Evaluation of Credit Risk in Indian Banks: A Comparative Approach Using Panel Data Regression and Random Forest
DOI: https://doie.org/10.10399/IJBE.2026977271
Koushika. T, Dr V.Vasanthakumar
Views: 82 | Downloads: 49
Keywords:
Credit Risk, Panel Data Regression, Decision Tree and Random Forest.
Abstract:
Purpose: This study evaluates the comparison of classical econometric panel data regression and robust machine learning techniques in the prediction of credit risk across selected banks in India.
Methodology: The research analyses the panel data from 2020-21 to 2024-25. The traditional fixed-effects regression model was validated against ML classifiers such as Random Forest and Decision Trees. To assess predictive accuracy in credit risk management.
Findings: The outcome of the study exhibits that an increase in NPA will cause a credit risk in the banking sector. While the financial indicators, such as Return on Assets (ROA) and Capital Adequacy Ratio (CAR), help to eliminate NPAs. Among these models tested were the Random Forest and the decision tree, which provide valuable insights for Banking and financial institutions, Risk Managers, Regulators, and Policy makers in terms of valuing AI tools to enhance financial oversight for decision-making in the banks.
Originality: This research fills a critical gap in the literature by examining benchmark performance in predicting non-linear machine learning ensembles against econometric tools within the robust supervisory regulatory framework in the banking sector
Practical Implication: Banking and financial institutions can significantly enhance predictive accuracy in the detection of fraud and default in risk management by taking proactive measures to mitigate NPAs. Besides, integrating AI & ML can provide early detection warning, and it facilitates a shift towards reactive to proactive risk management practices in the banks.