[R] Colours in silhouette plots (cluster package)

Patrick Connolly p_connolly at ihug.co.nz
Thu Aug 10 11:00:52 CEST 2006

I tried using colours in silhouette plots by specifying a vector of
colours instead of the default "gray" for the col parameter.

No bars are drawn, though the rest of the plot works as it did with
the grey bars.  On investigating cluster:::plot.silhouette, I came
across this part:

  if (do.col.sort && (lc <- length(col)) > 1) {
        if (lc == k) 
            col <- col[cli]
        else if (lc != n) 
            col <- rep(col, length = n)
       col <- rev(col[attr(x, "iOrd")])

Since attr(x, "iOrd") was NULL, I would end up with an empty col
vector which explained why I didn't get any bars.

When I removed [attr(x, "iOrd")], the plot worked how I think it's
meant to.  The silhouette object I used was created from one obtained
from the pam function in the same package.  I've not tried plotting
other types, so I can't say how general the phenomenon could be.

Could it be that the "iOrd" part belongs when some other conditions
apply or is it some left over code from a previous version?  Or is it
just that I've done something silly?

platform       x86_64-unknown-linux-gnu  
arch           x86_64                    
os             linux-gnu                 
system         x86_64, linux-gnu         
major          2                         
minor          3.1                       
year           2006                      
month          06                        
day            01                        
svn rev        38247                     
language       R                         
version.string Version 2.3.1 (2006-06-01)

Along with cluster with that distribution:
Packaged:      Wed May 17 09:56:22 2006; maechler
Built:         R 2.3.1; x86_64-unknown-linux-gnu; 2006-06-05 11:57:15;


   ___    Patrick Connolly   
 {~._.~}          		 Great minds discuss ideas    
 _( Y )_  	  	        Middle minds discuss events 
(:_~*~_:) 	       		 Small minds discuss people  
 (_)-(_)  	                           ..... Anon

More information about the R-help mailing list