[R] problem with corrgram function

[Eik Vettorazzi]

yes it is.
and a correlation of 0 isn't exactly white (#FFFFFF) either.
have a look at the panel.pie function.

the crucial part is

    ncol <- 14
    pal <- col.corrgram(ncol)
    col.ind <- round(ncol * (corr + 1)/2)

so an correlation near -1 maps to an index 0, which isn't a proper index
in R.
Alter these lines to
     ncol <- 15 #so 0 becomes "#FFFFFF"
     pal <- col.corrgram(ncol)
     col.ind <- round((ncol-1) * (corr + 1)/2)+1

hth.


Am 28.06.2011 13:11, schrieb Niels Janssen:
> Dear list,
> 
> I have a problem with the "corrgram" function. It does not seem to
> "color" large negative correlations, while the same correlation, if
> positive, provides no problems. Is this a bug?
> 
>     require(corrgram)   
>     a = seq(1,100)
>     b = -jitter(seq(1,100), 80)
>     cor(a,b) # r about -.96
>     c=as.data.frame(cbind(a,b))
>     corrgram(c, order=NULL, lower.panel=panel.pie,upper.panel=NULL,
> text.panel=panel.txt) # no color
>     
>     c$b = -1*c$b # flip direction of correlation
>     cor(c$a, c$b) # r now about +.96
>     corrgram(c, order=NULL, lower.panel=panel.pie,upper.panel=NULL,
> text.panel=panel.txt) #no problem with color.
> 
> Thanks!
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

-- 
Eik Vettorazzi

Department of Medical Biometry and Epidemiology
University Medical Center Hamburg-Eppendorf

Martinistr. 52
20246 Hamburg

T ++49/40/7410-58243
F ++49/40/7410-57790