Specificity for multiple-class classification
1 view (last 30 days)
have read the paper "A systematic analysis of performance measures for classification tasks" in which different statistics for binary classification as well as their extension to the multi-class setting isexplained.
They introduced micro- and macro-averaging for precision and recall. Unfortunately, they do not spend any word about specificity in the multi-class case.
Can I just use micro- and macro-averaging for specifcity in the same way as for recall and precision?