Timeline for How to assess statistical significance of the accuracy of a classifier?
Current License: CC BY-SA 3.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 13 at 11:48 | comment | added | Ruben van Bergen | Classifiers are sometimes used to assess whether a certain variable (or set of variables) contains information about another variable. For instance, in the context of analysis of neuroimaging data, a classifier might be used to determine whether the brain signals from a certain cortical region contain information about some aspect of a visual stimulus that the person was exposed to, which might suggest that this cortical region is involved in processing such information. In this context, a significance test against chance-level performance is very relevant. | |
| Dec 12, 2013 at 18:05 | comment | added | ely | Classifiers that just barely beat random guessing can be extremely useful in some situations. Thus, having some test that quantifies confidence in a classifier being better than chance is also useful. | |
| Oct 26, 2011 at 8:55 | comment | added | Simone | Yes, it is an highly problematic measure. I agree with you. | |
| Oct 26, 2011 at 1:27 | comment | added | Frank Harrell | Yes; a highly problematic measure. | |
| Oct 26, 2011 at 0:15 | comment | added | Simone | About of the proportion of classified correctly, do you mean the standard classification accuracy? thanks | |
| Oct 22, 2011 at 16:16 | history | answered | Frank Harrell | CC BY-SA 3.0 |