Paper Title
Comparative Analysis of Customer Churn Prediction Models in Banking

Abstract
This study provides a comparative analysis of machine learning algorithms used in forecasting bank churn, which is an important issue for financial organisations attempting to assist bank managers and top-level officials in developing strategies to reduce customer attrition and improve retention measures. The algorithms examined include Logistic Regression, Random Forest, SVC, Decision Tree, XGB, and ANN, which are all commonly used in predictive modelling. A collection of historical client information, including demographics, transactional history, and account details, was created using techniques similar to those used by large banks. Preprocessing measures were made to handle missing data, encode category variables, and normalise numerical properties. The dataset was then divided into training and testing subgroups for model building and assessment. Performance criteria like as accuracy, precision, recall, and F1-score were used to assess each algorithm's performance on testing data after training on the training set. According to the comparison analysis, Random Forest outperformed other algorithms in terms of predicted accuracy and overall performance, with an F1-score of 0.62 and an accuracy of 86.86%. Furthermore, SVC showed encouraging results, with an F1-score of 0.57 and an accuracy of 86.26%. Random Forest and SVC outperformed Logistic Regression, Decision Tree, XGB, and ANN, with accuracies of 80.66%, 80.30%, 85.00%, and 85.96%, respectively. These findings provide useful insights into the efficacy of various machine learning algorithms for bank churn prediction and can drive decision-making processes for financial institutions looking to improve client retention efforts. Keywords - ML algorithms, Bank churn, Finance, Customer retention, Regression, Forest, SVC, Decision Tree, XGB, ANN, Historical data, Preprocessing, Performance metrics, Decision- making.