----Messaggio originale----
Da: edzer.pebesma at uni-muenster.de
Data: 7-set-2013 15.22
A: <r-sig-geo at r-project.org>
Ogg: Re: [R-sig-Geo] R: Intamap and "honoring" original data
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 09/07/2013 02:14 PM, r_sig_geo at libero.it wrote:
Several:
1. don't ask questions on this mailing list without revealing your
identity and affiliation,
2. don't expect people to answer within 24 hours and/or in weekends,
3. don't simply repost your questions to the nearly 3000 list members,
but rather try to improve your question in case of no response after,
say, a week without response. See also
https://stat.ethz.ch/pipermail/r-sig-geo/2013-September/019254.html .
Hi to all. My problem is to force a Kriging prediction to
"honour" original data. This
because I need to interpolate values that represent the Bayesian
probability that a particular event occurs. For Instance, If for
a locality I computed a probability = 1, or p=0.6, I need to
possibly keep unchanged these values
the interpolation. I know this is a stretch according to what
Kriging
does. As I need to perform interpolation automatically, I'm using
package "intamap" that is great! I found the function
"unbiasedKrige" that should perform , if I understood well,
exactly what I need. Unfortunately, I find
that even by using this function (both using MOK and IWQSEL,
tested many
with different parameters) the predicted values are still too far
from
data at the same locations. What can you suggest me? Thanks.