Interpretable Machine Learning Models for Credit Risk Assessment

Abstract

For lending organizations, determining a borrower’s creditworthiness is essential to determining their capacity to repay loans. The prediction of credit scores using feature engineering and machine learning techniques is the main emphasis of this study. Using the Kaggle Family Credit Default Risk dataset, the AUC scores of several machine learning models are compared. Modern machine learning techniques, including well-established methods like Random Forest and Linear Support Vector Machines, can be effectively applied to credit scoring. Ensemble models, such as LightGBM, offer advantages like improved predictions and increased stability, making them well-suited for this specific use case. Combining predictions from multiple models often results in less noisy outcomes compared to using a single model, outperforming other techniques like XGBoost, SVMs, and logistic regression.

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 — 10 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.