Performance Evaluation Of Machine Learning Techniques For Prediction Of Graduating Students In Tertiary Institution

Abstract

Near accurate prediction of students’ future performance based on their historical academic records is important for effective pedagogical interventions. It is imperative to provide an enhanced prediction system that can assist educational institutions to identify and monitor students at different threshold and to focus on improving students that their threshold is less than graduation at early stage. Studies on the prediction of graduating students using data mining techniques have been widely carried out in the existing literature. The paper applied Baye’s theorem and Artificial Neural Networks (ANN) to build a predictive model for the likelihood of students’ graduation in a tertiary institution. The prediction was performed on four variables- Unified Tertiary Matriculation Examination (UTME), Number of sittings for O’level (NOS), Grade Points of O’level (Grade) and Mode of Entry (PreND). The implementation was carried out in Rstudio environment. The results showed that ANN had higher accuracy compared to Bayesian Classification. ANN performed better because of the learning rules it contains.