Template talk:Diagnostic testing diagram

Why is the diagram Transposed?

Same question, why was the matrix inverted? It is disturbing having a standard changing. Edit: I found out about the discussion change (not used to wikipedia sorry) I think a mention about the standard having been change would be nice JackRed6 (talk) 10:34, 4 June 2021 (UTC)[reply]

The decision to transpose the diagram is discussed here: Talk:Confusion_matrix#Confusion_Matrix_is_Transposed_compared_to_standard_practice. Odulon (talk) 14:05, 19 June 2021 (UTC)[reply]

I see that some reference equate FP = overestimation and FN = underestimation but some reference does the opposite.

This table has FP = over but the blue column on the right of [Confusion_matrix] has FP = under.

This is asking for troubles and I suggest removing the over and under estimation terminology.

Simple Symbol (talk) 22:47, 22 June 2021 (UTC)[reply]

This table was flagged as it contained some HTML {{sqrt}}s. I've replaced these with latex. To get the formatting to look nice I've had to use scriptstyle, tfrac, and textsf, but this is a matter of preference so I'm happy if this is modified. aricooperdavis (talk) 13:52, 16 July 2022 (UTC)[reply]

I think location of Negative Predicted Value(NPV) box and False Omission rate box(FOR) box are inconsistent

There is a great way of understanding this table, starting by 2×2 confusion matrix which contains absolute frequencies TP FN ; FP TN;

Normalizing each row by its sum, will create a 2x2 matrix that is just right-side of original matrix TPR FNR ; FPR TNR ;

Normalizing each column by its sum, will create a 2x2 matrix : PPV FOR ; FDR NPV;

But in the table, below the original 2x2 confusion matrix we have following matrix : PPV NPV; FDR FOR;

I don't understand this choice for location of NPV and FOR, ut just swapping them makes everything great

TheLoggicalOne (talk) 20:16, 23 September 2025 (UTC)[reply]