[R-sig-ME] spatial covariance structure in glmmTMB

[Alice Domalik]

In the end, I was able to run the following model: 

m1<-glmmTMB(count~waterdepth + temperature + chl.conc + (1|individual) + gau(coord + 0| group), family=list(family="truncated_nbinom1", link="log"), data=mydata) 

Where 'coord' are the spatial coordinates (UTM) represented as a factor, and 'group' is a single level. 

group<-factor(rep(1,n)) 

However, I get the error: 

Warning message: 
In glmmTMB(count ~ waterdepth + temperature + : 
Model convergence problem; non-positive-definite Hessian matrix. See vignette('troubleshooting') 

I didn't have this error when I initially ran my model without a covariance structure. 
Any advice on how I should move forward? 

Thanks!!! Alice 


From: "Ben Bolker" <bbolker at gmail.com> 
To: "Alice Domalik" <adomalik at sfu.ca> 
Cc: "Mollie Brooks" <mollieebrooks at gmail.com>, r-sig-mixed-models at r-project.org, "Kasper Kristensen" <kaskr at dtu.dk> 
Sent: Monday, October 9, 2017 3:05:19 PM 
Subject: Re: [R-sig-ME] spatial covariance structure in glmmTMB 

Did you see the part of the document (sub)titled "Adding coordinate 
information" ? (The example given in the vignette only does a 1-D 
example, but it gives instructions that should in principle work for 
2-D ...) 

On Mon, Oct 9, 2017 at 12:05 PM, Alice Domalik <adomalik at sfu.ca> wrote: 
> I did look at the vignette, and the notation wasn't clear to me. 
> If, for example, I wanted to add a spatial gaussian, would I write my model as: 
> 
> m1<-glmmTMB(count~gau(coordinates) + waterdepth + temperature + chl.conc + (1|individual), family=list(family="truncated_nbinom1", link="log"), data=mydata) ? 
> 
> It also isn't clear to me if spatial coordinates need to be in a particular format. 
> 
> 
> From: "Mollie Brooks" <mollieebrooks at gmail.com> 
> To: "Alice Domalik" <adomalik at sfu.ca> 
> Cc: r-sig-mixed-models at r-project.org, "Kasper Kristensen" <kaskr at dtu.dk> 
> Sent: Monday, October 9, 2017 6:42:48 AM 
> Subject: Re: [R-sig-ME] spatial covariance structure in glmmTMB 
> 
> Hi Alice, 
> Have you looked at the vignette on covariance structures? There's a section on spatial correlations. 
> 
> [ https://cran.r-project.org/web/packages/glmmTMB/vignettes/covstruct.html | https://cran.r-project.org/web/packages/glmmTMB/vignettes/covstruct.html ] 
> 
> cheers, 
> Mollie 
> 
> On Mon, Oct 9, 2017 at 3:22 PM, Alice Domalik < [ mailto:adomalik at sfu.ca | adomalik at sfu.ca ] > wrote: 
> 
> 
> Dear list, 
> 
> I am fitting mixed effects models using the package glmmTMB to investigate habitat use. 
> My current model has the form m1<-glmmTMB(count~waterdepth + temperature + chl.conc + (1|individual), family=list(family="truncated_nbinom1", link="log"), data=mydata) 
> I would like to add a spatial covariance structure based on geographic coordinates (mydata$x and mydata$y). 
> However, I do not know how to modify my model formula to include a spatial covariance structure. Does anyone have example code to show how this is done? 
> 
> Any help would be much appreciated! 
> 
> 
> 
> [[alternative HTML version deleted]] 
> 
> _______________________________________________ 
> [ mailto:R-sig-mixed-models at r-project.org | R-sig-mixed-models at r-project.org ] mailing list 
> [ https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models | https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models ] 
> 
> 
> 
> 
> 
> [[alternative HTML version deleted]] 
> 
> _______________________________________________ 
> R-sig-mixed-models at r-project.org mailing list 
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models 

	[[alternative HTML version deleted]]