[OSGeo-Discuss] how to test if spatial variables (layers) are independent

[karsten]

Hi GIS pros,
 
I am working on creating a model in my desktop GIS to characterize
agricultural sites for the purpose to create a layer that can show areas
that have similar natural / agricultural site characteristics that can be
used for site selection in agricultural research. Input layers will be for
example  population density, poverty density, land cover , soil types, tree
density , precipitation, hydrology, distance to market (for a farmer) ,
distance to road, agro ecological zones and others more. 
 
The idea is to cluster the layers first to avoid evaluating parameters that
are in high correlation (or dependant variables) multiple times but come up
with a set of layers (or clustered 'synthetic' layers) that can be an input
for the model evaluation run.
 
Now I did take some statistics classes 20 years ago but am lacking a clear
approach or best practices how to ensure that the layers (aka variables) are
independent. I would prefer to do this in QGIS or R statistical package.
However, if you know of any good OS tool that is a good fit for this please
suggest it to me.
 
I read then searching the internet about the 'mantel test' being suggested
by some , others talked about statistical correlation between a independent
and dependant variables. But I was not sure if that is the best way to do it
as in my case I would want to test the dependency of the variables instead
...
 
Would anyone have suggestions or experience with this that they can share
with me ?
 
Cheers
Karsten

Karsten Vennemann
Principal

Terra GIS LTD
SeattleUSA 
 <http://www.terragis.net/> www.terragis.net

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/discuss/attachments/20150414/2e9cd018/attachment.html>