EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2366. Sequence mining algorithm for students failure classification in multiple blended courses scenarios
Invited abstract in session WA-27: Learning Analytics using Mathematical Optimization and XAI, stream Mathematical Optimization for XAI.
Wednesday, 8:30-10:00Room: 047 (building: 208)
Authors (first author is the speaker)
| 1. | Joaquín Roa
|
| University of Chile | |
| 2. | Cecilia Saint-Pierre
|
| Universidad de Chile | |
| 3. | Thomas Peet
|
| EOL, Universidad de Chile |
Abstract
Over the past decade, the use of learning management systems (LMS) for online and blended courses has become widespread in higher education institutions, resulting in a high increase in the available data. The unprecedented volume of data has allowed researchers to develop new and improved algorithms for monitoring the learning process, becoming a useful tool for teachers, given the absence of face-to-face interaction with the students. In our work, we developed a new iteration of a classification method that uses sequence mining from event log data to predict lower-performing students during the realization of a course. Previous versions of this algorithm were based on non-continuous scenarios, using data available for various courses during a single semester. In the present study, we made optimizations to the algorithm, considering a greater diversity in the available training data, fine-tuned using performance metrics that minimize false negatives, and delimited a new, more general methodology, for present and future use cases. We tested our enhanced method in a second-year undergraduate blended course from the Engineering School at Universidad de Chile, with online content, activities, and assessments supported in the institutional LMS. Results show that the enhanced algorithm improves both accuracy and recall, allowing professors to identify the segment of lower-performance students before the course concludes, allowing for proactive measures to be taken.
Keywords
- Analytics and Data Science
- Computer Science/Applications
- Education and Distance Learning
Status: accepted
Back to the list of papers