Predicting students' academic performance and modeling using data mining techniques

Abstract

In educational institutions and universities, the issue of study interruptions can be addressed by using e-learning. As a result, this field has recently attracted a lot of attention. In this study, we applied four machine-learning methods to predict students' academic progress: logistic regression, decision trees, random forests, and Naive Bayes. The Open University Learning Analytics Dataset (OULAD), which contains a subset of the OU student data, was the source of the student data for all of these techniques. There is information regarding the students' VLE interactions as well as their demographics. Nowadays universities frequently use data mining techniques to analyze available data and extract knowledge and information that helps in decision making. The percentage split and the 10-fold cross-validation are used to measure and compare the prediction performance of four classifiers. When employing the percentage split, it was shown that the Naive Bayes classifier performs better than other classifiers, obtaining an overall prediction accuracy of 93%. This study aims to assist teachers in enhancing students' academic performance.