Hi Everyone,
I've recently discovered octave-gpr package, and I would like to use it. Until now I have been using R gstat for doing that kind of tasks - namely, kriging for predicting spatial data.
According to the terminology I used to work with - I need to create a variogram for a data set before I can create a krigging prediction.
I figured out that creating the variogram is done by the function
gpr_train, and the actually predicted surface is done by
gpr_predict .
Never the less I failed to use the functions, because I find the documentation a bit confusing.
My data is loaded from a file using the following structure:
"x", "y", "obseravation" columns.
next I want to create a "variogram" using
octave:30> g=gpr_train(xs,ys,t)
warning: gpr_train: some elements in list of return values are undefined
I don't really understand what the values of t should be.
For doing a krigging in R I would do the following:
#define coordinates
coordinates(sic.val) = ~x+y
#make variogram
v<-variogram(dataset$dayx~1,dataset["dayx"])
#find best fit with 120,200,10 as initial guesses
fit.Gau.v<-fit.variogram(v, vgm(120, "Gau",200,10))
#create kriging predictor
xv.Gau <-
krige.cv(dayx~1, sic.val["dayx"], fit.Gau.v, nmax = 40, nfold=5)
Can someone please give me an example how to do the same thing with the gpr_package ?
I was trying to find the demos here
but the link seems to be dead :http://artax.karlin.mff.cuni.cz/~hajej2am/octgpr.php
Thanks is advance,
--
Oz Nahum
Graduate Student
Zentrum für Angewandte Geologie
Universität Tübingen
---
Imagine there's no countries
it isn't hard to do
Nothing to kill or die for
And no religion too
Imagine all the people
Living life in peace