Big data and ensemble learning for effective student orientation in Morocco

Abstract

Guiding high school students toward suitable educational paths is a complex challenge, particularly influenced by academic performance. In Morocco, first-year high school students in the scientific branch face a crucial decision when selecting between science mathematics (SM), physics (SF), and Science of Life and Earth (SVT) paths. This decision is critical as it can significantly impact their future academic and professional success. To address the issue of suboptimal student orientation, this study proposes an automated, personalized approach leveraging big data technology. By employing ensemble learning techniques, including random forest and neural network models, we developed a classification system to predict students’ optimal paths based on their academic performance. Our models achieved exceptional performance, with precision, accuracy, recall, and F-measure scores of approximately 98.59%, 98.60%, 98.60%, and 98.58%, respectively. This research demonstrates the potential of our approach to enhance educational support and decision-making, ultimately improving student outcomes and reducing dropout rates caused by wrong orientation.