Timeline for Can I merge multiple linear regressions into one regression?
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Oct 6, 2022 at 12:59 | answer | added | Björn | timeline score: 0 | |
| Aug 6, 2022 at 13:04 | vote | accept | asmgx | ||
| Aug 6, 2022 at 12:53 | answer | added | zfy | timeline score: 1 | |
| Oct 3, 2021 at 21:52 | comment | added | Tim | Were the samples divided into groups randomly or there was some criteria for that? | |
| Oct 3, 2021 at 21:26 | comment | added | Christian Hennig | Although one can compute a single regression for all data points, if you include model assumptions such as i.i.d. normal errors, the model for all points combined can't be "correct" if the four individual models are correct (unless in reality they are all equal), because the combined model then can't be a single linear regression but would be a mixture of four of them, which is different. Nonetheless of course you can compute a regression that produces a "compromise" of them. | |
| Oct 3, 2021 at 21:08 | answer | added | Sextus Empiricus | timeline score: 4 | |
| Oct 3, 2021 at 13:13 | comment | added | asmgx | will it make a difference? they all have exact same number of records | |
| Oct 3, 2021 at 8:22 | comment | added | statsplease | Are all four models using the same dependent variable? Was the split of the data (into the four coloured groups) random or based on an observable variable? | |
| Oct 3, 2021 at 3:07 | history | asked | asmgx | CC BY-SA 4.0 |