I've 36 clusters with with some atributes and i can calculate the distance between all of them. I would like to create a graph were nodes get attracted to the nodes they are similar and repelled from those who they are different. Therefore i made a graph where I've 36 nodes and 630 edges (from all the nodes to all the nodes but themselves). I thought that ForceAtlas would be the way to go, but i just get all the nodes together or bouncing arround if i increase the speed.
Is force atlas the algorithm i'm looking for to generate the plot? if not any hint of what i'm looking for? if force atlas is the way to go, is there anywhere i can get more information to manipulate it to do what i want? or can you give a similar example such that the algorithm works, maybe my data just have to colapse.
thanks, sik.
small graph visualization based in similitudes
-
pegerp
- Posts:124
- Joined:21 Dec 2011 17:10 [phpBB Debug] PHP Warning: in file [ROOT]/vendor/twig/twig/lib/Twig/Extension/Core.php on line 1275: count(): Parameter must be an array or an object that implements Countable
Re: small graph visualization based in similitudes
Hi massich.
You could try adjusting the edge weights (distances) between nodes with a different algorithm. For example you could have the distance algorithm be 1 / x^2 so that the most closest nodes become connected with heavier edge than those whose distance to each others is greater.
More so you could try to remove redundant edges from the graph. Manually find some reasonable threshold and remove all edges that are not important enough (the weight is too low) to be part of the graph.
Both of these steps you'd do outside Gephi and then import the graph to Gephi to see what kind of network layout appears. ForceAtlas 2 is most of the time very usable for this kind of networks.
Good luck!
You could try adjusting the edge weights (distances) between nodes with a different algorithm. For example you could have the distance algorithm be 1 / x^2 so that the most closest nodes become connected with heavier edge than those whose distance to each others is greater.
More so you could try to remove redundant edges from the graph. Manually find some reasonable threshold and remove all edges that are not important enough (the weight is too low) to be part of the graph.
Both of these steps you'd do outside Gephi and then import the graph to Gephi to see what kind of network layout appears. ForceAtlas 2 is most of the time very usable for this kind of networks.
Good luck!