
Poisson Models 339
point process model, model.matrix.ppm returns a matrix with one row for each quadrature point,
and one column for e a ch of the canonical covariates appearing in (9.6). Note that, if the model
formu la involves transfor mations of the original covariates, then model.matrix(fit) gives values
of these transformed covariates.
The spatstat p a ckage also defines a new generic function model.images with a method for
"ppm" objects. This produces a list of pixel images of the canonical covariates.
> fit2 <- ppm(bei
~
sqrt(grad) + x, data=bei.extra)
> mo <- model.images(fit2)
> names(mo)
[1] "(Intercept)" "sqrt(grad)" "x"
Additionally spatstat ...