University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
The Impact of Synthetic Data Augmentation on Algorithmic Bias in Credit Risk Assessment
Mirzoev, Vladislav and Machado, Marcos Roberto (2025) The Impact of Synthetic Data Augmentation on Algorithmic Bias in Credit Risk Assessment.
 | There is a more recent version of this item available. |
|
PDF
669kB |
| Abstract: | Credit risk assessment increasingly relies on Machine Learning (ML) to automate decisions, yet imbalanced datasets, where loan defaults are significantly underrepresented, can compromise model precision and fairness. This research will investigate how synthetic data augmentation techniques, such as Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic (ADASYN) Sampling Approach, affect demographic bias in credit decision-making. By running experiments on the publicly available XYZ Corp dataset that we retrieved from Kaggle.com, we study the interdependence between data balancing techniques and potentially new bias. This correlation might play an important role in the meaningful usage of ML models in the field of credit risk assessment. The results indicate that both SMOTE and ADASYN effectively improved classification performance by addressing class imbalance. However, fairness evaluations revealed that these augmentation methods also had measurable effects on algorithmic bias. In some cases, we noticed reduced disparities; however, the disparities were also amplified, depending on the protected attribute. These findings emphasize the importance of combining performance-driven techniques with fairness assessments when deploying ML models in credit risk settings. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Business & IT BSc (56066) |
| Link to this item: | https://purl.utwente.nl/essays/107708 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page