A comparative study of machine learning algorithms for virtual learning environment performance prediction

Abstrak

Virtual learning environment is becoming an increasingly popular studyoption for students from diverse cultural and socioeconomic backgroundsaround the world. Although this learning environment is quite adaptable,improving student performance is difficult due to the online-only learningmethod. Therefore, it is essential to investigate students' participation andperformance in virtual learning in order to improve their performance. Usinga publicly available Open University learning analytics dataset, this studyexamines a variety of machine learning-based prediction algorithms todetermine the best method for predicting students' academic success, henceproviding additional alternatives for enhancing their academic achievement.Support vector machine, random forest, Nave Bayes, logical regression, anddecision trees are employed for the purpose of prediction using machinelearning methods. It is noticed that the random forest and logistic regressionapproach predict student performance with the highest average accuracyvalues compared to the alternatives. In a number of instances, the supportvector machine has been seen to outperform the other methods.