Timeline for What is the intuition behind the idea that for linear regression, the number of observations should exceed the number of parameters?
Current License: CC BY-SA 4.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 13, 2023 at 17:18 | history | edited | questionto42 | CC BY-SA 4.0 |
wording
|
| Dec 13, 2023 at 16:03 | history | edited | questionto42 | CC BY-SA 4.0 |
meta link
|
| Dec 13, 2023 at 15:54 | history | edited | questionto42 | CC BY-SA 4.0 |
more on the link between MLE and OLS, the linear and the logistic regression, see [If someone asks "What is the intuition behind the idea that for linear regression, ...", does the answer have to be only about the linear regression?](https://stats.meta.stackexchange.com/q/6615/287262)
|
| Dec 13, 2023 at 15:43 | history | undeleted | whuber♦ | ||
| Dec 13, 2023 at 2:15 | history | deleted | kjetil b halvorsen♦ | via Vote | |
| Dec 11, 2023 at 11:11 | comment | added | questionto42 | @whuber I think it does, but I am not sure, so I understand the remark and the downvotes. It is about the intuition. If you add too many parameters, you overfit a model of a logistic regression that is asymptotic and can decide between 0 and 1. If you add higher dimensions of parameters or add new parameters, this makes it easier for the model to hit 100 % of the observations, ending up with less accuracy for the predictions. Dimensionality reduction can avoid that a bit, or regularization. The question asks for an intuition so that the answer can go beyond the linear regression. | |
| Dec 9, 2023 at 23:33 | comment | added | whuber♦ | This doesn't seem to have much if anything to do with the question. Did you perhaps misread "linear regression" in the title as "logistic regression"? | |
| Dec 9, 2023 at 22:39 | history | answered | questionto42 | CC BY-SA 4.0 |