Reduce the size of my network

May 4, 2011 at 5:10 PM

Hello

I am trying to reduce the size of a network because seem to be so large that  my pc collapse when I am trying to make some graph or calculus, The network has 33.000 vertex.

I applied  the dynamic filtering, to concentrate only in  vertex that have  more than two indegrees, but I cant obtain a result, the PC continues thinking for a long time and never finished ( I have a AMD dual core 64). Also I try with the auto fill columns command specifically modify vertex visibility, but My Pc can finish the process. Maybe I should   buy a new PC but first I would try to avoid this alternative.
like to reduce the matrix and eliminate the vertex that in my research I don't have any interest.

The last alternative that I try  was to eliminate vertex manually (with delete) but I don't know how to erase the correspondent edges,and finally reduce the network size.
Thank you for your help and excuse if the question is so naive.

Coordinator
May 4, 2011 at 6:06 PM
Edited May 4, 2011 at 6:11 PM

Hello!

Yes, 33K vertices is a lot of nodes for most PCs.

Your strategy of removing low degree nodes (of which there are often many) is a good one.

NodeXL has two kinds of "filtering" that can sometimes cause confusion: spreadsheet and graph visualization filtering.

Spreadsheet filtering can be accomplished by editing the Vertices worksheet's "Visibility" column, setting the value = 0 for any vertex you wish to exclude from the calculations and the visualization.

This can also be accomplished via Autofill columns.  Select "Vertex Visibility" and choose to map it to a value like Degree or Betweenness Centrality that will help you select just the parts of the network that are most interesting to you.

Be sure to select the options for this setting:

Your setting should set a value higher than zero!  The higher the value, the few the vertices that will remain on the graph.

When a vertex is set to "Skip" it will be excluded from calculations and display in the network visualization.

Graph Visualization Filtering is accomplished through the Dynamic Filters dialog box access from the visualization pane.  This feature removes edges and vertices from view, but not from the metrics calculations.  These vertices are also using up RAM and slow down layouts!

The best way to deal with very large network data sets is probably to filter them before importing into NodeXL, since even those edges and vertices set to Visibility = 0 will consume space in the workbook.

 

May 4, 2011 at 9:46 PM

Thank You very much for your soon answer!, I will follow your instructions.