Graphics Reference
In-Depth Information
Nevertheless, a limit is still required because the time taken to draw a single plot
can be multiplied many times when producing plots of a large number of observa-
tions and when running batch jobs involving a large number of plots.
In R, complex plots, such as Trellis plots produced by the lattice package, can be
slowenoughtoseeindividualpanelsbeingdrawn,butmostusersfindthisacceptable.
he entire suite of figures for a medium-sized book can still be generated in much
less than a minute.
Output Formats
3.4.3
When producing plots for reports, it is necessary to produce different formats de-
pending on the format of the report. For example, reports for printing are best pro-
ducedusing PostScript or PDF(Bienz andCohn,
)versions of plots,but forpub-
lication on the World Wide Web, it is still easiest to produce some sort of raster for-
matsuchasPNG.herearemanyexcellent piecesofsotware forconverting between
graphics formats, which reduces the need for statistical graphics sotware to produce
outputinmanyformats;simplyproducewhateverformatthestatistical graphicssot-
ware supports and then convert it externally.
Nevertheless, there are still some reasons for statistical graphics sotware to sup-
portmultipleformats.Oneexampleisthatsotwarecanraisethebarforthelowest-
common-denominator format. For example, R performs clipping of output for for-
mats that do not have their own clipping facilities (e.g. the FIG format; Sutanthav-
ibul,
). Another example is that some formats, especially modern ones, provide
features that are unavailable in other formats, such as transparency, hyperlinks and
animation. It is not possibleto convert a morebasic format into a moresophisticated
format without adding information. Essentially this says that if you are going to aim
for a single format, aim high.
Finally, it is worth noting that a description of a plot in the original language of
a statistical graphics sotware system is a viable and important persistent storage op-
tion. For example, when producing plots with R, it is advisable to record the R code
that was used to produce the plot in addition to saving the plot in any 'traditional'
formats such as PDF or PostScript. One important advantage with retaining such
ahigh-levelformat isthat itisthen possibletomodifytheimageusinghigh-levelsta-
tistical graphics concepts. For example, an extra text label can be positioned relative
tothe scales ona plotbymodifying the original R code,but this sort of manipulation
would be inconvenient, inaccurate and hard to automate if you had to edit a PDF or
PostScript version of the plot.
Data Handling
3.4.4
he description of statistical graphics sotware in this chapter has largely ignored
theissueofwherethedatacomefrom.Ononehand,thisisdeliberatebecauseby
separating data from graphics there is a greater flexibility to present any data using
any sort of graphic.
