Timeline for How to select the best fit without over-fitting data? Modelling a bimodal distribution with N normal functions, etc
Current License: CC BY-SA 3.0
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 4, 2015 at 7:11 | vote | accept | MurphysLab | ||
| S Apr 3, 2015 at 1:08 | history | bounty ended | Glen_b | ||
| S Apr 3, 2015 at 1:08 | history | notice removed | Glen_b | ||
| Mar 27, 2015 at 10:08 | comment | added | Aleksandr Blekh |
You might want to check this answer (in case you will decide to go the R route). Some model selection criteria are mentioned in this answer. Finally, you may want to consider ensemble methods, which I briefly covered in this answer, which also contains link to Python-focused information. You can find more details on model selection and averaging in this answer.
|
|
| Mar 27, 2015 at 8:53 | answer | added | Chris Novak | timeline score: 7 | |
| Mar 27, 2015 at 4:17 | history | tweeted | twitter.com/#!/StackStats/status/581308995702550528 | ||
| S Mar 27, 2015 at 1:55 | history | bounty started | Glen_b | ||
| S Mar 27, 2015 at 1:55 | history | notice added | Glen_b | Draw attention | |
| Mar 24, 2015 at 18:09 | comment | added | MurphysLab | It's been suggested that I use the reduced chi squared X^2/(N-n-1) where N is the number of data points and n is the number of fitted parameters. However the small pentalty (+/-3) relative to the number of data points (91) doesn't intuitively seem like a particularly steep penalty for adding another Gaussian. | |
| Mar 22, 2015 at 23:20 | review | First posts | |||
| Mar 23, 2015 at 0:52 | |||||
| Mar 22, 2015 at 23:18 | history | asked | MurphysLab | CC BY-SA 3.0 |