Predicting Undergraduate Academic Success with Machine Learning Approaches

The opportunity to pursue tertiary education has increased in recent years, attributed to the initiatives and efforts made by governments, industry players, and educational institutions to make education more affordable and accessible. Hence, predicting undergraduate academic performance plays a crucial role in higher learning institutions’ success. This is because the predicted outcomes could provide valuable insights and benefits such as early intervention and support for students at risk of academic struggles, graduating on time, personalized learning, etc. This study selected two machine learning algorithms, i.e., Light GBM and Random Forest, to predict undergraduate academic success in a higher-learning institution. The preliminary result indicates that Light GBM is the best performer, obtaining the highest accuracy of 79.3% after hyperparameter tuning. The result is marginally better than the recently published works that used identical or highly similar datasets.