How to make Pierre de Vries' graph?

Jul 16, 2010 at 4:38 AM
Edited Jul 16, 2010 at 4:39 AM

Hi all,

First, thank you to the developers who made this awesome tool.

I have some data that I would like to format like Pierre's (one of the demo graps on NodeXL's homepage), but it seems like I would have to do that manually. The Fruchterman-Reingold layout is good for a small number of nodes, but when there are 100 or so, and there are distinct subgraphs, the subgraphs get overlaid and it is kind of messy. How do I get around this, or is it even possible? Was Pierre's graph arranged by hand?


Jul 16, 2010 at 4:52 PM
Hi Stimpy I'm afraid I did largely arrange the graph by hand. I started with the FR layout, and then used clustering to help me separate out the sub-graphs visually. After that it was a lot of moving around. However, this was quite a long time ago, and the tool has moved along substantially - thanks to Tony & the crew's efforts. I suspect it would be easier now. Good luck Pierre
Jul 17, 2010 at 12:59 PM
Dear group,
How do I get NodeXL to render multiple edges between the same two nodes? Thus, I want to represent the presence or absence of different kinds of social interactions between two persons.

>John R. Rodgers, Ph.D.

>Room M803

>Department of Pathology & Immunology

> Baylor College of Medicine

> Houston, Texas 77030

> 713-798-3903

> fax: 713-798-3700



> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

> ~One can study only what one has first dreamed about. Science is formed rather on a reverie than on an experiment, and it takes a good many experiments to dispel the mists of the dream"

> - Claude Bachelard,1938 "The Psychoanalysis of Fire", Beacon Press, p. 22 (Eng. translation by Alan Ross, 1964

> ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Jul 19, 2010 at 5:17 PM
Edited Jul 19, 2010 at 5:17 PM


NodeXL currently draws duplicate edges on top of each other, which makes them indistinguishable. We're talking about how we might add "intelligent edge routing" that would instead draw them next to each other, but for the moment I can only offer a couple of workarounds:

1. If you are interested mostly in the number of duplicate edges between nodes, you can merge duplicate edges (NodeXL, Prepare Data, Merge Duplicate Edges), which will add an Edge Weight column to the Edges worksheet. You can then autofill the Width column on the Edges worksheet (NodeXL, Visual Properties, Autofill Columns) based on the values in the Edge Weight column. Thus, edges between people who have more kinds of social interactions will be wider.

2. For small graphs, some people have had success manually increasing the Width of the duplicate edges while simultaneously reducing their opacity, which makes the duplicate edges visible even though they are on top of each other.

-- Tony

Sep 5, 2012 at 11:39 PM

Hi Tony,

The last post to this thread was from 2010, has there been any progress in this area to get the network to show multiple edges between the same two vertices?

I'm currently using the edge weight solution, but I want people to be able to see multiple types of interactions that took place typically at different times as well.




Sep 6, 2012 at 1:13 AM
Edited Sep 6, 2012 at 1:14 AM

Hello, Mike:

No, I'm afraid not.  I'll bring it up again in our next meeting.

-- Tony