LSTM Model to Predict Customer Churn in Banking Sector with SMOTE Data Preprocessing

Abstract

In any organization, the technique used to acquire new users is by Customer Relationship Management systems. In order to achieve more profitability with increase in customer retention is by maintaining a healthy association with them. Customer Churn is also known as Customer Loyalty or Retention. The inspiration behind churn forecast is to categorize and discover clients into churner & non-churner. A churned client implies there is a greater chance of the client is around to take off from the organization. A novel software can be utilized to discover the clients who will donate increased benefits for the organization. Moreover churn forecast can maintain a strategic distance from the misfortune of income by holding the existing clients. A few procedures are accessible for churn prediction with ensemble and hybrid models. This paper points to anticipate client Churn in banking sector with LSTM model and the data is preprocessed using SMOTE technique to overcome imbalanced information. The work is an extension to predicting customer loyalty in banking sector using Mixed Ensemble and Hybrid model. This paper proposes an accurate way to predict customer churn using LSTM model and the data is preprocessed using SMOTE technique. In this way the framework is more valuable for organizations to discover the clients with more chances to become churn. The results of the evaluation indicated that this is to be the case, the proposed systems for churn prediction performs with an accuracy of 88% and which is much better than the system without SMOTE technique.