Thanks so much for your input!
@a1222679 wrote:
- Course activity percentile (i.e. how the student compares to their peers within each course based on count of active days and using weblog counts as a tiebreaker when the number of active days is the same)
- % of enrollments to have accessed the course in the past week (we’re not too worried if a student hasn’t logged in to a course if none/few of their peers have either
I really like how you’re leveraging the weblogs data here. Initially, we were quite disappointed because we could only retrieve the last 30 days of weblogs data to store elsewhere, which caused a noticeable data gap in 2023. Our original plan was to use historical weblogs/requests data and backtest it in various ways, but I was always concerned about the size of the data and the many variables that came with this approach.
Following that, I considered using a technique somewhat similar to what you’ve described here—comparing students to their peers within each course—but I haven’t made much progress on that idea yet. Seeing your approach here is very helpful and I'll definitely be looking more into this in the near future. Based on other threats and posts on this site, using weblogs to compare peers on a class level appears to be the best way to use the data.
-
Our university has had previous attempts at measuring and improving student-retention that mostly involved SIS data. We might end up leveraging some SIS data alongside the Canvas Data information that we pull, but I wanted to lean in pretty heavily to what's in canvas this go-around. I really like your approach here.
Once again, thanks for your input!