Graphics Reference
In-Depth Information
his chapter describes the requirements for a modern statistical graphics system for
the production of static plots. here is a discussion of the production of complete
plots, customizingplots, addingextra output toplots andcreating entirely newplots.
Statistical graphics is described as an extension of a general graphics language.
here is an emphasis on the importance of support for sophisticated graphics facili-
ties such as semitransparent colours, image compositing operators and the complex
arrangement of graphical elements.
Static displays of information continue to be the primary graphical method for the
display and analysis of data. his is true both for presentation purposes, where the
vast majority of data displays produced for articles and reports are still static in na-
ture,andfordataexploration,wheremanyimportantstatisticaldiscoverieshavebeen
made based simply on static displays (e.g. Cleveland's barley data discovery using
Trellis plots; Cleveland,
). he recent advances in dynamic and interactive dis-
plays (e.g. Swayne et al.,
; heus,
) provide us with wonderful additional
tools, but static displays still play a fundamental role in the presentation of data.
here are very many sotware packages (some of them statistical) that provide
ways to produce static displays of data. his is good, because these computer pro-
grams allow us to produce more complex graphics, and graphics in greater volumes,
than was ever possible when working just with pen and paper. But how good is the
sotware for displaying data? More importantly, how good could the sotware be?
What should we expect from our statistical graphics sotware?
hischapteraddressesthesequestions bydiscussing theimportant features which
sotware for the static display of data should provide. In addition, there are descrip-
tionsofwaystoprovidethosefeatures.Foreachtopic,therewillbeanabstractdiscus-
sion of the issue followed by concrete examples implemented in R (R Development
Core Team,
).he useof R is natural for me due to my personal familiarity with
the system, but it is also justified by the fact that R is widely acknowledged as being
pretty good at producing static displays of data, and, to my knowledge, some of the
ideas can only be demonstrated in R.
The Grammar of Graphics
A comprehensive overview of statistical graphics is provided by Wilkinson's Gram-
mar of Graphics (Wilkinson,
,
).Wilkinsonoutlinesasysteminwhichstatis-
ticalgraphicsaredescribedinahigh-level,abstractlanguageandwhichencompasses
more than just static graphical displays.
his chapter provides a different view, where statistical graphics is seen as an ex-
tension of a general graphics language like PostScript (Inc.,
) or SVG (Ferraiolo
etal.,
).hisview islowerlevel,moreexplicit about the basic graphical elements
which are drawn and more focused on static graphics.
To emphasize the difference, consider a simple barplot of birth rate for three dif-
ferent types of government (Fig.
.
). A Grammar of Graphics description (or part
