Hi all
I’m trying to a fit a SPDE model to a set of observations in the Himalayas. When I run a simple intercept only model with the SPDE term, the estimate of the intercept is quite different (~ -0.21) to a model with an intercept but no SPDE (~ -0.17). The mean of the response variable (mb_mwea) is about -0.17.
At first I thought that this was due to either me using the default mesh settings, or the clustered nature of the data (like this post: https://groups.google.com/g/r-inla-discussion-group/c/RqZ5v8jCkzo). So I've tried to improve on the mesh and to incorporate priors on the range and sd, but I still get roughly the same estimate for the intercept term.
I’d appreciate any advice on what I might be doing wrong here. The data are relatively normal looking, but somewhat leptokurtic, so it could be that using a Gaussian likelihood is not the best choice.
I’ve attached the dataset and the R script here.
Thanks in advance!
Simon
--
You received this message because you are subscribed to the Google Groups "R-inla discussion group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to r-inla-discussion...@googlegroups.com.
To view this discussion on the web, visit https://groups.google.com/d/msgid/r-inla-discussion-group/a2553098-1f05-4e07-ba4b-f32a902aea3dn%40googlegroups.com.
On 23 Jun 2021, at 08:11, Finn Lindgren <finn.l...@gmail.com> wrote:
Hi Simon,
On 23 Jun 2021, at 08:14, Finn Lindgren <finn.l...@gmail.com> wrote:
Note: in the illustrative basic example, I forgot to say the domain was centered around location zero.