- 4-Week Groupings
- 13-Week Groupings
- “Waves” [breakpoints at ~Jun 20, ~Oct 20, Apr 21, Jul 21… .so far]
So far, I’ve looked at the whole period 2020-2021 (well, through 10/2/2021). But now I want to get at how trends may have changed over time.
Check out this graph for reference:
Doing it weekly is to granular, but I am trying to figure out the best way to aggregate it.
Should I try regular time periods or use the “waves” that we’re seeing in the data?
I’m voting “waves” more to capture seasonality with the clear “breaks” shown above that may not be properly captured by a 13 week groupings.
IMO, 4 week groups might still be too granular, but would be my second choice as it should still capture the seasonality trends adequately.
Is there a way to show waves by geographic area? We hear that the waves move through different geographies at different times. Is that doable?
My plan was to look at the 11 states plus NYC that I covered in this post:
One issue, of course, is that huge geographical swathes just don’t have many people.
New York (this excludes NYC)
It’s basically the most populous states, jumping over NC because they suck at reporting their data.