Graphics Reference
In-Depth Information
shows all the correlations—that is, all the edges—as a colored matrix with
the detailed data values and all edges are equally visible.
Figure 4-22:
Using this force-directed graph of stock correlations, the
assigned task requires analyzing links, but some links are difficult to see.
Tip
An adjacency matrix can be effective when the analysis is focused on
individual edges and groups of similar edges.
However, while showing the overall clustering pattern, the force-directed
diagram (
Figure 4-22
) makes it more difficult to see all the links. Some links
are very long, and others are not visible at all. Furthermore, the intuitive
understanding of distance between nodes is useful for understanding the
overall graph but is not effective for comparing individual links.
Look closely at the relationship between Symantec (SYMC) and Oracle
(ORCL) versus eBay (EBAY) and IBM. In the matrix, eBay-IBM is stronger
(0.4, light orange) than Symantec-Oracle (-0.3, reddish). But in the
force-directed diagram, Symantec-Oracle is much closer together than
eBay-IBM, and the link between Symantec-Oracle is difficult to perceive.
Sorting of matrix diagrams helps reveal some of the higher-level patterns.
You can do simple sorting using a tool such as Excel and more advanced
