Optimization and prediction of corporate credit rating through advanced feature selection based on AI and deep learning

Abstract

Accurate prediction of corporate credit ratings is essential for financial decision-making and risk management. This study offers a comprehensive evaluation of six machine learning algorithms—Random Forest (RF), eXtreme Gradient Boosting (XGBoost), Logistic Regression (LR), Support Vector Machine One-vs-One (SVM OVO), Support Vector Machine One-vs-All (SVM OVA), and Multi-Layer Perceptron (MLP)—in the context of corporate credit rating classification. A key contribution of this work is the introduction of a feature selection method based on RF, which significantly enhances model performance. We assess the effectiveness of these algorithms by comparing their predictive accuracy with all available features and with a reduced set of selected features. Our findings reveal that feature selection leads to substantial improvements in both classification accuracy and model efficiency, with XGBoost emerging as the top performer, closely followed by RF. This research provides valuable insights into how dimensionality reduction can improve the interpretability and robustness of credit rating models, offering practical guidance for more effective financial risk assessment.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 11 citations exist upstream.