Student Academic Performance Prediction

Abstract

Given the increasing number of students who attend traditional and non-traditional classes that deploy internet-based educational resources and environments, large volumes of data are being generated on a daily basis. As a result, more researchers are now working with Educational Data Mining (EDM) methods to understand learning processes and behaviors of learners. The problem that led to this research is the need to make use of unused data that is collected during education and learning processes by gaining insights in order to support students in regards to their academic performance and in taking actions to prevent or warn students from failure. The main focus of this research is on how EDM can support student learning in regards to student academic performance, engagement, and intervention. The research mainly addresses the appropriate EDM methods used to predict student academic performance. Modeling and evaluation of several classifiers were conducted. As a result, Random Forest classifier has been chosen as the best model to be deployed in an interactive R Shiny application.