Review of theoretical advancements in AI/ML classification models for credit risk assessment

Abstract

In the realm of credit risk assessment, the utilisation of artificial intelligence (AI) and machine learning (ML) classification models has become increasingly prevalent. This paper thoroughly investigates latest advancements in AI/ML classification models for credit risk assessment, which are crucial for assessing the creditworthiness of individuals and businesses. Key findings reveal that modern AI/ML techniques, particularly Random Forest and XGBoost, outperform traditional logistic regression methods. Additionally, interpretability techniques, including Shapley Additive exPlanations (SHAP) and feature importance analysis, improve the understanding and transparency of model predictions. This paper synthesises recent research findings and industry developments to provide practitioners and researchers with insights into model selection, evaluation metrics and explanation techniques, thereby contributing to the ongoing evolution of credit risk management strategies in the financial sector.