Loan Default Prediction Using Machine Learning: An Interpretable Economic Cycle-Aware Approach

Mangla, Karan Rakesh (2025) Loan Default Prediction Using Machine Learning: An Interpretable Economic Cycle-Aware Approach. Masters thesis, Dublin, National College of Ireland.

Preview

PDF (Master of Science) Download (1MB) | Preview

Abstract

In case of loan defaults, financial institutions are at a significant risk. Old fashioned credit rating metrics, which may be linear regression or expert opinion, do not keep up with the dynamic macroeconomic environment and behavior trends. The research paper suggests a machine learning solution to predict the probability of loan default based on integrating Extreme Gradient Boosting (XGBoost) with the Synthetic Minority Oversampling Technique (SMOTE) along with a new and interpretable set of macro-financial features dubbed Economic Cycle Risk Score (ECRS). The ECRS measures borrower exposure to recession, interest rate shocks, and macro-cycle stress, which allows risk analysts to understand the interaction between the economic condition and borrower behavior. The proposed model employed on 255,347 records of borrowers resulted in an accuracy of 0.89 and a 5 per cent improvement over the baseline models, implying that the model can be effective in detecting potential defaulters. Through experimental analysis, it has been demonstrated that the combination of SMOTE with minority classes increased the model during the detectivity of minority classes. The current paper provides a credit risk model that is interpretable, cycle conscious and combines sophisticated machine learning with economic arguments, which can serve as a flexible solution to financial institutions operating in volatile markets.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Anant, Aaloka UNSPECIFIED
Uncontrolled Keywords:	Loan Default Prediction; Credit Risk Modeling; Machine Learning; XGBoost; SMOTE; Economic Cycle Risk Score (ECRS); Financial Analytics; Explainable AI
Subjects:	H Social Sciences > HG Finance H Social Sciences > HG Finance > Banking H Social Sciences > HG Finance > Credit. Debt. Loans. Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	08 Sep 2026 09:10
Last Modified:	08 Sep 2026 09:10
URI:	https://norma.ncirl.ie/id/eprint/9878

Actions (login required)

View Item