Graphics Reference
In-Depth Information
Nevertheless, interactive linked graphics are usually more flexible in exploratory
data analysis applications. Linking the panel plot to barcharts or mosaicplots of the
conditioning variables and/or adjunct variables or brushing over a shingle variable
is more flexible, though these techniques lack the global overview and the possibility
of static reproduction.
Parallel Coordinate Plots
6.4
Parallel coordinate plots, as described by Inselberg (
, Chapter III.
same vol-
ume), escape the dimensionality of two or three dimensions and can accommodate
many variables at a time by plotting the coordinate axes in parallel. hey were intro-
duced by Inselberg (
) and discussed in the context of data analysis by Wegman
(
).
Geometrical Aspects vs. Data Analysis Aspects
6.4.1
Whereas in Inselberg (
, Chapter III.
same volume), the geometrical prop-
erties of parallel coordinate plots are emphasized to visualize properties of high-
dimensional data-mining and classification methods, this section will investigate the
main use of parallel coordinate plots in data analysis applications. hemost interest-
ing aspects in using parallel coordinate plots are the investigation of groups/clusters,
outliers, and structures over many variables at a time. hree main uses of parallel
coordinate plots in exploratory data analysis can be identified as the following:
Overview
No other statistical graphic can plot so much information (cases and variables)
at a time. hus parallel coordinate plots are an ideal tool to get a first overview of
a data set. Figure
.
shows a parallel coordinate plot of almost
cars with
variables. All axes have been scaled to min-max. Several features, like a few very
expensive cars,threeveryfuel-e
cientcars,andthenegative correlation between
car size and gas mileage, are immediately apparent.
Profiles
Despite the overview functionality, parallel coordinate plots can be used to visu-
alize the profile of a single case via highlighting. Profiles are not only restricted to
singlecasesbutcanbeplottedforawholegroup,tocomparetheprofileofthat
group with the rest of the data.
Using parallel coordinate plots to profile cases is especially e
cient when the co-
ordinate axes have an order like time.
Figure
.
shows an example of a single profile highlighted - in this case, the
most fuel-e
cient car.
