hence, reduce churn rate. Although a number of algorithms have been proposed, there is still room for performance improvement. Therefore this paper evaluates existing individual and ensemble Neural Network based classifiers and proposes an ensemble classifier which utilizes Bagging with Neural Network in order to improve performance measures resulting in better accuracy for churn prediction. This work employs two benchmark datasets, obtained from GitHub, for comparison and evaluation of the proposed model. An average accuracy of 81% is achieved by the proposed model.