Re: [igraph] Randomisation of existing graph for statistical test

[Gábor Csárdi]

Tergagene,

I think you can use degree.sequence.game() with method="vl" to sample
a number of networks uniformly, while keeping the degree distribution
fixed. Then calculate your measure(s) of interest on these networks
and show where the measure(s) on the real network fall in this
distribution. According to my limited statistics knowledge this is
basically a permutation test.

The only restriction is that this method always generates connected
graphs, and works only for undirected ones.

Maybe you can do the same with rewire(), but
degree.sequence.game(method="vl") has been shown to be a uniform
sampler. Please don't forget to cite the Viger-Latapy paper (see url
on ?degree.sequence.game) paper if you decide to use this. Thanks.

Best,
Gabor

2009/3/18 Shin Changjin <address@hidden>:
> Hi Guys,
>
> This might be a simple question.
> I constructed a network.
> To show this network is more likely correct, I want to compare 'original
> network' with 'randomised original network'.
> Would you give me any advice for my study?
>
> I am trying to use rewire in R for this analysis.
>
> cheers,
>
> tergagene
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
>
>



-- 
Gabor Csardi <address@hidden>     UNIL DGM