Enhancing Bank Loan Approval Efficiency Using Machine Learning: An Ensemble Model Approach

Abstract

Lending is a major source of income for banks, but identifying worthy borrowers who will consistently repay loans is a constant problem. From a pool of loan applicants, conventional selection procedures frequently fail to find the most qualified individuals. To make loan applications faster, we created a new system that uses machine learning to automatically find people who qualify for loans. This comprehensive analysis involves data preprocessing, effective data balancing using SMOTE, and the application of various machine learning models, including Decision Trees, Support Vector Machines, K-Nearest Neighbors, Gaussian Naive Bayes, AdaBoost, Gradient Boosting, Logistic Regression, and advanced deep learning models like recurrent neural networks, deep neural networks, and long short-term memory models. We thoroughly evaluate the models based on accuracy, recall, and F1 score. Our experimental results demonstrate that the Extra Trees model outperforms its counterparts. Furthermore, we achieve a significant 0.62% increase in accuracy over the Extra Trees model by using an ensemble voting model that combines the top three machine learning models to predict bank loan defaulters. An intuitive desktop application has been developed to enhance user engagement. Remarkably, our findings indicate that the voting-based ensemble model surpasses both current state-of-the-art methods and individual ML models, including Extra Trees, with an impressive accuracy of 87.26%. Ultimately, this innovative system promises substantial improvements and efficiency in bank loan approval processes, benefiting both financial institutions and loan applicants.