Timeline for Cross validation on a single model (not model comparison)
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 14, 2020 at 9:19 | comment | added | cdalitz | I have elaborated on this in my answer. Beware however, that is more a suggestion than an answer, because I am not sure about the bias or consistency of the computed values. | |
| Oct 14, 2020 at 9:17 | answer | added | cdalitz | timeline score: 2 | |
| Oct 13, 2020 at 19:09 | answer | added | cbeleites | timeline score: 1 | |
| Oct 13, 2020 at 14:14 | comment | added | Dylan_Gomes | Sure, that would be a statement about a single model. But I do not see jacknife variance discussed in any of the forums/posts about LOO or K-fold CV. It isn't included as the output to any of the common functions. An internet search of the term also doesn't convey a connection to CV. Can you explain how this jacknife variance is connected to CV, and how one might obtain it, as an answer to this question? | |
| Oct 13, 2020 at 13:07 | comment | added | cdalitz | What do you mean with "a statement about a single model"? A confidence interval for the prediction rate? If yes, this can indeed be obtained with leave-one-out: it is the Jackknife variance. | |
| Oct 13, 2020 at 4:34 | history | edited | Dylan_Gomes | CC BY-SA 4.0 |
clarify question
|
| Oct 13, 2020 at 1:50 | history | edited | Dylan_Gomes | CC BY-SA 4.0 |
rephrase question so that it is clearer
|
| Oct 12, 2020 at 22:27 | history | edited | Dylan_Gomes | CC BY-SA 4.0 |
added new knowledge
|
| Oct 12, 2020 at 22:14 | history | asked | Dylan_Gomes | CC BY-SA 4.0 |