rpart with circular data?

Hi Rainer,

In a similar situation I used the two components of the normal vector
of the surface ("northing" & "easting"). I.e. for a horizontal plane
both are 0, for a vertical slope facing south northing=-1and
easting=0, etc. This descartian decomposition of the slope vector
avoids the problem of circularity present in the widespreadly used
polar (aspect, slope) decomposition, and thus seems to suit ecological
problems much better to me. However I have not looked into this much,
I am also very interested in the opinion of others.

B?lint


On Mon, May 12, 2008 at 11:35 AM, Rainer M Krug <r.m.krug at gmail.com> wrote:

Hi

 I am planning to use a CART analysis with rpart() to analyse the
 impact of, among others, slope, altitude and aspect on mortality
 rates.
 My question is:
 Is there a p[roblem with using aspect as a predictor as it is circular?
 And if it is a problem (which I suspect), is there a transformation I
 could use to transform aspect?

 Thanks

 Rainer

 --
 Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation
 Biology, UCT), Dipl. Phys. (Germany)

 Plant Conservation Unit
 Department of Botany
 University of Cape Town
 Rondebosch 7701
 South Africa

For aspect, vegetation ecologist usually use an angular transformation, don?t have the reference at hand but that's the road that Dean L. Urban used for his CART analysis of Sequoia Nat. Park
   
  

B?lint Cz?cz <czucz at botanika.hu> wrote:
  Hi Rainer,

In a similar situation I used the two components of the normal vector
of the surface ("northing" & "easting"). I.e. for a horizontal plane
both are 0, for a vertical slope facing south northing=-1and
easting=0, etc. This descartian decomposition of the slope vector
avoids the problem of circularity present in the widespreadly used
polar (aspect, slope) decomposition, and thus seems to suit ecological
problems much better to me. However I have not looked into this much,
I am also very interested in the opinion of others.

B?lint


On Mon, May 12, 2008 at 11:35 AM, Rainer M Krug wrote:

Hi

I am planning to use a CART analysis with rpart() to analyse the
impact of, among others, slope, altitude and aspect on mortality
rates.
My question is:
Is there a p[roblem with using aspect as a predictor as it is circular?
And if it is a problem (which I suspect), is there a transformation I
could use to transform aspect?

Thanks

Rainer

--
Rainer M. Krug, PhD (Conservation Ecology, SUN), MSc (Conservation
Biology, UCT), Dipl. Phys. (Germany)

Plant Conservation Unit
Department of Botany
University of Cape Town
Rondebosch 7701
South Africa

------------------------------------------------------------------------