Advanced Framework for Loan Approval Predictions Using Artificial Intelligence-Powered Financial Inclusion Models

The loan approval procedure is an essential activity for banks as loans contribute significantly to their revenues. But managing the risks of loan defaults and appropriately assessing loan requests are becoming more and more difficult. Machine learning techniques have shown to be excellent in processing enormous volumes of data. Thus, scientists have started researching approaches for anticipating loan approval processes. The research employs machine learning techniques for both binary classification and regression analysis to evaluate financial risk factors influencing loan approval dataset. Three machine learning models—LightGBM, Random Forest, and CatBoost—are trained using Bayesian optimization for hyperparameter tuning. Classification models are evaluated using F1-score, recall, accuracy, and precision, while regression models are assessed based on MAE, MSE, RMSE, and R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score. Experimental findings indicate that LightGBM delivers the highest classification accuracy, 96.23% while maintaining a strong balance between precision and recall. In regression tasks, CatBoost demonstrates superior performance, achieving the lowest prediction errors and the highest R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score 0.8820. A comparative evaluation against baseline models, including Naïve Bayes and XGBoost, underscores the effectiveness of the proposed models, with LightGBM and CatBoost significantly enhancing classification accuracy and precision. Despite challenges such as class imbalance and dataset biases, the study emphasizes the transformative potential of AI-driven decision-making systems in facilitating fair and data-driven loan approval processes.