Thanks These big networks are hard! My past experience is with networks of a couple hundred nodes, at most
Peter Peter Flom Peter Flom Consulting http://www.statisticalanalysisconsulting.com/ http://www.IAmLearningDisabled.com -----Original Message----- From: igraph-help-bounces+peterflomconsulting=mindspring....@nongnu.org [mailto:igraph-help-bounces+peterflomconsulting=mindspring....@nongnu.org] On Behalf Of Gábor Csárdi Sent: Wednesday, April 04, 2012 1:34 PM To: Help for igraph users Subject: Re: [igraph] Working with large networks and how to sample from a graph? On Wed, Apr 4, 2012 at 7:26 AM, Tamás Nepusz <[email protected]> wrote: >> One idea I had was to take a small random sample from the network (say 5,000 nodes) but I am not sure exactly how to do this in igraph. > > Well, it depends on how you want to do it. You can try selecting 5000 nodes randomly from the entire network and then take the subgraph; this is relatively simple: > > library(igraph) > vs <- sample.int(vcount(g), 5000)-1 > g2 <- subgraph(g, vs) > > However, if your graph is large and sparse enough, there is a chance that the resulting graph will not be connected at all, and then your estimates will bear no resemblance at all to the "real" betweenness values. Well, I'm not convinced that there is any kind of sampling that will tell you much about betweenness values in the original network. (Unless you network structure is special and you can use this fact in the sampling.) I would recommend doing some simulations first, with (say) snowball sampling. Gabor [...] -- Gabor Csardi <[email protected]> MTA KFKI RMKI _______________________________________________ igraph-help mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/igraph-help _______________________________________________ igraph-help mailing list [email protected] https://lists.nongnu.org/mailman/listinfo/igraph-help
