Missing year intervals in Dynamic Occupancy Models with INLA

[Chris Smith]

Dear INLA support,

     I am considering using INLA to analyze a multi-year dataset using occupancy for a small Mongolian mammal. I was intrigued by the recent paper "Spatio-temporal occupancy models with INLA" (2024), but couldn't find any mention in the paper if it can handle missing time intervals.  I have data from 2019, 2022, 2023, 2025 and my biggest challenge has been that ML packages like Unmarked can't deal with missing time intervals. I was wondering if this is possible in INLA, since I know other Bayesian methods have been cited as capable of dealing with this (though I can find no published examples).

Thanks for any help you can give,

Chris

[Finn Lindgren]

Hi Chris,

shouldn't be a problem in principle. Assuming you want the model to be indexed by 2019:2025, and you're using a group="ar1" or similar model, and sf data (only affects the "geometry" input below which is sf specific), with inlabru you can say

   + my_model_component(geometry, model = spatial_model, group = year, group_mapper = bru_mapper(fm_mesh_1d(2019:2025), indexed = TRUE), control.group=list(model="ar1"))

The details will differ a bit depending on precisely what component model you want.

(I strongly recommend using the inlabru interface for this, as doing it in plain inla() is a pain to debug, and most issues are solved by the inlabru interface automatically)

Finn

--
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, visit https://groups.google.com/d/msgid/r-inla-discussion-group/52f89d2d-59f6-4af1-aa5c-0f447abe0d54n%40googlegroups.com.

--

Finn Lindgren
email: finn.l...@gmail.com