Dear S-users,
We have built GAM models from presence-absence data of forest communities from
a relatively large dataset (~11000 observations) and with a binomial family of
distribution, using Splus6.2 in windows.
We would like to test for spatial autocorrelation in the residuals. We have
tried many libraries found on the Internet for Splus (spatial by R. Reich) or R
(ape, ade4, spdep, geoR, spatial), as well as a solution in ArcGIS, but all
methods seem to fail at building a distance matrix of 11000x11000.
Ideally we are looking for tests such as Moran or Geary.
Would sub-sampling be a solution ?
Anybody able to help ?
Cheers,
Anthony
--------- I support: www.spacedoor.net ---------
Anthony Lehmann, Dr.
Centre Suisse de Cartographie de la Faune (CSCF)
Terreaux 14, CH-2000 Neuchâtel
SUISSE
http://www.cscf.ch/carto
http://www.cscf.ch/grasp
------------------------------------------------
|