[R-sig-Geo] automatic assessment of linear coregionalization model

[Emanuele Barca]

Dear friend,

I am trying to garner a result similar to that of autofitvariogram() 
function with fit.lmc() function.
But results are not good.
The following is my code:

g <- gstat(id = "Balance", formula = Balance ~ 1, data = mydata.sp, set 
= list(nocheck = 1))
g <- gstat(g, id = "Ancillary", formula = Ancillary ~ 1, data = r.sp, 
set = list(nocheck = 1))
v.g <- variogram(g)
g.fit <- fit.lmc(v.g, g, vgm("Exp",
         model.df$range[2]), fit.method = 1, correct.diagonal = 1.01) # 
fit.lmc = TRUE,
g.fit

the outcome is
data:
Balance : formula = Balance`~`1 ; data dim = 74 x 1
Ancillary : formula = Ancillary`~`1 ; data dim = 18683 x 1
variograms:
                      model     psill    range
Balance[1]             Nug 10.864143     0.00
Balance[2]             Exp  8.128785 12946.87
Ancillary[1]           Nug 43.803405     0.00
Ancillary[2]           Exp 52.100889 16156.90
Balance.Ancillary[1]   Nug 21.598834     0.00
Balance.Ancillary[2]   Exp 20.375770 16151.70
set nocheck = 1;

the results are too weird. Is a data problem?


thanks in advance

emanuele barca
-------------- next part --------------
A non-text attachment was scrubbed...
Name: data.rar
Type: application/x-rar-compressed
Size: 11684 bytes
Desc: not available
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20200630/5d1ce8c7/attachment.bin>