> However, it seems that using the "corr" excludes the
possibility to
> plot the correlation-matrix using a matrix, right? I think this is a
> bit unfortunate. In this case, the user would still need to program it
> itself or use an additional external package, right?
In git I've added an option --matrix=matname for "corr" (as for
"freq"
and "summary"), which should take care of this point. Example:
<hansl>
open data4-10
matrix X = {dataset}
corr --matrix=X --plot=display
</hansl>
Great, it works fine. I also like that the variables names are
automatically added! ;-)
> However, I would also favor a monochrome option for the heat plot
as I
> had in my little function. Especially when it comes to publication,
> one often needs to provide black-white figures. So I would favor
> another option like "--monochrome" or so.
OK, but do you have a suggestion for how to distinguish negative from
positive correlations in grayscale? (Simply running a scale from white =
-1 to black = +1 doesn't seem like it'd be very intuitive.)
Mmhh, I had a look how R deals with this. But you're right, it doesn't
seem very intuitive...
I really happy about this nice little feature, Allin!
Artur