I will give it a try, this sounds like a sensible suggestion, although -
for the first suggestion - my network already has weights (the strength of
a relationship). Need to find out how to create (how does one assign
coordinates and create noise?) and assign (how does one 'place a node'?).
Have some work with layouts to do. Many thanks.
Bob

Kindly,
Bob Pap


On 29 August 2013 12:22, Tamás Nepusz <[email protected]> wrote:

> > Maybe some has faced the following situation: vertices in a  network
> have an attribute (think of - for example - a sport they play frequently),
> which has a role in how they are related (within-attribute edges are more
> frequent and stronger than outside-attribute).  Would it be possible to
> layout the network in a way that graphically 'groups' the attributes (sport
> teams) and then still shows the vertices and their edges in a sensible way?
> You can probably use the weighted Fruchterman-Reingold layout in igraph.
> Set up the edge weights such that edges within a group have significantly
> larger weight than edges between groups, and then run the layout algorithm.
> That should do the trick. In case the layout algorithm is not able to
> figure out an appropriate layout on its own, you can "help" it by providing
> it with a seed layout where the nodes in the different groups are already
> separated to some extent. For instance, if you have k groups, then choose k
> points along the circumference of a circle with a sufficiently large
> radius, and place the nodes in group i around point i, adding some random
> noise to avoid nodes ending up at exactly the same position. Then pass this
> seed layout to the weighted Fruchterman-Reingold layout with appropriate
> weights.
>
> All the best,
> --
> T.
>
>
> _______________________________________________
> 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

Reply via email to