Investigation of early detection methods for dropout students by using LMS log data

Abstract

Digital human resource development is an important issue, and e-learning that allows for learning at any time and from any location is well-suited for recurrent education and reskilling. However, although e-learning has the advantage of enabling a large number of students to take courses, it has a high dropout rate. Hence, this study focuses on analyzing e-learning course log data pertaining to basic computer science education. We examine methods to address missing values, and identify sessions that exhibit significant signs of dropout by employing a self-organizing map. Subsequently, based on the analysis findings, we developed a system for the early detection and identification of students displaying dropout signs.