Designing for Fairness in Higher Education Early WarningSystems
Proceedings of the Canadian Conference on Artificial Intelligence · 27.05.2024
DOI: 10.21428/594757db.c7aec3c1 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the Canadian Conference on Artificial Intelligence, online erschienen am 27.05.2024. Verfasst von Kelly McConvey und 1 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref).
Abstract (Originalsprache)
Early Warning Systems are algorithms trained to predict the probability of a student graduating successfully. Despite the growing reliance on algorithms by Higher Education Institutions and significant interest in the development of these algorithms by the research community, there remain several critical questions related to the fairness of Early Warning System algorithms. In this case study, we assessed a model built on the Open University Learning Analytics dataset (OULAD) for gender and disability-related performance disparities. We found that our model displayed significant disparities be-tween classes, indicating a higher likelihood of unnecessary interventions for women and individuals with recognized disabilities; and a lower likelihood of receiving interventions when needed. We then tested two mitigation approaches and found that a reductions approach is a viable solution for mitigating unfairness in Early Warning System.
Zitieren
APA
McConvey, K., & Guha, S. (2024). Designing for Fairness in Higher Education Early WarningSystems. Proceedings of the Canadian Conference on Artificial Intelligence. https://doi.org/10.21428/594757db.c7aec3c1
Vancouver
McConvey K, Guha S. Designing for Fairness in Higher Education Early WarningSystems. Proceedings of the Canadian Conference on Artificial Intelligence. 2024. doi:10.21428/594757db.c7aec3c1
BibTeX
@article{McConvey2024,
author = {McConvey, Kelly and Guha, Shion},
title = {Designing for Fairness in Higher Education Early WarningSystems},
journal = {Proceedings of the Canadian Conference on Artificial Intelligence},
year = {2024},
doi = {10.21428/594757db.c7aec3c1}
}
EUVIMED