AI-Driven Credit Scoring Using Alternative Data: Unlocking Financial Access In Emerging Markets

Ng'Ang'A, Allan Arthur (2025) AI-Driven Credit Scoring Using Alternative Data: Unlocking Financial Access In Emerging Markets. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (951kB) | Preview
Preview	PDF (Configuration Manual) Download (779kB) | Preview

Abstract

One of the biggest issues that plagues emerging markets is that traditional credit scoring methods often exclude individuals who lack formal financial data and history which goes to limit these underserved populations access to credit. This study investigates the use of alternative financial data such as mobile money usage, financial behaviour and other key demographic factors to develop an AI-powered credit scoring model that is aimed at improving financial access for millions of underserved people in emerging markets.

Through the use of the 2024 Finaccess household survey from Kenya, I have assessed, trained and evaluated multiple machine learning models, including Logistic Regression, Random Forest, XGBoost, and LightGBM, in an attempt to predict creditworthiness by using self-reported loan applications as a substitute variable. The initial results revealed that Logotsic Regression outperformed the other models, achieving a recall of 68.2% and F1of 0.34. Unfortunately, there was significant class imbalance, where loan denials were way underrepresented and that compromised the fairness and effectiveness of the models.

To address this imbalance, we applied SMOTETomek resampling alongside class-weighted learnings. After accounting for the under-sampling, Logistic Regression still maintained strong performance with recall of 68.7% and F1 score of 0.3335 while random forest showed substantial improvement post SMOTETomek. SHAP explainability revealed that factors like mobile money usage, age and education are very influential predictors. The findings from this study highlight the promise and power that alternative financial data has in enhancing credit scoring and advancing equitable access to credit for underserved populations and boosts financial inclusion in emerging markets.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Onwuegbuche, Faithful UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence H Social Sciences > HG Finance > Credit. Debt. Loans. H Social Sciences > HG Finance > Fintech T Technology > T Technology (General) > Information Technology > Fintech Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in FinTech
Depositing User:	Ciara O'Brien
Date Deposited:	20 Aug 2026 11:28
Last Modified:	20 Aug 2026 11:28
URI:	https://norma.ncirl.ie/id/eprint/9577

Actions (login required)

View Item