Abstract - Educational data mining has gained popularity due to its ability to provide useful knowledge hidden in data of students’ records for better educational decision-making support. During the last years, a variety of methods have been applied to develop accurate models to monitor students’ behavior and performance, while most of these studies examine the efficiency of supervised classification methods. In this work, we propose a new ensemble-based semi-supervised method for the prognosis of students’ performance in the final examinations at the end of academic year. Our experimental results reveal that our proposed method is proved to be effective and practical for early student progress prediction as compared to some existing semi-supervised learning methods.