many network metrics make more sense when compared against the benchmark of a random network. The thing is, the Erdos-Renyi is not necessarily the best benchmark for some observed networks, for example in social network analysis: this is because the degree distribution of the E-R can be very different from the one of the observed network. I was wondering: is there a way that I can, in Gephi, create a random graph with a given number of nodes, number of edges and degree distribution? Or, even better, to create a random graph equivalent to a given observed graph already loaded into Gephi?
I know this is done when computing modularity, with stubs of individual nodes being rewired at random to generate a random-null. Is there any other way? Thanks!Statistics:Posted by albertocottica — 03 Nov 2012 11:09
]]>