[R] venneuler plots may be inaccurate

Carl Witthoft carl at witthoft.com
Fri Jan 6 16:08:11 CET 2012


Hi,
I just started playing w/ venneuler, and have a question about the last 
method listed for creating inputs.  This is the method where the source 
is a matrix X, for which each column represents a set, and the 
co-occurrence is defined by the rows.  So, for example:

Rgames> foo
       [,1] [,2] [,3] [,4] [,5]
  [1,]    0    1    1    0   NA
  [2,]    0    1    1    0   NA
  [3,]    0    1    1    0   NA
  [4,]    0    1    1    0   NA
  [5,]    0    1    1    0   NA
  [6,]    0    1    1    1   NA
  [7,]    0    1    1    1   NA
  [8,]    0    1    1    1   NA
  [9,]    0    1    1    1   NA
[10,]    0    1    1    1   NA
[11,]    1    0    1    1   NA
[12,]    1    0    1    1   NA
[13,]    1    0    1    1   NA
[14,]    1    0    1    1   NA
[15,]    1    0    1    1   NA
[16,]    1    0    1    0   NA
[17,]    1    0    1    0   NA
[18,]    1    0    1    0   NA
[19,]    1    0    1    0   NA
[20,]    1    0    1    0   NA
Rgames> plot(venneuler(foo[,1:3]))

The result I get is at http://home.comcast.net/~cgwcgw2/venndiagram.pdf

I would not expect either A or B to extend outside C, and they shouldn't 
overlap either.
Any ideas?   So far as I can tell, the first method (define weights for 
each 'overlap' region, e.g. "A&B"=.2, seems to work correctly).

thanks
Carl
-- 

Sent from my Cray XK6
"Pendeo-navem mei anguillae plena est."



More information about the R-help mailing list