0

I have a confusion matrix as below:

enter image description here

When I use the "caret" package, it uses Fail as the "Positive" class to calculate sensitivity and specificity since it appears first in the row and column. Neither Fail or Pass is more important to me than the other. Still, Intuitively it looks wrong to me as I automatically think of Pass to be the "Positive" class.

What is the industry standard in this case? Should I change my confusion matrix, or keep the confusion matrix as is and change the setting so that Pass is the 'Positive' class?

Or should I leave everything as is, and it's fine to report it this way?

camille
  • 16,432
  • 18
  • 38
  • 60
Roz
  • 195
  • 2
  • 10
  • 1
    You can rearrange it as you want ie lets say your confusion matrix is called `M` you could simply do `M[2:1, 2:1]` – Onyambu Jul 21 '21 at 20:12
  • If that column is a factor, you can change the order of levels, or if it's not you could make it one. Beyond that, it's hard to do more than guess without a [reproducible example](https://stackoverflow.com/q/5963269/5325862) – camille Jul 21 '21 at 20:14

0 Answers0