Grade Prediction Considering Learning Log Relationship

Abstract

Analyzing learning log data from digital platforms helps identify at-risk students and provide personalized academic support. In this study, we aim to improve prediction accuracy by considering the temporal and contextual relationships among learning logs. We introduce a Transformer-based approach that processes sequences of tokenized learning logs. Our experiments show that the proposed method achieves higher prediction accuracy than previous methods. This result highlights the effectiveness of modeling these sequential relationships.