Timeline for Abnormally large confidence interval with binomial gam when p->1 at max of predictor range
Current License: CC BY-SA 4.0
6 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 16 at 9:22 | vote | accept | Lars M. Korslund | ||
| Apr 9 at 20:59 | answer | added | Gavin Simpson | timeline score: 5 | |
| Apr 9 at 20:42 | comment | added | Gavin Simpson |
@ChristianHennig Wrong gam; this is with mgcv:gam. Here k is the required basis dimension. For a univariate smooth the actual basis used will be k-1 with the default penalty as a sum-to-zero constraint is applied to the smooth.
|
|
| Apr 9 at 19:34 | comment | added | Christian Hennig |
The gam help says, as far as I can see, that s has parameters x, df, spar, but not k. So I wonder what the k is doing here. In any case I suspect the problem is with the smoother. Have you tried lo instead of s? Also parameters of s may make a difference, but as I wrote, I don't see k there.
|
|
| S Apr 9 at 18:58 | review | First questions | |||
| Apr 9 at 19:16 | |||||
| S Apr 9 at 18:58 | history | asked | Lars M. Korslund | CC BY-SA 4.0 |