A Study on Real-Time Prediction of Course Dropout Students Using LMS Logs

• Published in: 2024 16th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) pp. 224 - 230
• Main Authors: Ohkawauchi, Takaaki, Tanaka, Eriko
• Format: Conference Proceeding
• Language: English
• Published: IEEE 06.07.2024
• Subjects: Accuracy; dropout; Informatics; Input variables; Learning management systems; log data; Machine learning; Measurement; prediction; Predictive models; Prevention and mitigation; Real-time systems; Time-frequency analysis
• DOI: 10.1109/IIAI-AAI63651.2024.00051

In recent years, there has been a rise in attempts to predict students' credit acquisition using machine learning. However, many of these attempts have been evaluated through cross-validation within the same academic year and course after the semester ends and results are available. However, there is a practical need for methods to predict real-time credit acquisition and AI models that can be universally applied to different students and subjects. In this study, we developed a system to identify at-risk students in real time by leveraging constantly updated learning management system (LMS) logs. As a result, we confirmed that we can make continuous and increasingly accurate predictions for different students as the new academic year progresses.