Order of outputs of ConfusionMatrix

[ashleve]

Hi,
is there any explanation on order of output generated by ConfusionMatrix?

Here is the example of binary classification from the docs:

>>> from torchmetrics import ConfusionMatrix
>>> target = torch.tensor([1, 1, 0, 0])
>>> preds = torch.tensor([0, 1, 0, 0])
>>> confmat = ConfusionMatrix(num_classes=2)
>>> confmat(preds, target)
tensor([[2., 0.],
        [1., 1.]])

Judging by the output thensor, I assume the order in matrix is:

[ True Negative, False Negative ]
[ False Positive, True Positive ]

Did I get it right?
Why is the order like that exactly?
For example, on wikipedia it looks like this:

I suggest to add some explanation to the docs about order and interpretation of the outputs, becuase currently this is confusing.

[justusschock]

Hi this is actually the same as wikipedia gives just not with positive and negative but with an arbitrary number of classes. I.e. You have predicted class on the x axis and actual class on the y axis. Both start with class zero and the diagonal is always the number of correct predictions for the corresponding class