Moodle now implements learning analytics using machine learning.These models can go beyond simple descriptive analytics to provide predictions of learner success, and ultimately diagnosis and prescriptions (advisements) to learners and teachers.
In Moodle 3.4, the system ships with two built-in models:
- Students at risk of dropping out, and
- No teaching activity
This initial release has some serious limitations:
- Models included have be ‘trained’ on a site with previously completed courses, ideally using the Moodle course completion feature. Given that the current models cannot make any predictions on a site until this is done, if you are not already using course completion you would have to start doing this now to build up some data.
- The current prediction models require that courses have fixed start and end dates. So it won’t work with rolling enrolment type courses. If you run your courses on this basis, it means this initial analytics is of no use to you.
- Models and predictions are only visible to teachers and administrators at present (which I think makes sense).
We haven’t had time to look at how this works in-depth in the real world yet … for example what happens if you have a course that uses course completion and set start and end dates, but you then reset it for the next intake? Is the previous data lost form the analytics store? Is it combined?
After understanding the requirements and limitations, if you think Analytics is something of use to you, have a read of the documentation page that outlines the basic operation:
We will be implementing this in our MoodleBites online training courses site in 2018, across a huge range of 8 week courses, to see how much useful data it really provides in our context. We will report back here to let you know our thoughts and share our experiences … 🙂