Newman's modularity clustering for graphs

Question

I am interested in running Newman's modularity clustering algorithm on a large graph. If you can point me to a library (or R package, etc) that implements it I would be most grateful.

best ~lara

For those interested in answer: http://stats.stackexchange.com/questions/1915/newmans-modularity-clustering-for-graphs — mbq, Aug 24 '10 at 21:50

score 9 · Answer 1 · answered Oct 25 '12 at 19:07

Use the igraph package for R: http://igraph.sourceforge.net/doc/R/fastgreedy.community.html this implements a fast algorithm for community finding using the newman-girvan modularity maximization method.

your code will look like this:

library(igraph)
# read graph from csv file
G<-read.graph("edgelist.txt", format="ncol")
fgreedy<-fastgreedy.community(G,merges=TRUE, modularity=TRUE)
memberships <-community.to.membership(G, fgreedy$merges, steps=which.max(fgreedy$modularity)-1)
print(paste('Number of detected communities=',length(memberships$csize)))
# Community sizes:
print(memberships$csize)
# modularity:
max(fgreedy$modularity)

score 1 · Answer 2 · answered Jun 04 '14 at 08:19

1

I'm not quite sure whether the open-source data visualization tool, Gephi, is running with this algorithm. As I know, it runs with the algo in paper: Fast unfolding of communities in large networks

It's also a modularity based methods.

answered Jun 04 '14 at 08:19

zihaolucky

196
1
11

score 0 · Answer 3 · edited Jul 28 '11 at 09:40

0

There's a method in the excellent networkx package that returns a Newman-Watts-Strogatz small world graph.

edited Jul 28 '11 at 09:40

shanethehat

15,460
11
57
87

answered Sep 25 '10 at 01:01

Richard Careaga

628
1
5
11

3

the question was to cluster a graph, not to generate it - which is what that function does – lynxoid Sep 21 '11 at 22:31

Newman's modularity clustering for graphs

3 Answers3