Page Views with no page ID

[AndrewYardy]

I'm wanting to identify if select students have viewed a named page in a canvas site.  

I've tried this from multiple angles (analytics and users) but can see a call that would link student id with page ID and date time stamp.

url = 'https://canvas.newcastle.edu.au/api/v1/courses/' xxxxx + '/analytics/activity'

This call does not return a page ID as shown below the id is None.

{'date': '2023-06-08', 'id': None, 'participations': 0, 'views': 1}

Can anyone point me in the right direction please?

[James]

@AndrewYardy 

This information is contained in the Access Report. That report is not available through the API and must be done one student at a time.

I wrote script that allows you to obtain the access report for the entire course. It has to be ran within the browser since that's where the access report is available, but you don't have to check each student individually.

I also use Canvas Live Events to track when users view content. It isn't guaranteed to work, but it catches most things. There is overhead involved in setting that up and downloading the results to a database, but it makes querying quicker. It cannot be used to go back and find results before it was installed, but it might be helpful moving forward.

[AndrewYardy]

 Hi James,

Many thanks.  The access report is helpful but doesn't provide the exact timestamp we need to make a call on student engagement. 

I'll checkout the Live Events data. 

Kind regards,
Andrew

[James]

@AndrewYardy 

Before you rule out the access report completely, be aware that there is information downloaded by Canvas that it does not show on the web. If you look at the JSON that is delivered, it has the first time they accessed the report as well as the last time. That JSON object also contains the full timestamp for the last viewed -- not the unprecise "1 hour ago" or "10 hours ago" that it shows on the web.

You can access the JSON data by appending .json to the end of the URL. That is, instead of the path ending in usage, make it end in usage.json. Depending on your browser, you may need to copy/paste the JSON into a JSON to CSV converter to make it into a table to understand it.

My script puts all that information, including both timestamps, and exports it as a CSV file. You can then set up pivot tables or filter off assignment names.

The two main downsides of the Access Report are that it is only available through Canvas (no API) and that does not give all of the times that someone looked at something (unless that number is 2 or fewer). If someone looks at something 30 times, you still only know the first and last times. Often, knowing the first time is well after when they should have looked at it is enough for what I'm working with.

You can potentially view pages (as an admin), but then you're hunting for a needle in a haystack, especially if it happened a while ago. It can be beneficial, though. I had a student who was logged in multiple places at the same time, submitting work from different IPs (in different countries) within seconds of each other, and turning in work that was like nothing they had done before (and didn't really answer the discussion prompt). The access report doesn't have IP addresses, but that's not always reliable. I have three ISPs at home so that I can get some semblance of high speed internet and my IP could change between requests.

The live events is only good moving forward. You cannot go back and look at things that happened before you set it up. It is also a best-effort. If there is network issues or the Canvas servers are too busy, you may not receive the data.

I originally set it up so I could provide data to my statistics class so they could try and figure out what behaviors are predictive of success in the class. I give them 30+ variables including how many times a student looked at the grades page or used a mobile device. They have 60 minutes to look through the data and prepare a slideshow. The last 10 minutes are the group presentations to the class.

While that was the original intent, I can see how often users are active in the system (more than just logging in). I've used it on the student side and the instructor side.

The live events is kind of like the page view logs except that it can happen in near-real time and you can filter out some junk to only receive what you want to receive. It includes the IP address used in the access. It can really big, though. We're a small school and the AWS costs are typically under $2 a month for the data I gather. It's taking about 4-5 GB for a month's worth of data. February 2024's data was 2.7 million records and 5 GB. That's just mostly access-related events being sent.