[R-sig-Geo] spatstat - how to choose an optimal bandwidth for smooth.ppp ?

[Adrian.Baddeley at csiro.au]

Mathieu Rajerison <mathieu.rajerison at gmail.com> writes:

> >I know how to determine an optimal bandwidth for the kernel density of
> >points using mean suared error method as possible with splancs package.
> >
> >But I was wondering if there was an equivalent method for the density of
> >weighted points as performed by spatstat::smooth.ppp

smooth.ppp does not compute the density of weighted points. That is computed by density.ppp with the argument 'weights'. 

smooth.ppp computes a kernel-smoothed average (Nadaraja-Watson smoother) of the ***mark values** of the point pattern. For example with the 'longleaf' pattern of trees marked by their diameter at breast height, plot(smooth.ppp(longleaf)) will show you a map of the spatially-varying average diameter of trees. 

In the next release of spatstat (1.23-4) to be released in a couple of days, I will include some code for automatic bandwidth selection for smooth.ppp by least squares cross-validation. 

Adrian Baddeley