Graphics Reference
In-Depth Information
Figure
.
.
Faceting and degree of interest. he let view shows players since
, faceted by the league
in which they played. Note the brief existence of the Federal League (FL:
-
). On the right
a binned scatterplot of the putout-to-assist ratio is shown. Ater selecting players playing post-
in
the let panel,thelet view shows the interesting cases in black,whereastheright view uses degree of
interest to define the vertical paneling (“interesting” cases are in the lower row)andusesbrightness to
show the counts of points in the binned scatterplot
categories and works best with binary variables. Systems for displaying aggregated
data elements are discussed in Sect.
.
. We could have used a summary function
instead to display the mean degree of interest, mapping it to a brightness scale. Sec-
tion
.
gives details on the dataset being displayed, with notes on the meanings of
the variables used.
he rightmost view in Fig.
.
splits the data into two panels, one for the selected
and one for the unselected subsets. his faceting should not be considered a special
functionality butshouldfitintoexisting faceting schemes,sothat, forexample,atrel-
lis view would simply incorporate the degree of interest as another variable within
the tablelike structure. As a further example, Fig.
.
shows how a binary selection
status is incorporated into the faceting structure. Baseball has had many different
leagues over its history, which are separate pools of teams; each team is a member of
one league only in any year and plays almost exclusively members of its league. Note
that post-
, the putout-assist scatterplot resolves itself into two clusters, whereas
pre-
there is good evidence for at least another cluster. We investigate that more
closely later.
Brushing Scatterplot Matrices
and Other Nonaggregated Views
9.3
One of the earliest linked views works to achieve wide attention was the scatterplot
brushingtechniqueofBecker,Cleveland,andWilks(Beckeretal.
).Byarranging
