For imbalanced classification (say 85:15), what is good value of F1 score? An answer https://stats.stackexchange.com/a/217343/285091 says "Experiments indicate that the sweet spot here is around 0.76, where the F-measure is 0.87." Is there any publication indicating good F1 score for imbalanced classification?
Asked
Active
Viewed 185 times
0
-
There is not concept of a "good or bad" score without context of where we are applying this. In certain settings maybe 60% is the state of the art, in other settings 95% might be unacceptably low. I answered a similar question (for AUC-ROC) here: https://stats.stackexchange.com/questions/483185 *exactly* the same logic applies. – usεr11852 Sep 12 '20 at 23:59
-
@usεr11852 Similar to your reference on ROC in your mentioned answer, I am looking for a reference for F1 score. – ewr3243 Sep 13 '20 at 00:49
-
please note that I specifically say: "*These values are by no means set-to-stone and they are given without any context.*" the author of the passage himself says: "*there is no “magic” number, only general guidelines*". I have seen no publication make any such claims and even if I saw them I would seriously question their applicability. Do literature review for your particular problem and/or build a base-line model. It will *really* help you in contrast to some paper that puts numbers out that might be irrelevant for your task or used as a stick to beat your project team. – usεr11852 Sep 13 '20 at 00:58