Talk:Loss functions for classification
Add topic | This article is rated C-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
Range of f
[edit]A bit confusing is that in the intro, it is stated that f: Xset --> Yset, where Yset = {-1; 1}, while later on, also 0 is in the range of f and I suspect (I am not an expert) that in general f can take real values. Maybe some expert can clarify this in the text?
Moreover, further down, in Eq. (2), another function f appears - this time of a variable eta (instead of xvec) with a different meaning (auxiliary function for the loss function instead of the classifier function) and a different domain than Xset. I propose to replace this (or the f of the classifier) by another letter like g.
Could somebody add a discussion how exactly a scoring rule differs from what is written here?
[edit]Maybe the focus is different, but please add some insights Biggerj1 (talk) 23:15, 26 January 2024 (UTC)
- Talk:Scoring rule#Discussion on possible merging of this page, and might be the best place to continue the merge discussion. Klbrain (talk) 19:02, 14 June 2024 (UTC)
- No merge was the outcome over there. Klbrain (talk) 10:51, 10 November 2024 (UTC)
Strange wording of the first sentence
[edit]loss functions for classification are computationally feasible loss functions representing the price paid for inaccuracy of predictions in classification problems
I thought price is paid for something beneficial, i.e. accuracy. Jack who built the house (talk) 12:45, 21 June 2026 (UTC)