Predicting academic performance in traditional environments at higher-education institutions using data mining: A review

Abstract

The purpose of this review paper is to survey the academic literature of the past five years in the area of data mining applied to the educational domain in a traditional classroom-based environment at Institutions of Higher Education (IHEs). We describe Educational Data Mining (EDM) and its main methods
and techniques as they are applied to predicting academic performance in a traditional setting, and we proceed then to review 56 primary-research articles on the subject. We also examine 5 other review papers that have preceded this one in the last five years. To our knowledge, this is the first review article to focus exclusively on applying EDM to predict academic performance in a traditional setting. We determine that classification is by far the most popular method used by the primary-research studies, followed by clustering and association rule mining. We conclude that the success experienced by researchers abroad in predicting academic performance can be replicated in Ecuador, provided that we avoid the pitfalls of overreliance on software and that we do not underestimate the complexities and need for human intervention that are involved in an EDM project. Moreover, the shortage of Big Data expertise in the country will need to start getting addressed.