<![CDATA[The importance of academic classification in online learning platforms is increasingly recognized as it helps in assessing student performance, early detection of issues, and identifying factors that influence academic success. This study uses the Open University Learning Analytics Dataset (OULAD) to predict students' academic success in various classification areas, including Distinction vs Non-Distinction, Withdrawn vs Non-Withdrawn, Pass vs Non-Pass, and Pass vs Fail. The aim of this research is to compare machine learning and deep learning techniques, such as Random Forest, Gradient Boosting, AdaBoost, LightGBM, and Voting Classifier, with a deep learning model based on Dense Neural Networks (DNN) to produce the best possible predictions. Relevant features are also selected using feature selection and dimensionality reduction strategies, including autoencoders and Recursive Feature Elimination (RFE). The results show that LightGBM and Gradient Boosting perform best in several classifications, with an accuracy of 75.47% for Pass vs Fail. On the other hand, DNN requires further refinement but shows potential in handling more complex classifications. In addition to identifying students at risk of failing, this method provides a deeper understanding of the variables affecting academic success in online learning environments.]]>
Copyrights © 2025
