Before moving into specific methods to partition our network graphs, we should understand why we choose to do so. There are at least two main reasons for segmenting our graphs in this fashion:
To highlight patterns in the data based on underlying statistical or behavioral patterns. This can be especially essential when we have a dense network where patterns are not obvious to the naked eye.
To make a network more visually attractive through the use of size and coloring options.
Note that these two are not mutually exclusive aims, but can be used to great advantage together. So the end goal of the efforts taken for partitioning or clustering should be to enhance the viewer's ability to interpret the graph. Clustering methods will do this through specialized algorithms that interpret network patterns into distinct groupings (clusters) of similar nodes. Partitioning is typically more manual, but also seeks to deconstruct the network into meaningful groupings...