Re: [igraph] on VertexClustering

[Tamás Nepusz]

Hello,

> I'm trying to get some graph clustering done, so I create a graph from
> my graphml and try to run VertexClustering.
VertexClustering is just a data structure, not a particular clustering 
algorithm. VertexClustering instances are returned by several graph clustering 
methods, but they are all implemented as methods of the Graph class. Look for 
methods starting with "community_" in the Graph class to get a rough idea of 
the available algorithms.

E.g.:

>>> g = igraph.Graph.Read_GraphML("myfile.xml")
>>> cl = g.community_fastgreedy()
>>> print type(cl)
igraph.clustering.VertexClustering
>>> print cl.membership

Some clustering algorithms return a VertexDendrogram instead of a 
VertexClustering, but since VertexDendrogram is derived from VertexClustering 
in igraph 0.5, you can still use its "membership" property.

By the way, if you want to construct a VertexClustering by hand, you need two 
arguments: the first argument specifies the graph, the second argument is the 
membership vector (i.e. the cluster index for each vertex).

Cheers,
T.