CODEWITHSUNDEEP
spatstat
Richard M
Richard M

Reputation: 25

how can I do a point process model in spatstat on 1 km resolution

I would like to do a point process analysis to model point patterns on 1 km square resolution. I have three layers of covariates as pixel images named; rd, pd and ras. The PPP object is on a polygonal boundary enclosing [-3.1523926, -2.5752286] x [53.31128, 53.70412] units

I have tried the quadscheme function Q <- quadscheme(data, method="grid", eps=1) passed the quad class Q into the ppm formula, model1 <- ppm(Q~ras/rd+pd)

It cants return any results. any help is highly appreciated

Upvotes: 0

Views: 142

Answers (1)

Adrian Baddeley
Adrian Baddeley

Reputation: 2973

You say the ppm command didn't return any results. But you used an assignment, model1 <- ppm(...). This would not produce any printed output. You could print the result: try typing model1 or print(model1) or coef(model1).

You say you want to "model point patterns on 1 km square resolution". Does this mean that the point pattern data are (a) coordinates rounded to the nearest 1km, (b) indicators of the presence/absence of points within each 1km pixel in a pixel grid, (c) spatial coordinates recorded with reasonable accuracy but which you want to analyse by discretising to 1km pixels?

If (c) is correct, then you could use the spatstat function slrm instead of ppm. Just slrm(data ~ ras + pd + rd) should work.

The command quadscheme(data, method="grid", eps=1) does not discretise the spatial coordinates and does not help you with objective (c). It's complicated; I recommend you don't use quadscheme at all, as this is a rather advanced feature. Instead, if data is already a point pattern of class "ppp", you could use data instead of Q in your call to the model-fitting function ppm.

Example data would help.

Upvotes: 0

Related Questions