Graphics Reference
In-Depth Information
Figure
.
.
Two-dimensional layout of the small protein interaction network shown in Fig.
.
teins TLG
and YIP
are positioned close to the center of the plot, signifying their
central role in this network.
he next example comes from the UCI machine learning repository. he data set
consists of features of handwritten numerals (
-
) extracted from a collection of
Dutchutility maps. hereare
patterns perclass,and
variables characterizing
the pixel intensity of the underlying digital image of the digit have been collected.
he pixel intensities are categorical and take values in the
to
range. his is an ex-
ample where linear techniques such as principal component analysis fail to separate
the classes (see top panel of Fig.
.
).
he next set of plots in Fig.
.
shows the layouts of a few large graphs that have
been used for testing graph partitioning algorithms (Walshaw,
).he first graph
is comprised of
vertices and
edges, the second of
vertices and
edges, and the third of
vertices and
edges. hey are derived from
computational mechanics meshes and characterized by extreme variations in the
mesh density and the presence of “holes.” he layouts shown are based on weighted
graphs that were built by considering for each vertex its ten nearest neighbors in the
Euclidean metric and calculating exponentially decreasing weights. It can be seen
that the layouts capture, to a large extent, the underlying structure of the graphs in
terms of density and the presence of “holes.”
Bipartite Graphs
4.3.4
As noted inSect.
,the graph representation of a contingency table and of a categori-
cal data set has some special features, namely, the node set V can be partitioned into
two subsets. For example, in the case of a contingency table, the categories of one
variable form the first subset and those of the other variable the second one. Notice
thatthereareonlyconnections between membersofthesetwosubsets.Ananalogous
situation arises in the case of categorical data, where the first subset of nodes corre-
sponds to the objects (e.g., the sleeping bags) and the second subset to the categories
