Uneven time series length in spatio-temporal model

[Sylvan Benaksas]

Hi,

does the spatiotemporal random effect take into account uneven amounts of data across time?

 I am running a spatiotemporal model where I am joining together three datasets, two of these run from 2000-2020 while the third runs from 2009-2020. So half way through the amount of data will increase significantly.

Can the spde effect take this into account and realise that the response did not just increase majorly after 2009?

My initial crude solution is the inclusion of a binary factor, with levels before and after the inclusion of the third dataset,

some example code:

gtime <- as.integer(seq(2000, 2020, length = 7))
gtime
mesh.t <- inla.mesh.1d(loc=gtime, degree = 1)

dim(A2 <- inla.spde.make.A(mesh, group=dat1$year, loc=loc,
                           group.mesh=mesh.t, n.group = mesh.t$n))

 inla(... f(s.z, model=spde2, group=s.z.group, control.group=list(model='ar1', hyper=h.spec))...)

Any help would be greatly appreciated,

Sylvan

[Finn Lindgren]

If the observations are conditionally independent, given the spatial field, then you shouldn’t need to do anything special at all. You just have more observations. the spatiotemporal field exists whether it’s observed or not.

Finn

On 15 Jun 2022, at 15:16, Sylvan Benaksas <sylva...@gmail.com> wrote:

Hi,

--
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/67d5ad87-b3c7-445e-a2b5-ab7d87b2ede3n%40googlegroups.com.