
I am construction a network with repeated interactions, meaning that node A may set up edge with node B repeated. At the same time, the interaction can be either positive and negative, i.e. node A may positively interact with node B twice, as well as negatively
4 times.
Now my question is: if I set the edge color upon the category of the interaction (supposing positive is blue and negative is black), what is the final edge color shown to me? Will Nodexl count the repeated time in the category to decide the winner, or simply
show based on the first or the last interaction under the category? Thanks.
I am doing some analysis of dynamic network.


Jun 6, 2013 at 5:44 PM
Edited Jun 6, 2013 at 5:49 PM

I assume that you are using NodeXL's Autofill Columns feature and you are specifying that you want Edge Color autofilled using a particular column. I'm not clear on the rest of it, though. Please tell me the following:
 Does the column contain categories ("Positive" and "Negative") or numbers (2, 4, 5)?
 If A interacts with B twice, are there two edges between A and B, or just one edge?
 Tony



 Does the column contain categories ("Positive" and "Negative") or numbers (2, 4, 5)?
I record pos/neg to 1/0.
 If A interacts with B twice, are there two edges between A and B, or just one edge?
two in data set. Other attributes like "year/month" can distinct the two by time.
Thanks



The Autofill Columns feature looks at the smallest and largest number in your Interaction column, which are 0 and 1. It assigns black to 0 and blue to 1. (The usual colors are orange and blue, but it sounds like you chose different colors.) If there were
a range of values in the Interaction column, then edges with a value of 0 would become black, edges with a value of 1 would become blue, and edges with values between 0 and 1 would become some color between black and blue. You don't have any inbetween values,
so all edges will be either black or blue in your graph.
NodeXL assigns colors to each edge individually. It does NOT look for duplicate edges, or what you call repeated interactions. If there are two edges between A and B, one with an Interaction value of 0 and another with an Interaction value of 1, then the first
edge will be black and the second edge will be blue.
Unfortunately, NodeXL draws duplicate edges on top of each other, so you can't tell one from the other. (It should bend them out slightly so you can distinguish them, but it doesn't do that.) The result is that you'll get a black edge on top of a blue edge,
or a blue edge on top of a black edge, depending on the order the edges are listed in the Edges worksheet. This obviously isn't what you want to happen.
Because of this, it would be better for you to combine your duplicate edge rows and create an Average Interaction column before using Autofill Columns. In my example, you would combine the two edges between A and B into a single edge with an Average Interaction
value of 0.5. You would then tell Autofill Columns to use the Average Interaction column, and NodeXL would assign a single color (dark blue) to the combined edge.
 Tony



Thank you, Tony. Your answer is very helpful.
 ZC

