Bank Churn Prediction: A Holistic Churn Analysis on Imbalanced Data Using ML Techniques

In today’s modernized world, there is heavy use of technologies in almost every sector of service, which can be used to tackle the challenges faced by the industry. One such example is the bank, which faces the problem of its customers defecting. Bank customer churn has proved to be a critical problem in the banking sector and thus, preventing customer churn is of utmost importance to the bank. This research aims at making the process of identifying potential churners much easier through rigorous training of various machine learning models. The primary focus is to compare the efficacy of four machine learning approaches—Artificial Neural Networks (ANN), Random Forest, XGBoost and Support Vector Machines (SVM)—in forecasting bank customer turnover. It additionally evaluates the effectiveness of addressing class imbalance using the Synthetic Minority Over-Sampling Technique (SMOTE) with the above mentioned classifiers. The comparative analysis includes assessing the performance of the four classifiers without using SMOTE and then comparing it with their performance with SMOTE in terms of metrics such as accuracy, precision, and recall. The study aims to provide insights into the impact of addressing class imbalance on prediction accuracy and the effectiveness of different classifiers in predicting customer churn while considering multiple evaluation metrics.