Cloud-based STEM Student academic success prediction Web application.
Abstract: Student success can be significantly improved through the early detection of struggling students and the taking of preventive measures. Studies have shown that Machine Learning (ML) has the potential to help address challenges like academic underachievement, dropout rates, and graduation delays. Nevertheless, there needs to be more research on deploying these ML models for educators. The purpose of this paper is to address this issue by presenting a comprehensive approach to deploying ML models using cloud computing technologies. The paper is divided into two modules. The first module compares the test accuracy performance of five well-known ML ensemble models. The ensemble models used 1044 student datasets to classify student academic success into two categories- successful and not-successful. The results of the comparative analysis indicate that Random Forest provides the highest test accuracy of 73.20%. The second module deployed the Random Forest model to the streamlit cloud.