Predicting Student Performance to Improve Academic Advising Using the Random Forest Algorithm
Mirna Nachouki, Mahmoud Abou Naaj
Abstract
The Covid-19 pandemic constrained higher education institutions to switch to online teaching, which led to major changes in students’ learning behavior, affecting their overall performance. Thus, students’ academic performance needs to be meticulously monitored to help institutions identify students at risk of academic failure, preventing them from dropping out of the program or graduating late. This paper proposes a CGPA Predicting Model (CPM) that detects poor academic performance by predicting their graduation cumulative grade point average (CGPA). The proposed model uses a two-layer process that provides students with an estimated final CGPA, given their progress in second- and third-year courses. This work allows academic advisors to make suitable remedial arrangements to improve students’ academic performance. Through extensive simulations on a data set related to students registered in undergraduate information technology program gathered over the years, we demonstrate that the CPM attains accurate performance predictions compared to benchmark methods.