Emerging technologies, particularly artificial intelligence (AI) and machine learning (ML) algorithms, present valuable opportunities to analyse learning management system data (LMS) and considered as the corner stone of Learning analytics (LA). The aim is to analyse student performance during a course or a whole academic year. In particular, it identifies at risk students and enables educators to timely support this student’s category and can provide clear guidance to improve teaching and learning strategies. This is why implementing an Early-Warning System is very crucial in this context to alert at risk student with weak performance, during first course sessions, to mitigate potential failures. The primary objective of this paper is to develop and implement an early-warning system that assists educators in identifying these students requiring attention and prompts them to be aware of their academic progress, thereby facilitating timely interventions to reduce the risk of failure. This study target web technologies course for the third-year engineering level in Esprit school of engineering, Tunisia. The experiments involve testing and comparing the performance of various classifiers, with a focus on Logistic Regression (LR), Random Forest (RF), Naive Bayes (NB) and K-Nearest Neighbours (KNN). The classification process considers factors such as students’ engagement in different activities over time, the scores obtained in these activities, class attendance, and final results. The ultimate goal is to early predict student performance by categorizing them into two groups: those requiring additional support (convocation) and those who achieve well, the initial weeks of the course.