1477. Data-driven dynamic predictive model, to assess the probability of students failing online/blended courses
Invited abstract in session TE-12: Learning Analytics, cluster Analytics and Data Science.
Tuesday, 16:15-17:45Room: FENP201
Authors (first author is the speaker)
| 1. | Cecilia Saint-Pierre
|
| Universidad de Chile | |
| 2. | Thomas Peet
|
| EOL, Universidad de Chile | |
| 3. | Galina Deeva
|
| KU Leuven | |
| 4. | Richard Weber
|
| Department of Industrial Engineering, FCFM, University of Chile |
Abstract
Learning management systems (LMS), massively popular in higher education institutions, whether for online or blended courses, and with large amounts of data, have led to the development of computational models to help instructors improve student learning experiences and monitor student progress through various indicators and visualizations, based on learning analytics techniques.
In this study, a sequential classification technique was applied on event log data, collected by the LMS platform implemented at the University of Chile and analyzed by time windows, reflecting student behavior during a course. The presented algorithm generates a pass/fail probability with a predictive model trained by behavioral data from previous versions of the same course and the consequent final student scores. The daily addition of student-generated data to the platform allows the model to create dynamic predictions helping instructors identify underperforming students early in the course.
Keywords
- Data Science
Status: accepted
Back to the list of papers