Enhancing Student Performance in South African Higher Education through Machine Learning and Big Data Analytics: A Theoretical Framework

Abstract

The paper explores using machine learning algorithms and big data analytics to enhance academic and psycho-socio performance among students. It suggests that South African higher education can benefit from appropriate technology and data application to address issues like lack of personalised support, varied student readiness, and resource constraints affecting performance. Employing adaptive structuration theory (AST), derived from Giddens' structuration theory, the study examines how institutions incorporate new technologies to improve student outcomes adaptively. Through a thorough literature review encompassing student performance factors, teaching methodologies, technology integration, and big data analytics (BDA), the research proposes establishing an IT infrastructure capable of integrating diverse student data types for machine learning analysis. This data-driven approach aims to personalise curricula, identify at-risk students, and enhance pedagogy to bolster learning outcomes. By bridging technological capabilities with practical implementation, the study offers a framework for local universities to make informed, data-driven decisions tailored to their challenges. It underscores the potential for data analytics to create supportive, personalised learning environments conducive to student success within the South African higher education context. The key findings show that it is possible to accurately predict student performance through machine learning algorithms even with different data sets. The analysis of the collected data showed that students' marital status and academic achievements in the previous years affected the study results. Applying big data analytics in South African higher education institutions can potentially enhance student support and resource distribution. However, there are issues of data heterogeneity and ethical issues.