Moodle Analytics Models
This walkthrough video provides an overview of main analytics models available in Moodle LMS (version 4.0.1):
Upcoming activities due
Courses at risk of not starting
Students who have not accessed the course recently
Students who have not accessed the course yet
Students at risk of dropping out
Key takeaways from this: Moodle Analytics capabilities are well designed and implemented with the following advantages
Explainability (why a course not analysable, how prediction made with indicator values)
Actionable (send message, “Go-to-activity” link)
Ability to provide feedback of the generated predictions
Tracking model’s performance / effectiveness
Model clarity and customizable
Complete control of model training and deployment
However, it appears that the model training and deploying may consume a lot of server resources, and the model accuracy is not as high as expected (due to the algorithm used, and/or the amount and quality of data available). As a result, these capabilities are often overlooked and disabled altogether by system administrators.