multiclass_auc() computes the one-vs-rest ROC AUC of every class, together
with the macro and micro averages.
Usage
multiclass_auc(truth, prob_predictions)
Arguments
- truth
A vector with the observed classes.
- prob_predictions
A tibble of predicted class probabilities as returned by
predict(type = "prob"), i.e. one .pred_<class> column per class.
Value
A named numeric vector with one AUC per class, plus macro and micro.
Details
Each class is scored against all remaining classes pooled together. macro is
the unweighted mean of the per-class AUCs and micro is the AUC obtained by
pooling every one-vs-rest decision into a single binary problem.
