Graphics Reference
In-Depth Information
it may be much more effective to filter out Richard during this initial
computation. In the e-mail data set, all links to Richard have been removed.
A number of scenarios exist in which it may be desirable to filter out links
during the data preparation stage, such as removing some of the weaker
links in massive graphs to make it faster to process the graph in later steps.
For example, in the case of correlated time series for stocks, the result
shown in Figure 3-4 is a fully connected graph with a link for every possible
combination. Processing the data so that only the top few links for each
stock are retained can result in a completely different visual layout later.
Figure 3-4: Same data set of stock correlations. In the left image, all links
were used. In the right image, only the top three links per node were output
and used.
At this point, these two data sets can be output into the appropriate file
formats that can then be read by graph software. You can learn more details
about the programming in Chapter 8.
 
 
Search WWH ::




Custom Search