AI-Powered Smart Loan Risk Prediction System: A Predictive Approach Using Machine Learning

This paper presents a Loan Risk Prediction System designed to analyze borrower information and predict the likelihood of loan default. The system integrates data pre- processing, machine learning techniques and an AI-enhanced suggestion mechanism to provide both predictive accuracy and actionable insights. Developed in Python and deployed via Streamlit, the application enables loan officers to interactively evaluate borrower risk, while also supporting bank management in making informed strategic decisions and enhancing portfolio oversight. A Random Forest classifier was selected for its proven ability to handle complex, nonlinear datasets effectively. The final model achieved an accuracy of 87.4 percent on unseen test data, with a high recall for default cases, thereby minimizing the risk of overlooking high-risk borrowers. Overall, the system demonstrates how AI-powered tools can enhance transparency, improve decision making efficiency, and deliver actionable recommendations in agricultural loan management, offering a scalable and practical prototype for real-world banking integration.