Predicting academic success by using context variables and probabilistic classification

Copyright © 2017 John Wiley & Sons, Ltd The aim of the study reported here was to predict students' grades based on context and personal state variables. Motivation for the study derives from the need to provide accurate recommendations about both educational resources and activities that match students' requirements and expectations. The proposed prediction method takes advantage of information associated with the context variables of the students, proposes the identification of clusters that group students with similar attributes, and estimates the final grade based on a probabilistic model. The findings show that the proposed model outperforms other existing models in terms of error accuracy.

keywords: clustering, context, e-learning, prediction, probability