The new version is much improved! But I don't think you should lean on this number as hard as you do.

How accurate do you think the information is about who is trans? I went looking through the Gun Violence Archive to see what it was like, and it only took a few minutes to find that they had misgendered someone. Not someone trans. They just had a shooter, who was a (cis, apparently) woman listed as a man.

Given this kind of quality control, if four have been confirmed to have been committed by transgender people that could only reasonably be considered a lower bound. (Well, maybe the lower bound is three, because one of those is in doubt.)

I note also that you go with the 4/3561 number instead of 1 or possibly 2 out of 188 public mass shootings. 0.11% sounds a lot better than 0.53% or 1.06%, doesn't it?

But really, the bottom line is that the data quality is mediocre and the sample size is really small (confidence intervals are huge), so it's absurd to do anything important with these numbers.

The only thing one can do with these numbers is to state clearly and forcefully that there isn't an epidemic. There really isn't. "We're in the midst of an epidemic but the sample size is too small to be able to notice" is not a sensible statement: intrinsic to the idea of epidemic is that it is occurring widely. It's not.

(Mass shootings as a whole reasonably could be termed an epidemic.)