Educational data mining for predicting academic performance in sub-Saharan African undergraduate stem students: a scoping review

This systematic literature review examines the effectiveness of educational data mining (EDM) in predicting the academic performance of undergraduate STEM students in Sub-Saharan African higher education institutions. The study analyses 41 peer-reviewed articles to identify key input factors and student involvement factors influencing academic success. The research categorizes these factors into four main themes: previous and current class performance, demographics, socio-economic factors, and e-learning behaviours. A conceptual framework is developed based on student involvement and Input-Environment-Outcomes theories to guide the investigation, highlighting the interplay of student characteristics, educational environment, and academic outcomes. The methodology employs a systematic literature review (SLR) using the PRISMA guideline, with searches conducted across Scopus, Google Scholar, and ProQuest databases. The findings demonstrate that while static data (demographics, prior records) are commonly used, dynamic data (e-learning behaviours) offer a more realtime perspective and understanding of student engagement and predictive accuracy.