Optimasi Model Credit Scoring dengan Random Forest dan XGBoost

Abstract

This study highlights the importance of addressing class imbalance in credit score model datasets to achieve accurate risk prediction, especially in identifying rare default cases The SMOTETomek technique is proven to be effective in addressing class imbalance, significantly improving the performance of Random Forest and XGBoost models, especially in terms of recall (the ability to identify positive cases) Both models, after being optimized with SMOTETomek and Grid Search, showed excellent classification performance with accuracy above 93%, high precision, and significant recall improvement ROC analysis shows that XGBoost has slightly superior discriminatory ability compared to Random Forest in distinguishing between positive and negative classes