[R-sig-Geo] kde output from ks reversed

[andrewaduff]

Thanks for all your help Robert - I appreciate it as Im sure you a busy man!

> Is this proper use of the hpi values from R? 

> I do not know that 

Ill check with the author of that tool as it seems his conversions most
logically would have been using your resampling suggestion in ArcObjects. 
Since I am looking at a number of methods for home range analysis including
brownian bridge, fixed kernels, and others.  I want to make sure that I am
comparing the resultant surfaces correctly.

>1) resampling the result 
 
This seems less than optimal, but as long as the values were rescaled to a
true pdf it seems that it could work fine.

2) A better solution is to provide equally spaced evaluation points, using a
pre-defined square-cell grid 

I do have concern that while all three rasters (r, r2, and r3) show the same
spatial pattern, the values of the sum of the surface are different.  I
guess the key here is rescaling by the sum of the cell values so that they
sum to 1 (a true pdf) and then I can create a volume surface.

 ## Get the volume
udvol_hpi <- getvolumeUD(r)

It looks like I will have to work on getting these rasters into class
"khrud" or "kbbhrud".  Ill do some research on that and if I figure it out I
will post.  All these different software packages (within R and outside R)
can make home range comparison tricky.  Adehabitat goes a long ways to bring
everything together.
-- 
View this message in context: http://r-sig-geo.2731867.n2.nabble.com/kde-output-from-ks-reversed-tp6037090p6044237.html
Sent from the R-sig-geo mailing list archive at Nabble.com.