Re: [igraph] python ig.Graph.multilevel_community

[Tamás Nepusz]

Dear Lucio,

> I am trying to run the community detection algorithm multilevel_community on 
> the attached graph using the commands:
> 
> com1 = g.community_multilevel(weights='weight')
> com2 = g.community_multilevel(weights=g.es['weight'])
> 
> In both case, the algorithm seams to not take in account the edges weights 
> returning each node as a single cluster.
This is not what happens on my machine:

>>> g = load("DI_clean.pickle")
>>> com1 = g.community_multilevel(weights="weight")
>>> len(com1)
63
>>> g.vcount()
1619

This indicates that you have 63 clusters for 1619 vertices. Also, the 
modularity is positive:

>>> com1.q
0.021505758059457203

Please confirm that the commands above indeed produce 1619 clusters in your 
machine; if this is the case, let us know which platform you are using and what 
version of igraph so I can try reproducing your issue.

Cheers,
Tamas