A Fuzzy Logic-Based Early Warning System for Identifying At-Risk Students under Uncertainty and Imprecise Indicators
BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE · 30.07.2026 · 10 (3) · 154-162
DOI: 10.64290/bima.v10i3.1677 ↗Kurz gefasst
Fachartikel in der Zeitschrift BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE, online erschienen am 30.07.2026. Verfasst von Umar Ahmed und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Academic institutions face challenges in identifying at-risk students because academic vulnerability often involves uncertain, incomplete, and imprecise indicators that conventional Early Warning Systems may not adequately capture. This study developed and evaluated a Fuzzy Logic-Based Early Warning System (FL-EWS) to provide interpretable student-risk assessments under uncertainty. The system employed a Mamdani-type fuzzy inference engine with 28 rules and six input variables: GPA, study hours, attendance, class participation, social engagement, and socioeconomic status. It was implemented using Python’s scikit-fuzzy library and evaluated on a dataset of 1,000 student records, of which 771 were classified as at risk. Performance was assessed using accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC-ROC), with comparisons against conventional machine learning models. The FL-EWS achieved an accuracy of 0.365, precision of 1.000, recall of 0.175, an F1-score of 0.298, and an AUC-ROC of 0.719. It classified 26.2% of students as low risk, 65.9% as moderate risk, and 7.9% as high risk. Attendance (−0.424) and GPA (−0.376) showed the strongest correlations with the risk label. Although the system provided human-readable explanations to support targeted interventions, conventional models achieved higher accuracy, ranging from 0.885 to 1.000. The low recall indicates that the FL-EWS may fail to identify many at-risk students, while the perfect precision should be interpreted cautiously. These findings demonstrate the potential of fuzzy logic for transparent risk assessment but highlight the need to improve predictive performance and validate the results. The study concludes that FL-EWS may complement existing early warning systems, with future research focusing on adaptive membership functions, real-time dashboards, additional student-support indicators, and multi-institutional validation.
Zitieren
APA
Ahmed, U., Abubakar, M., & Baba, U. A. (2026). A Fuzzy Logic-Based Early Warning System for Identifying At-Risk Students under Uncertainty and Imprecise Indicators. BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE, 10(3), 154-162. https://doi.org/10.64290/bima.v10i3.1677
Vancouver
Ahmed U, Abubakar M, Baba UA. A Fuzzy Logic-Based Early Warning System for Identifying At-Risk Students under Uncertainty and Imprecise Indicators. BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE. 2026;10(3):154-162. doi:10.64290/bima.v10i3.1677
BibTeX
@article{Ahmed2026,
author = {Ahmed, Umar and Abubakar, Mohammed and Baba, Usman Ahmad},
title = {A Fuzzy Logic-Based Early Warning System for Identifying At-Risk Students under Uncertainty and Imprecise Indicators},
journal = {BIMA JOURNAL OF SCIENCE AND TECHNOLOGY GOMBE},
year = {2026},
volume = {10},
number = {3},
pages = {154--162},
doi = {10.64290/bima.v10i3.1677}
}
EUVIMED