Closed corneliusroemer closed 1 year ago
The latest updates on your projects. Learn more about Vercel for Git ↗︎
Name | Status | Preview | Updated (UTC) |
---|---|---|---|
flu-frequencies | ✅ Ready (Inspect) | Visit Preview | Jun 28, 2023 11:20am |
There's something funny going on with nested circles in the preview:
fixed
Lollipop radius hence signifies statistical weight as opposed to counts of the variant in question, which seems more intuitive to me
As a result, there's less of a trend of big lollipops at the top and small ones at the bottom.
In particular, lollipops of non-observed variants were not shown at all - despite the fact that we do know something: that out of N observations none were of the variant in question.
Also, this PR removes the pseudocount of 1 that is added to radius - I'm not sure why this is necessary: if there is no observation, there is no lollipop anyways, so the radius will be >=1 anyways.
Before:
After: