This study aims to predict student performance using machine learning models based on the Open University Learning Analytics Dataset (OULAD). The dataset includes various features such as student assessments, virtual learning environment (VLE) interactions, and demographic data. Several machine learning models were applied, including Logistic Regression, Linear Discriminant Analysis, Random Forest, and Neural Networks, to predict student outcomes (Pass, Fail, or Distinction). The results demonstrate that models incorporating both weighted grades and pass rates outperformed those relying on single features. Although Neural Network models achieved the highest accuracy, they faced challenges in predicting failure cases. This paper offers insights into the performance of different models and proposes directions for future improvements.
)
