[R-sig-Geo] fuzzy kappa

[Paulo Flores Ribeiro]

Hi Hugo,
Many thanks for the reply. I don't think it's the same. If I'm not 
mistaken, the "weighted" of Cohen's Kappa in the psych package works on 
the confusion matrix of classifications, assigning different weights to 
values ​​further away from the diagonal. Fuzzy kappa has another 
purpose. A classic example (which I read elsewhere) to show the 
potential usefulness of fuzzy kappa is the comparison of two rasters 
with the appearance of a chessboard: if the two boards are aligned, the 
match is perfect; if one of them is rotated 90 degrees, the mismatch is 
total but, however, the mosaic pattern remains exactly the same! Fuzzy 
kappa aims to get around these issues by providing an assessment of the 
mosaic pattern that is closer to a human perspective, rather than a 
purely "mathematical" pixel-to-pixel correspondence assessment.
I've seen articles where the authors claim to have adopted this "fuzzy 
kappa" approach to assess the accuracy of predictions in land use change 
models (which is my case), and I was hoping there would be a 
package/function to accomplish this in R. I'll be very surprised if it 
doesn't exist, as R got me used to finding "answer for everything" in it ;)
Cheers,
PauloFR


Às 10:23 de 24/11/2021, Hugo Costa escreveu:
> Just wondering if fussy kappa is the same as weighted Kappa. If yes, 
> you can try package psych (I used it years ago).
> Hugo
>
> Андрей Чинилин <achinilin using rgau-msha.ru> escreveu no dia terça, 
> 23/11/2021 à(s) 16:47:
>
>     Well, I could be wrong, but I suspect, that you can try "goofcat"
>     function from "ithir" package.
>     Link to pkg - https://bitbucket.org/brendo1001/ithir/src/master/.
>     Andrey
>     23.11.2021, 18:03, "Paulo Flores Ribeiro"
>     <paulo.flores.mail using gmail.com>:
>
>         According to Hagen‐Zanker (2009:
>         https://doi.org/10.1080/13658810802570317), «The Fuzzy Kappa
>         statistic
>         expresses the agreement between two categorical raster maps. The
>         statistic goes beyond cell‐by‐cell comparison and gives
>         partial credit
>         to cells based on the categories found in the neighborhood.»
>
>         Does anyone know of a package/function to compute such a Fuzzy
>         Kappa
>         statistic in R?
>
>         Thanks in advance,
>
>         PauloFR
>
>         _______________________________________________
>         R-sig-Geo mailing list
>         R-sig-Geo using r-project.org
>         https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>     -- 
>     С уважением,
>     Андрей Чинилин
>     _______________________________________________
>     R-sig-Geo mailing list
>     R-sig-Geo using r-project.org
>     https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
	[[alternative HTML version deleted]]