[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [igraph] Implementation of subgraphs() method in VertexClustering
From: |
Tamás Nepusz |
Subject: |
Re: [igraph] Implementation of subgraphs() method in VertexClustering |
Date: |
Thu, 6 Mar 2014 22:33:01 +0100 |
> For these subgraphs, what happens to the edges that connect members of
> different communities? Are they preserved?
Well, they are preserved in the *original* graph of course (since the original
graph is not modified when you invoke subgraphs()), but none of the subgraphs
will include these edges.
> let's just assume that nodes 1, 2, 3, 4, B are in one community/subgraph,
> and nodes 6, 7, 8, 9, A are in the other. How does igraph place the edges
> B-6, B-7, B-8 and B-9?
They won’t be added to any of the subgraphs.
> And how does igraph take them into account when it
> comes to calculating various network metrics of the subgraphs such as
> density, clustering, etc?
Since they are not added to any of the subgraphs, they are effectively treated
as nonexistent. Network metrics of the subgraphs are calculated as if the
subgraph was standing “on its own”. For instance, the density of the subgraph
consisting of nodes 1-2-3-4-B in your figure would be 1.0 since this subgraph
has 5 nodes and all the possible edges exist between then, while the density of
the subgraph consisting of nodes 6-7-8-9-A would be 0.4 because there are four
edges between these nodes, there could be at most 10 possible connections, and
4/10 = 0.4.
All the best,
Tamas