Bank Loan Approval Prediction Using Machine Learning for Fraud Detection with Explainable AI

This research presented a machine learning-based approach for predicting bank loan approval using applicant financial and personal information. The proposed system utilized multiple input attributes, including age, annual income, employment experience, credit score, debt-to-income ratio, existing EMIs, loan amount, property value, loan tenure, marital status, and loan purpose, to classify loan applications into approved and rejected categories. Nine supervised machine learning algorithms, namely Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, Naïve Bayes, AdaBoost, Gradient Boosting, and Extreme Gradient Boosting (XGBoost), were implemented and comparatively evaluated using the same dataset and preprocessing strategy. The experimental evaluation demonstrated that the Decision Tree classifier achieved the best overall prediction performance with a testing accuracy of 97.98%, together with excellent precision, recall, F1-score, ROC-AUC score, and computational efficiency. Based on this comparative analysis, the Decision Tree classifier was selected as the final prediction model and successfully deployed through a Flask-based web application. The developed application enables real-time loan approval prediction while incorporating fraud detection and Explainable Artificial Intelligence to improve transparency and user confidence. The combination of machine learning, explainability, and web deployment provides an efficient, reliable, and practical solution for intelligent loan approval prediction in modern banking environments. The proposed system can assist financial institutions in improving decision accuracy, reducing processing time, and minimizing the risks associated with manual loan evaluation.

This research presented a machine learning-based approach for predicting bank loan approval using applicant financial and personal information. The proposed system utilized multiple input attributes, including age, annual income, employment experience, credit score, debt-to-income ratio, existing EMIs, loan amount, property value, loan tenure, marital status, and loan purpose, to classify loan applications into approved and rejected categories. Nine supervised machine learning algorithms, namely Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, Naïve Bayes, AdaBoost, Gradient Boosting, and Extreme Gradient Boosting (XGBoost), were implemented and comparatively evaluated using the same dataset and preprocessing strategy. The experimental evaluation demonstrated that the Decision Tree classifier achieved the best overall prediction performance with a testing accuracy of 97.98%, together with excellent precision, recall, F1-score, ROC-AUC score, and computational efficiency. Based on this comparative analysis, the Decision Tree classifier was selected as the final prediction model and successfully deployed through a Flask-based web application. The developed application enables real-time loan approval prediction while incorporating fraud detection and Explainable Artificial Intelligence to improve transparency and user confidence. The combination of machine learning, explainability, and web deployment provides an efficient, reliable, and practical solution for intelligent loan approval prediction in modern banking environments. The proposed system can assist financial institutions in improving decision accuracy, reducing processing time, and minimizing the risks associated with manual loan evaluation.