MODELING AT-RISK LEARNER DETECTION IN ONLINE EDUCATION: DEEP LEARNING FRAMEWORK WITH LSTM AND ATTENTION MECHANISMS

Abstract

Identifying and supporting at-risk students is a major challenge in a digital education environment. This study examines the use of deep learning methods, specifically Long Short-Term Memory (LSTM) neurons with cognitive mechanisms to identify at-risk learners based on their involvement in periodic assessment and engagement with learning components in online learning environments. It also takes into account the importance of the dependencies of temporal factors, thus augmenting accuracy of prediction. The findings highlight the potential of advanced data analytics techniques to improve support strategies for learners on virtual learning platforms, ultimately leading to greater learner success and retention in turn. By using the test results, the study highlights the robustness of the LSTM model in predicting learner's achievement and provides insight into the factors that have the greatest predictive impact. The model performance indicates that the approach of LSTM along with attention mechanism is effective in capturing the periodic behavior of the learner on virtual platforms and the early predictions would be useful to administrators for designing timely intervention and improve retention rates of learners.