[R] ZIB models

[Sundar Dorai-Raj]

Rizkalla, Carol wrote:

 > I attempted to contact Drew Tyre, but the email I have for him is no
 > longer in service.
 >
 > Hopefully someone can help.
 >
 >
 >
 > I'm using obs.error in R to model turtle occupancy in wetlands.
 >

What is "obs.error in R"? I get the following:

 > help.search("obs.error")
No help files found with alias or concept or title matching
'obs.error' using regular expression matching.

 > I have 4 species and 20 possible patch and landscape variables, which
 > I've been testing in smaller groups.
 >
 >
 >
 >
 >>zib.out<-obs.error(y=painted,m=numvis,bp=zvars,pcovar=7)
 >
 >
 >
 >
 > I get the following error message, with all species, all variable
 > groups, standardized and unstandardized data...
 >
 >
 >
 > Error in optim(par = rnorm(pcovar + qcovar), fn = obs.error.LL, method =
 > "L-BFGS-B",  :
 >
 >         L-BFGS-B needs finite values of fn
 >
 >
 >
 > I have a large sample size, so we're not sure that this is related to a
 > convergence problem, but we can't figure out what this error means.
 >
 >

Hi Carol,

(Please see the posting guide)

As the message says, your fn ("obs.error.LL") returned a non-finite 
value which is not allowed. Without seeing your code, it will be 
impossible to say where this happened. Either turn the trace on or step 
through your function to see where infinite values may creep in.

BTW, if you are fitting a ZIB model (zero-inflated binomial) you may be 
interested to know that Jim Lindsey's gnlm package does this through his 
fmr function (I think; I haven't actually done this kind of modelling in 
a while). The package can be downloaded from

http://alpha.luc.ac.be/~jlindsey/rcode.html

Be sure to get the rmutils package while you're there.

--sundar