CODEWITHSUNDEEP
rspatstat
Ian Dewan
Ian Dewan

Reputation: 3

Replicated point patterns on linear networks in spatstat

I have point pattern data from a replicated experiment where the points in each replicate are constrained to the same linear network (the data are from daily surveys of a bike path for snakes: each day gives a separate point pattern of locations where animals are found).

I know that in spatstat it is possible to fit point processes to multiple point patterns simultaneously (with mppm), and to fit point process models on linear networks (with lppm); is it possible to do both simultaneously? As far as I can tell, mppm will not accept lpp objects: is there another way of fitting this type of model?

Upvotes: 0

Views: 190

Answers (1)

Adrian Baddeley
Adrian Baddeley

Reputation: 1984

This is not yet fully supported in spatstat.

However, you can do most of what you want by converting the lpp objects to quadrature schemes using linequad and then using these quadrature schemes instead of ppp objects in the hyperframe. Example:

X1 <- spiders
X2 <- runiflpp(25, domain(spiders))
A <- linequad(X1)
B <- linequad(X2)
f <- function(x,y)x
H <- hyperframe(X=solist(A,B), Z=list(f,f))
fit <- mppm(X ~ Z, data=H)

Most methods for mppm will work correctly; except that you can't simulate or predict the fitted model, because it doesn't know that it's supposed to be on a network.

If you have a long list of ppp objects, then instead of converting the point patterns to quadrature schemes one-by-one, you could use solapply( , linequad).

Upvotes: 2

Related Questions