I have used random forest classifies to classify 4 groups (or class). For each group (target class) the AUC has been generated. Using "Test&Score" (from orange) provides AUC for each target class (Lets say target classes are A, B, C and D) and also average over classes.
How's AUC been generated for a given target class while there are 3 more groups to compare with?
Does the AUC for target class ( let's say "A") is calculated considering other 3 ("B", "C", and "D") groups as one group (one negative group)? Because AUC is calculated for normalized data (0 to 1), assuming 0 as negative and 1 as positive cases.