Kamada-Kawai from Covariance Matrix

Feb 15, 2012 at 11:39 PM
First, thanks for making NodeXL free so graphical display of concepts can be accelerated. Will definitely donate a decent amount. For a current project, I would like to be able to generate a Kamada-Kawai plot for about 100 vertices based on a covariance matrix for these variables. Specifically, there is a paper (http://www.santafe.edu/media/workingpapers/07-08-030.pdf) which shows in Fig. 8 one of the nicest KW plots I have ever seen (with group colors). The question to the forum is: using NodeXL, what layout and type ( directed/undirected) options, plus filtering covariance values, would result in a similar structure. The fact that there are individual clusters in this figure with between-object edge lines makes me wonder if the authors binarized based on a cut point for covariance value and filled an adjacency matrix before plotting in network-graph software. Any suggestions you have on construction of plots with a similar structure would be appreciated.
Jul 8, 2012 at 3:03 AM
Edited Jul 8, 2012 at 3:10 AM

This is resolved.   FYI - for anyone who needs to input a correlation or covariance matrix into NodeXL and then generate a network image for the between-dimension covariances(correlations), here is how to do it:

1. generate the square matrix

2. list the dimension names in the first row and first column

3. place zeroes in the diagonal and *either* the lower or upper triangular.

4. import as Excel sheet and then plot, and you should have no problems.