Timeline for How to choose an error metric when evaluating a classifier?
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Aug 13, 2012 at 12:11 | history | edited | Michael R. Chernick | CC BY-SA 3.0 |
added 1 characters in body
|
| Aug 13, 2012 at 11:19 | comment | added | sebp | Comparing the performance of two classifiers on the same dataset is another topic to argue about. Especially, in the case of ROC and AUC there are a couple of methods to compare either ROC curves as a whole or the AUC estimates. Those are essentially statistical tests with the null hypothesis that the ROC/AUC does not differ. Cross-validation vs. bootstrap is another interesting topic, I recently saw a paper (dx.doi.org/10.1016/j.csda.2010.03.004) about that. I guess if you consider all the aspects at once, it can get pretty intimidating. | |
| Aug 13, 2012 at 11:11 | history | edited | user10525 | CC BY-SA 3.0 |
Hyperlink
|
| Aug 13, 2012 at 10:26 | history | answered | Michael R. Chernick | CC BY-SA 3.0 |