I just finished this article regarding massive log data and it’s use in predicting student outcomes in online courses. This comprehensive study pulled LMS log data (Moodle) from all courses during an entire academic year and clustered students into groups based upon predictive features. Not surprisingly, the study found that the most successful students logged in earlier, viewed external resources and completed quizzes earlier than their lower performing classmates.
As an experiment, I posted a link to the article into ChatGPT and asked for a summary. The summary also provided me with the following table. Although the article did not specifically mention interventions, I did find this information useful and posted below.
| LMS Data | Why it Matters | Action |
|---|---|---|
| Low logins in first 1–2 weeks | Early inactivity predicts poor performance | Send a personalized check-in message asking if they need help getting started |
| Few clicks on readings/resources | Indicates lack of engagement or confusion | Suggest specific resources, provide study tips, or remind them why resources matter |
| Late or missing assignments | Strong signal of time-management struggles | Offer deadline reminders, recommend planning tools, or invite them to office hours |
| Active in course but not submitting work | May be unsure about expectations | Clarify instructions, provide an example submission, or encourage questions |
| High engagement (frequent logins, many resource views) | Predicts strong performance | Provide enrichment opportunities (bonus materials, leadership roles in discussions) |
| Drop in activity after initial weeks | Risk of burnout or external distractions | Send supportive message: “I noticed your activity dropped — anything I can do to help?” |
| No activity checkpoints (quizzes, reflections) early on | Hard to detect at-risk students | Add small, low-stakes tasks early to generate engagement data |
MOODLE RESOURCES AND PLUG-INS
To take this experiment to the next level, I decide to find some plug-ins to support the above data. The article mentioned an Early Warning System (EWS) plug-in from Moodle. I was not able to find this particular plug-in, but I did find these:
- Student Tracker: Lists users who have not logged into Moodle for configurable number of days
- Students at risk of dropping out: Built-in analytic model provided by Moodle to identify students at risk for dropping out by analyzing several data points including views, submissions, feedback and social interactions)
- MoodleMoot Session: Using Moodle Analytics to predict student success