[R-sig-ME] glmmadmb function

[Ben Bolker]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

  [taking the liberty of cc'ing to r-sig-mixed-models]

  How badly/why do you need to use glmmadmb?  So far I don't see
anything in the model that requires it (you've got binary data, so
zero-inflation won't be relevant ...)  I'd recommend glmer (which
takes about a minute even with all of the fastest settings I know
about, so glmmadmb is likely to be much slower).  I might even
recommend gamm4, which would allow you to fit smooth responses to
ResDist ...

  There are some updated binaries of glmmadmb that might fix the
problem -- I haven't gotten around to putting them into the package
yet, but if you say why you need glmmadmb I can post some information
about how to update the binaries on your system ...

L <- load("CoyoteAllData.RData")
library("ggplot2"); theme_set(theme_bw())
library("mgcv")
ggplot(all.data,aes(x=ResDist,y=Used,colour=Sex,shape=Status,linetype=Status))+
    geom_point(position=position_jitter(height=0.05),alpha=0.2)+
        facet_wrap(~LandCover)+
        geom_smooth(method="gam",family=binomial)

nrow(all.data)
library("lme4")
system.time(g1 <- glmer(Used~Sex+Status+LandCover+ResDist+(1 |ID),
            data=all.data, family=binomial(link="logit"),
            control=glmerControl(optimizer=nloptwrap)))

table(all.data$Used)

On 14-11-20 04:52 PM, Sharon Poessel wrote:
> Dr. Bolker,
> 
> I'm attempting to use the glmmadmb function to run the following
> model:
> 
> globmod5 = glmmadmb(Used~Sex+Status+LandCover+ResDist+(1 | 
> ID),data=all.data, family="binomial", link="logit")
> 
> I'm attaching the dataset.  "Used" is a 0,1 response variable,
> "ResDist" is a continuous variable, and all other variables are
> categorical.  Each row represents a location of a coyote (1 =
> actual locations, 0 = random locations).
> 
> I'm using glmmADMB version 0.8.0, R version 3.1.0, and Windows 7
> 64-bit OS.
> 
> I'm getting the following error message when I run the model, which
> is similar to what others have received:
> 
> Error in glmmadmb(Used ~ Sex + Status + LandCover + ResDist + (1 |
> ID),  : The function maximizer failed (couldn't find STD file)
> Troubleshooting steps include (1) run with 'save.dir' set and
> inspect output files; (2) change run parameters: see
> '?admbControl' In addition: Warning message: running command
> 'C:\Windows\system32\cmd.exe /c 
> "C:/.../R/win-library/3.1/glmmADMB/bin/windows64/glmmadmb.exe"
> -maxfn 500 -maxph 5' had status 1
> 
> As previously suggested, I also tried
> admb.opts=admbControl(shess=FALSE,noinit=FALSE) and
> save.dir="tmp", but I get the same error messages.  Previous posts
> have indicated that older versions of the package may not give this
> message.  Is there anything else I can try using the current,
> newest version of the package?
> 
> Also, if I run this same model in glmer, I get no error messages,
> so I don't think it's the structure of my model.
> 
> Thank you for your help.
> 
> Sharon Poessel
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iQEcBAEBAgAGBQJUbrFGAAoJEOCV5YRblxUHrA4H/iJtibQCxd0rjirwJaYWzckd
fHdG21VVvtI9d/D4mmkDaGJmX6529aru9vvGVRAI7JreEIwrqNBD9+mXycAwpSm2
+QFCP5TzAEgHDbidueXoiY1swDug9TJ9sySGnLbNK/v85M0wd3ddrj6uTJjQsRK9
VTz4TaZml8XNTDLSx21LRUvLx73y4ZGIoc+2gOOWI+jjrBCIyV6cC5aBGyT5+Qz3
PjGv1FefPx4pk1hD1QrX/x+U9VepjQGZo0NEOJt6Uk/q8+zQDBIRCKu6aw4NB6te
ODEEwA0latJheYOTcMr0TFL4MRs3MOrp07h/Tk/2MgcLdOK93BWMjGWJfa4mCKw=
=m35N
-----END PGP SIGNATURE-----