Communities - Graph Analysis and Visualization - page 404

Start by swapping the order of the Vertex and Group columns on the Group
Vertices worksheet. Click the drop-down button on the Vertex column to
sort it A to Z so that it can be used for lookup. Switch to the Vertices
worksheet and add a column to the right of the graph data. Add a formula
to the cells in the new column that will look for the Vertex in the Group
worksheet, and populate each cell with the name of the clique group or an
empty string if not found:
=IFERROR(VLOOKUP(Vertices[[#This Row],[Vertex]],
GroupVertices[[Vertex]:[Group]],2,FALSE),"")
Create a new column in the graph data area of the worksheet and copy from
the cells you added in the previous step. Use Paste Values to add a copy of
the group references into the new column and label the column Clique.
It would be useful to explore the demographic characteristics of cliques.
However, you already looked at gender in the data and determined that
women do not often post together, so sex will not be a factor. What about
locale? Facebook locale consists of ISO codes for language and country
joined by an underscore. To explore language and country separately,
extract a new column for each:
=IF(LEN(Vertices[[#This Row],[Locale]])>0,
RIGHT(Vertices[[#This Row],[Locale]],2),"")
=IF(LEN(Vertices[[#This Row],[Locale]])>0,
LEFT(Vertices[[#This Row],[Locale]],2),"")
Exporttheenhanced GraphMLfileandimportitintoGephi.InthePartition
tab, map the Node color to Clique. Map null (meaning a node that is not
part of a clique) to white. Inspect the resulting visualization in the preview
mode. As shown in Figure 11-17 , the result is arguably beautiful, but is it
useful? It's difficult to make sense of cliques in the middle where the nodes
are densest. A clique has a very hard-edged definition. It would be ideal if
the visual boundaries of the cliques were equally hard-edged.