problems estimating INT_Ab in dynamic 2-sps models

posts related to the RPresence library, which may not be of general interest to users of 'classic' PRESENCE.

problems estimating INT_Ab in dynamic 2-sps models

Postby amurphy » Mon May 20, 2019 4:50 pm

PRESENCE version: 12.29
RPresence: 2_12.29
Interface: Mac

I'm using the dynamic two species model in R and estimates I get for gamma and epsilon AB/Ab are identical to each other despite including the INT_Ab term, even when using the code available in the help file.

Help file code:

## fit some multi-season, 2-species models (using different data)
mod0<-occMod(model=list(psi=psi~SP+INT,
gamma =gamma~SP+INT_Ab+INT_Ba+INT_B_a+INT_Baa, epsilon=epsilon~SP+INT_Ab+INT_Ba+INT_B_a+INT_Baa,
p =p~SP+INT_o+INT_d+SP:INT_o),
data=bearcoy,type="do.2sp.1")

Any ideas?
amurphy
 
Posts: 5
Joined: Wed Feb 13, 2013 2:51 pm

Re: problems estimating INT_Ab in dynamic 2-sps models

Postby jhines » Mon May 20, 2019 8:48 pm

The dynamic 2sp model has lots of possible parameters and is very "data-hungry". Did you see any warning messages about convergence or the var-cov matrix? Do the variances of the beta parameters look reasonable?
Without seeing your data, my guess is that you don't have enough data to estimate all of the parameters. (I'd be happy to take a look if you send me a copy, though.) Did you try simpler models before that one? I usually start with simple models and move to more complicated ones to see how much complexity the data/model can handle.

Jim
jhines
 
Posts: 451
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: problems estimating INT_Ab in dynamic 2-sps models

Postby amurphy » Mon May 20, 2019 9:00 pm

So I do get an error when I run the full model but this error occurs every time I use INT_Ab, even if I don't have anything else on any other parameter.

For example, I get identical estimates for gammaAB and gammaAb, even when the model is psi~SP,gamma~SP+INT_Ab,epsilon~SP,p~SP.

Thank you for offering to look it over; I'll send you an email.
amurphy
 
Posts: 5
Joined: Wed Feb 13, 2013 2:51 pm

Re: problems estimating INT_Ab in dynamic 2-sps models

Postby jhines » Tue May 21, 2019 11:42 am

Thanks for sending the code. I've fixed the bug in RPresence and uploaded the new version to the web-server. Please re-install the package and try again.

Note: The USGS server does not update as often as I do, so the best way to install the latest version is:

install.packages('https://www.mbr-pwrc.usgs.gov/software/bin/RPresence_12.31.zip',repo=NULL)

Jim
jhines
 
Posts: 451
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: problems estimating INT_Ab in dynamic 2-sps models

Postby jhines » Tue May 21, 2019 12:36 pm

Sorry for the error in the last post... the RPresence version number should have a '2.' before the '12.31'. So, it's:

install.packages('https://www.mbr-pwrc.usgs.gov/software/bin/RPresence_2.12.31.zip',repo=NULL).

Note: You can omit the version number to get the latest version.

Note2: If the www.mbr-pwrc.usgs.gov server is down, you can get the software from:

www.bushcreekfarm.com/software

Jim
jhines
 
Posts: 451
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA


Return to RPresence

Who is online

Users browsing this forum: No registered users and 1 guest