E-learning has advantages over traditional education thanks to its flexibility and scope. This study aims to test whether it is possible to predict students’ final outcomes solely based on their interaction with virtual resources at different stages of the course. Early prediction of the outcome of students at the first stages of the course is an advantage, allowing teachers to react quickly and provide the necessary support for the students who are in need of help. The effectiveness of different machine learning and deep learning models in predicting student performance throughout the stages has been evaluated using the OULA dataset. The models will predict whether a student will pass or fail, earn a distinction, or drop out of the course prematurely. The study shows that it is possible to predict student performance based on their interactions with virtual resources during each stage of the course.