74 lines
2.4 KiB
Plaintext
74 lines
2.4 KiB
Plaintext
|
\documentclass[a4paper]{article}
|
||
|
|
|
||
|
|
%\VignetteIndexEntry{Venn Diagrams with gplots}
|
||
|
|
|
||
|
|
\title{Venn Diagrams with \texttt{gplots}}
|
||
|
|
\author{Steffen M{\"{o}}ller}
|
||
|
|
|
||
|
|
\begin{document}
|
||
|
|
\SweaveOpts{concordance=TRUE}
|
||
|
|
|
||
|
|
\maketitle
|
||
|
|
|
||
|
|
Venn diagrams\footnote{https://en.wikipedia.org/wiki/Venn\_diagram}
|
||
|
|
allow for a quick overview on the number of elements that multiple
|
||
|
|
sets share. And when those elements are representing traits of
|
||
|
|
real objects, like observations in biomedical sciences/marketing/...,
|
||
|
|
then this may direct researchers to further investigations/decisions.
|
||
|
|
|
||
|
|
The {\tt gplots} package provides Venn diagrams for up to five sets.
|
||
|
|
The R code to produce the diagrams is not complicated. The plot
|
||
|
|
function behaves alwas the same, depending only on the number of
|
||
|
|
overlapping circles to draw. Its input is a table that is produced
|
||
|
|
by another function. The function {\tt venn()} calls one after the
|
||
|
|
other and is the only one to be seen by the user. The values shown
|
||
|
|
are returned invisibly.
|
||
|
|
|
||
|
|
The {\tt venn()} function accepts either a list of sets as an argument,
|
||
|
|
or it takes a binary matrix, one column per set, indicating for every
|
||
|
|
element, one per row, the membership with every set.
|
||
|
|
|
||
|
|
The common form with overlapping circles only works with up to three
|
||
|
|
sets, as seen here:
|
||
|
|
\begin{center}
|
||
|
|
<<fig=TRUE,echo=TRUE>>=
|
||
|
|
suppressMessages(library(gplots))
|
||
|
|
venn( list(A=1:5,B=4:6,C=c(4,8:10)) )
|
||
|
|
@
|
||
|
|
\end{center}
|
||
|
|
|
||
|
|
The names of columns or the list elements are the set names.
|
||
|
|
To squeeze extra circles in, those circles need to become ellipses.
|
||
|
|
This works for four sets
|
||
|
|
\begin{center}
|
||
|
|
<<fig=TRUE,echo=TRUE>>=
|
||
|
|
v.table<-venn( list(A=1:5,B=4:6,C=c(4,8:10),D=c(4:12)) )
|
||
|
|
print(v.table)
|
||
|
|
@
|
||
|
|
\end{center}
|
||
|
|
|
||
|
|
and maybe even more impressively also for five.
|
||
|
|
\begin{center}
|
||
|
|
<<fig=TRUE,echo=FALSE>>=
|
||
|
|
venn( list(A=1:5,B=4:6,C=c(4,8:10),D=c(4:12),E=c(2,4,6:9)) )
|
||
|
|
@
|
||
|
|
\end{center}
|
||
|
|
|
||
|
|
The man page of {\it venn()} lists options to change the appearance of
|
||
|
|
the plots, e.g., the names of the sets may be omitted and sizes changed.
|
||
|
|
However, there is ample of opportunity to extend the functionality of
|
||
|
|
this package. To mind come
|
||
|
|
|
||
|
|
\begin{itemize}
|
||
|
|
\item more dimensions (next)
|
||
|
|
\item colors
|
||
|
|
\item variation of size of circles with the number of members the set has
|
||
|
|
\item density plot rather than numbers, identification of individual entries
|
||
|
|
\end{itemize}
|
||
|
|
|
||
|
|
The prime personal interest is more in the increase of dimensions. Please
|
||
|
|
send patches for features you are most interested in.
|
||
|
|
|
||
|
|
\end{document}
|
||
|
|
|