Graphics Reference
In-Depth Information
•
Other representations
—Depending on the data and the objective,
you should consider other representations.
• Chord diagrams can be effective for analysis of bidirectional flows
between pairs of nodes in directed graphs.
• Adjacency matrices can be very effective for comparing and
clustering edges (but not path analysis).
• Treemaps and hierarchical pies can be an intuitive way to analyze
hierarchies, particularly when there is a magnitude.
• Parallel coordinate plots can be effective for analyzing graphs with
many variables where relationships between all the variables are
more important than the graph structure.
Summary
The first level of analyzing a graph is to get a high-level understanding of
thegraph.Creatingsomeoverallstatisticsandgeneratingagraphlayoutcan
provide a good understanding of the types of relationships in the graph.
You should consider different kinds of layouts, depending on the objective
and the nature of the graph, ranging from a variety of node-and-link layouts
to other types of layouts such as adjacency matrixes, chord diagrams,
treemaps, and, in some cases, parallel coordinates.
Once you have a layout, you must consider what data you want to reveal
about the nodes and edges, and how to reveal it. Chapter 5 provides a
discussion on the use of size, color, labels, line styles, and so on, to show this
information in the graph.
