A Data-Driven Exploration of Factors Affecting Student Performance in a Third-Level Institution

Abstract

This paper describes an application of data mining techniques to the analysis of student academic records, collected at Limerick Institute of Technology, with the goal of acquiring clearer, evidence-based understanding of how a variety factors affect students' examination performance. To this end, a comprehensive dataset has been prepared. It has been analysed using a variety of machine learning/data mining techniques, in order to examine it from multiple perspectives. Of the techniques used, Bayesian networks have been found to be best in terms of yielding results that are comprehensible and meaningful. The results of this work provide a useful snapshot of factors affecting performance for the student group analysed, as well as demonstrating a process by which other institutions may analyse their own student groups.