Data stack and factors

[Jussi]

Dear all,

I'm wondering whether INLA treats factors incorrectly or am I missing something. In this toy example there is a variable with 6 factor levels, which I put into a stack and feed to INLA, which returns estimates for 7 levels!? However, if I give the data directly to INLA I get 5 as should:

data(InsectSprays)

n = nrow(InsectSprays)

loc = matrix(runif(n),n,2)

mesh = inla.mesh.create.helper(points.domain=loc, max.edge=c(0.1,0.5))

spde = inla.spde2.matern(mesh)

index = inla.spde.make.index("spatial", mesh$n)

A = inla.spde.make.A(mesh, loc, index=1:n)

stack = inla.stack(data=list(count=InsectSprays$count), A=list(A, 1),

                   effects=list(c(index, list(intercept=1)), list(spray=InsectSprays$spray)))

summary(inla(count ~ -1 + intercept + spray, data=inla.stack.data(stack), control.predictor=list(A=inla.stack.A(stack))))

summary(inla(count ~ spray, data=InsectSprays))

[Jussi]

I just updated INLA. Now I get 6 estimates, so still something wrong.

[Finn Lindgren]

Dear Jussi,

inla.stack() does not currently support factors. You can transform them
into 0/1 variables instead and build your model using that.

Another way is to use model.matrix() and feed that into inla.stack() as
a matrix, which will allow you to keep most of the rest of the code
intact. I'll see if I can find the example I wrote for that a long time
ago...

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 post to this group, send an email to
> r-inla-disc...@googlegroups.com.
> Visit this group at
> http://groups.google.com/group/r-inla-discussion-group?hl=en-GB.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[Jussi]

Ok, I see. Thank you for the answer.

[Elias Teixeira Krainski]

We have an example with factors on inla.stack() on
any answer on semi continuous data at this list.