# [S] scaled Venn diagrams?

Carl Schmertmann (schmertmann@fsu.edu)
Wed, 22 Jul 1998 10:52:36 -0400

S-newsers:

Has anyone written S+ code to draw scaled Venn diagrams as a means of
visualizing data?

Here's a concrete example. I am trying to draw a figure representing living
arrangements of the US elderly population. I have three (T,F) variables for
each elder, indicating whether or not the elder lives with a spouse, with a
son, with a daughter. There are 2^3=8 possible living arrangements, each
of which can be represented by a triple of logical variables
(Spouse,Son,Daughter).

The 2x2x2 array of counts in 1000s (apologies if these don't end up aligned

, , No Spouse
No Son Son
No Dau 15257 1543
Dau 1359 212

, , Spouse
No Son Son
No Dau 19827 2037
Dau 1318 372

I'm looking for an algorithm that takes an array like this, and plots 3
squares (or rectangles, circles, ellipses ...) such that

(1) each square's area is proportional to the total population with that
attribute (in this case 23554 with spouse, 4164 with son, 3261 with daughter)

(2) the areas of the intersections are proportional to the populations with
the corresponding combination of attributes (e.g., area of
intersection of 'Spouse' square and 'Son' square proportional to 2409;
area of intersection of all 3 squares prop. to 372)

Any suggestions (including algorithms for 4+ attributes!) are welcome. I
subscribe to the digest version of S-news, so off-list replies will arrive

- Carl
------------------------------------------------------------------
Carl Schmertmann
Dept of Economics & Ctr for the Study of Population
Florida State University
Tallahassee FL 32306-2240 USA

Ph +1 850 644 7100
Fax +1 850 644 8818
Email schmertmann@fsu.edu
Email2 schmert@coss.fsu.edu
------------------------------------------------------------------

-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news