SYSTEMATIC REVIEW OF THE USE OF MACHINE LEARNING IN PREDICTING STUDENT ACADEMIC SUCCESS IN SOUTH WESTERN NIGERIAN UNIVERSITIES

Abstract

Machine learning (ML) has become an increasingly important tool in educational research, particularly in predicting student academic performance. While this field has matured in many developed countries, its application in Nigeria remains emergent and fragmented. This systematic review synthesizes existing literature on the use of ML models to predict academic success among students in Nigerian universities. To evaluate the types of ML models used, features employed for prediction, model performance metrics, and methodological quality of studies conducted within the Nigerian university context. Following the PRISMA 2020 guidelines, a systematic search was conducted across major academic databases including Google Scholar, AJOL, IEEE Xplore, Science Direct, and Scopus. Studies published between 2010 and 2024 that used ML algorithms to predict academic performance in Nigerian universities were included. Data were extracted and synthesized using a narrative approach. Twenty-seven studies met the inclusion criteria. Decision Trees, Naïve Bayes, Logistic Regression, and Support Vector Machines were the most commonly applied models. Frequently used features included CGPA, WAEC/UTME scores, gender, and attendance. Reported prediction accuracies ranged from 70% to 92%, with ensemble models and XGBoost showing the highest performance. However, most studies lacked rigorous validation, large datasets, or feature explain ability, and few explored deployment in real-world academic systems. Machine learning holds significant potential to enhance academic prediction and student support systems in Nigeria. However, the field is still developing, with gaps in data quality, model robustness, and institutional integration. It is recommended that future research should focus on adopting advanced techniques, improving dataset diversity, and promoting real-world application in policy and university administration.