Analysing University at-Risk Students in a Virtual Learning Environment using Machine Learning Algorithms

Abstract

Students at risk in universities are becoming a rising global issue. These are students who have a high likelihood of dropping out of their respective academic programs. Due to the importance and impact on students, if interventions are not implemented, research on students at risk is garnering widespread attention in the literature. Early identification of these at-risk students is essential for intervention to lessen the likelihood of dropout. On a virtual learning environment dataset, this study compared Adaboost with five other machine learning algorithms, including Random Forest, Logistic Regression, Support Vector Machine, and Decision Trees, to detect students at risk. This study focused on training and evaluating the six machine learning models adopted, employing performance evaluation metrics such as F1 score, Confusion Matrix, Recall, Precision, ROC, error rate, and accuracy. Adaboost was found to be the top-performing algorithm, having the highest accuracy, F1 score, Precision, and Recall.