Multi-stage mortgage default prediction using ensemble machine learning: a comparative framework

Afsaneh Azimi () and Navid Khaledian ()
Additional contact information
Afsaneh Azimi: University of Westminster
Navid Khaledian: University of Luxembourg

Digital Finance, 2025, vol. 7, issue 4, No 20, 1093-1118

Abstract: Abstract Mortgage default prediction is a critical component of credit risk management in the financial sector. Traditional statistical methods, such as logistic regression, often fail to capture complex borrower behavior and adapt to rapidly changing economic conditions. In this study, we present a robust machine learning framework that employs advanced ensemble methods, including LightGBM, XGBoost, Gradient Boosting, AdaBoost, Random Forest, and Extra Trees to enhance the prediction of mortgage delinquencies and foreclosures. A novel contribution of this work is the development of three specialized models targeting distinct stages of mortgage distress: the Early Alert Analysis Model (30–89 days delinquent), the Critical Risk Assessment Model (90 + days delinquent), and the Foreclosure Predictor Model. The models are trained and validated using a comprehensive dataset of 44,168 U.S. mortgage records, enriched with over 100 features from the FHFA NMDB® and macroeconomic indicators from the Federal Reserve Economic Data (FRED). Leveraging automated machine learning (AutoML) via PyCaret and advanced feature engineering techniques, our models achieve exceptional predictive performance, with AUC scores reaching up to 0.9998. The results demonstrate the superiority of machine learning approaches over traditional models in identifying early-stage risk, severe delinquencies, and imminent foreclosures. This research provides actionable insights for financial institutions and highlights the value of integrating adaptive, interpretable ML systems into modern mortgage risk assessment frameworks.

Keywords: Mortgage default; Credit risk; Ensemble learning; Automated machine learning; Financial risk prediction (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s42521-025-00160-5 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:digfin:v:7:y:2025:i:4:d:10.1007_s42521-025-00160-5

Ordering information: This journal article can be ordered from
https://www.springer.com/finance/journal/42521

DOI: 10.1007/s42521-025-00160-5

Access Statistics for this article

Digital Finance is currently edited by Wolfgang Karl Härdle, Steven Kou and Min Dai

More articles in Digital Finance from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().