Admin Analytics and New Course Analytics on our Data Access Platform

[carlycurran]

Admin Analytics has been migrated to DAP

We’re excited to announce that Admin Analytics for all customers in all regions has been migrated to our Data Access Platform (DAP), the same source as Canvas Data 2. 

It's important to note that this exclusively a backend change. The data source for Admin Analytics has been transitioned to Instructure’s Data Access Platform. You should see no major differences in your Admin Analytics data, although we did fix a couple of data classification bugs when we moved to this new data source: some mobile page views and participations have been added to the dataset. 

Data Freshness 

Your Admin Analytics data will now have a target latency of 8 hours. In some cases, data will refresh more quickly than this, depending on the Canvas table the data comes from. There is a huge volume of submissions data, for example, so this table often takes the most time to process.  

New Analytics will be migrated to DAP during the month of November

We’re now working on migrating New Analytics (course-level analytics) to use DAP. We will transition customers to this data pipeline in a rolling manner through the month of November. 

Data Freshness 

When the migration is complete, New Analytics will have a target latency of 8 hours, with the exception of Reports, which are CSVs that can be run with near real-time data.

Large Course Support 

As part of the DAP migration, we are also working to support large courses in New Analytics. Currently, there are some restrictions on the number of enrollments and assignments (course weight), which make it impossible to load New Analytics when a course has many students and/or many assignments. This is not the experience we want for teachers of large courses – they often need access to this data the most! 

Moving New Analytics to DAP will be part of the work needed to support large courses. We plan to complete the work for large course support by the end of November. 

[carlycurran]

I need to make a clarification on data freshness of reports for New Course Analytics: Reports all have an target latency of 8 hours. Some may have fresher data than this, but 8 hours will be average. Previously, two reports were 24-hours latent and the other three were near real-time. 

[stimme]

@carlycurran That is a useful note. Would you consider implementing new data refresh timestamps in course-level New Analytics? I know the old timestamps were removed because they could be misleading. I worry that, especially for most instructors (who don't follow all the release notes and blog posts in Instructure Community), the lack of any timestamp is just totally opaque. Instructors may reasonably assume the dataset is live.

[carlycurran]

@stimme This is totally valid point. One issue is that not every data point refreshes at the same time, so we can't say something like: your data was updated at X:XX PM.

In the reports tab, we do say: "Run a report as CSV. Average data latency is about 8 hours." I think you're saying that would be beneficial information for the tool as a whole (i.e., not just noted in that tab). Am I getting that right?

[TrinaAltman]

@carlycurran Just to chime in on this as well, yes there should be something on the tool overall indicating how old the data could be so it is visible no matter which tab they are looking at. Otherwise, I think many or most users would assume it's current. We have that concern with ANY of the tabs where they'd be looking at data, including e.g., Online Attendance.

[stimme]

@carlycurran Yes, that is right. A prominent reminder that "average data latency is about 8 hours" would be helpful on every tab.