[R-sig-Geo] Getis Ord G test in 3 dimensions

[Roger Bivand]

On Wed, 15 Mar 2017, Juanma José via R-sig-Geo wrote:

>> Dear all
>
>> I am working on a 3D home range estimator, we want to calculated home 
>> ranges and core areas on primates that were followed sleeping site to 
>> sleeping site. We collected GPS locations and animals height at equal 
>> intervals along the day. We already calculates home ranges and core 
>> area in 2D using Delaunay triangulation and then applying a hot spot 
>> analysis in ArcGis( Getis Ord G*) using the perimeter of the triangles. 
>> Then, to calculate it on 3D we first created a 3D convex hull formed by 
>> tetrahedrons using animals locations (coordinates in meters)  and 
>> height.
>
>> we are wondering if it is possible to use the function Local G or 
>> Global G test (package Spdep) for 3D locatios (X,Y,Z) as we previously 
>> did in 2D (Hotspot with rendering, ArgGis)

I replied to this off-list, but the OP hasn't taken my points into 
account:

Please note that hotspots are often seriously misleading unless missing 
values have been taken into account. For instance in your case, the 
presence of a predator may lead to clustering of prey birds in 3D, or 
insect-eating birds will cluster in voxels with higher insect densities. 
The case appears to be 2.5D, not 3D, which will undoubtedly affect 
the absence of hotspots underground.

Getis G/G* is a 2D measure, so you'd first have to show that the same 
statistical properties apply to 3D. Really you need 4D as time is crucial. 
My guess is that kernel densities generalize better than LISA, and there 
are better non-parametric statistical bases for carrying out inference 
from 3D kernels. There are quite a lot of R contributed packages providing 
KDE techniques: http://vita.had.co.nz/papers/density-estimation.pdf - in 
the listing you can see the #D. Choosing bandwidths is like choosing 
distance rings for G* but better founded. Note that many homerange 
estimators use a form of KDE, I think.

Please also see spatstat and its book. G* is probably not what you should 
be using.

Hope this clarifies,

Roger



>
> Thanks in advance
>> 
>> 
>> Juanma
> 
> 
> Juan Manuel José Domínguez
> Postdoctoral Researcher
> 
> Conservation Ecology Program
> School of Bioresources & Technology
> King Mongkut´s University of Technology Thonburi
>
> 49 Soi Tienthalay 25
> Bangkhuntien-Chaithalay Road
> Thakham, Bangkhuntien
> Bangkok 10150
> Thailand
> http://cons-ecol-kmutt.weebly.com/
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

-- 
Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; e-mail: Roger.Bivand at nhh.no
Editor-in-Chief of The R Journal, https://journal.r-project.org/index.html
http://orcid.org/0000-0003-2392-6140
https://scholar.google.no/citations?user=AWeghB0AAAAJ&hl=en
http://depsy.org/person/444584