[R-sig-ME] predict.glmmTMB when "cloglog" link is used.
r@turner @end|ng |rom @uck|@nd@@c@nz
Sun Mar 22 22:31:53 CET 2020
In an off-list message it was suggested to me that perhaps the value of
p2 should be exponentiated. Indeed p1 and exp(p2) are "reasonably
similar". I really don't understand how the necessity for
exponentiation could arise, however.
When the "probit" link is used to fit the model, predict.glmmTMB() seems
to work just fine.
I have stared a bit at the code for predict.glmmTMB() but the subtleties
are too great for me to be able to figure out what is going on/wrong.
On 22/03/20 9:01 pm, Rolf Turner wrote:
> Please consider the follow examples:
> X <- dget("demoDat.txt")
> fit1 <- glmmTMB(cbind(Dead,Alive) ~ (0+Trt)/Dose +
> fit2 <- glmmTMB(cbind(Dead,Alive) ~ (0+Trt)/Dose +
> p1 <- predict(fit1,type="response")
> p2 <- predict(fit2,type="response")
> The vector p1 appears to have "reasonable" entries i.e. they look (as
> they should) like probabilities:
>>> p1 >  0.019609546 0.194977679 0.745729561 0.972612901 0.994688457
>>  0.999807998 0.999963583 0.022220571 0.120654723 0.833383226
>>  0.994545468 0.999092456 0.999975082 0.106684473 0.261453062
> The entries of p2 do not seem "reasonable". In particular they all less
> than or equal to zero:
>>  -3.244681e+00 -1.597702e+00 -3.235952e-01 -6.683303e-04
>>  -5.930060e-33 -2.134861e-103 0.000000e+00 -3.360384e+00
>>  -3.222809e-01 -1.400287e-02 -6.947338e-07 -3.769347e-21
> Am I misunderstanding something, or is there a bug in predict.glmmTMB?
> The data set in question is attached as "demoDat.txt".
> Thanks for any enlightenment.
More information about the R-sig-mixed-models