Customer churn prediction using ensemble learning with neural networks

Abstract

Modern-day companies have to keep in mind their immensely large customer base. Their businesses strictly deal with the masses. Companies have to maintain a record of how their product or service is reaching the people. And to know Customer Behavior around the product or service provided. This vital piece of information will give understanding about customer's interests aligned with respect to the product or the service provided. Companies have to keep an eye on customers leaving. Churn Rate is a measure vital to understand the growth or fall of a company. Predicting customers who are willing to churn even before they are going to churn will be quite remarkable. Based on prediction, companies now can target and work better on customers predicted to be leaving. This paper suggests an Ensemble Learning based Model using Neural Networks. Three models of Neural Networks will share their prediction and most occurred prediction is selected to be the ftnal prediction. Proposed model scored 84% accuracy which is a decent improvement over individual model.