weighting edges

Question

Hi,

Only just discovered this program today and have to say it's pretty awesome. I know little of network design and such, but I think I'm not doing too badly after an afternoon.

I hope I haven't missed it, but I had a question about edges and weighting. I have a correlation matrix I've generated in excel. It's not just ones and zeros, so I manually created an edge table of interest and have imported it into yEd with the scores as weights, but it hasn't quite worked out. Will the program handle negative weightings? What I was trying to do, with the organic layout for example, would be something where a positive correlations would be clustered and separate from negative ones. The absolute value of the correlation (ie. distance from zero being no relationship) could be represented by thickness.

If there's no way to do this automatically, is there a way to get the program to recognize manually set colour and line thickness? I did the same analysis making my positives green and negatives red and increasing thickness as a function of absolute value, but no layout algorithm (except I think for that UML? maybe?) seems to have anything to do with recognizing colours or thicknesses. Any suggestions? Though even if not, program's still awesome.

Answer 1

Maybe I overlooked something, but as far as I remember, none of yEd's automatic layout commands has a setting for per-edge weights, and none has the ability to derive such values from visual properties like color or line width. The only features that consider edge weights are the Properties Mapper and the tool Centrality Measures.

That said, I agree that having a setting for edge weights in Organic layout would be a nice feature. The layout algorithm itself can handle such weights already but the corresponding user interface controls are missing. Since adding these shouldn't be too complicated, I suggest to create a feature request for this: http://yed.yworks.com/support/qa/ask?cat=1