A systematic review of the literature on machine learning application of determining the attributes influencing academic performance

Academic institutions operate in an extremely demanding and competitive environment. Some difficulties confronting most schools are delivering high-quality education to the students, developing systems for evaluating student performance, analyzing performance, and recognizing the future demands of their learners. Also, due to the paradigm shift due to the computerization of school data management, educational stakeholders, including the machine learning (ML) community, have taken an interest in analyzing performance traits using academic and non-academic factors. This systematic literature review identifies various machine learning methods based on 84 selected publications. It shows how researchers have been able to pattern-map student characteristics and their influence on school performance. An attempt is made to answer how the overall study coverage of student characteristics and the ML methods are employed to predict students’ performance. An analysis of the 84 papers highlights that, student characteristics predominantly influencing performance are academic and demographic attributes. The study further shows that classification and decision trees are the most widely used methods and algorithms. The review also reveals population and practical knowledge gaps due to a lack of research on basic academic performance and prescription of intervention plans for averting poor performance through mapping these influential characteristics to student accomplishment. To bridge these perceived gaps, the scope of the population sample needs a benchmarked dataset and embedding the appropriate intervention outlines that will map the learner’s performance early in their school life.