[R-sig-Geo] centroids in 3-space?

Duncan Elkins duncan at duncanelkins.com
Mon Sep 10 21:22:56 CEST 2007


Hello, list-
  Forgive me if this has been covered before- I've just joined and
can't seem to find a way to browse the list archives by keyword rather
than date.
     I'm trying to analyze a bunch of location data from lab
experiments that's roughly in the form of
[ObservationID,x,y,z, subjectID, treatment]

What I'd like to do is plot the cloud of observations for each fish in
3-space, subset that by treatment (or day of the trial), determine
some summary stats on the enclosing polygons and the centroids for
each, then calculate the distance matrix from each centroid to the
others.  This would be to test the hypothesis that the average
subject-subject distance is less under some treatments than others.
     Some of this looks pretty straightforward to do in 2-space,
especially if I use something like LoCoH to get the polygons, but I'm
wondering if there's something canned to do it and use the z-values,
too.  Can someone point me in the direction of a good resource for
this sort of analysis, either in R, ArcGIS, or both?

Thanks,
Duncan Elkins




More information about the R-sig-Geo mailing list