Newbi question - shaping nodes by attributes

[Frieder]

Hello everybody,

Is it possible to determine nodeshape/color by attribute? For example if I surveyed a multicultural team, could i get Gephi to show female teammembers with a certain ethnic background who have been working in the team for xxx time in one shape and male teammembers with a certain ethnic background etc. in another? I know this can be done in Visone and I think it is an essential feature. So far I can only find the three different node shapes in Gephi. As a workaround the attributes could be part of the label but then it gets quite crowded and messy looking...

thanks and great work

Freddy

[admin]

Hi, we can't use shapes for the moment: only sizes and colors.

[Frieder]

Hi, we can't use shapes for the moment: only sizes and colors.

Ok, got it. How about "batch colouring" by attribute, like it is possible for the node labels. I think that would be very good, but have no clue about programing and can't do it myself, so I will not complain further

Freddy

[Frieder]

just found that the partition tool alows me to group by attribute, but if I see it correctly I have to choose between parameters - either modularity class OR some other attribute, is this correct? In this case it would not be useful if I wanted to see if group formation is determined by a certain attribute, or is it possible to group (and color) by one attribute and then, in a second step, run modularity class?

thanks in advance

Freddy

[Frieder]

Help, anyone?

thanks

Freddy

[elijah]

Hi Freddy,

I'm not clear what you're trying to do. If you're trying to color by attribute in modules, then you can first partition by the modularity (using the partition filter) and then color by attributes within the partitioned piece of the network. Or you can partition and sub-partition, using sub-queries, and work with individual pieces of a network. If, however, what you're looking to do is to color different parts of a network using the same set of attributes, then you still have to do that manually, piece by piece.

Could you describe an example of your workflow, then I could give some better advice.

[Frieder]

Hi Freddy,

I'm not clear what you're trying to do. If you're trying to color by attribute in modules, then you can first partition by the modularity (using the partition filter) and then color by attributes within the partitioned piece of the network. Or you can partition and sub-partition, using sub-queries, and work with individual pieces of a network. If, however, what you're looking to do is to color different parts of a network using the same set of attributes, then you still have to do that manually, piece by piece.

Could you describe an example of your workflow, then I could give some better advice.

Hi Elijah,
ok, I'll try to be more clear. Lets say I want to conduct a SNA on a network of foreign and german exchange students (I am making this up...) and I have serveral atribute for each of them such as sex, age, country of origin, time spent in Germany, degree of fluency in English etc. It would be nice to "batch color" them by combinations of these attributes e.g. all German male fluent english speakers in one color, all Chinese female non fluent english speakers in another and so on. It is possible to have a selection of attributes displayed as a label but once you have more than a few plus a lot of nodes it becomes very crowded.
This way it would be imediately visible should group formation be determined by e.g. ethnic background or sex.
I hope this makes more sense now - as indicated in the header I am a newbie...

thanks

Freddy

[eduramiba]

Hi Freddy,
Here is an idea for that. You could combine (merge) the 2 or more columns into a new column using the 'Join with separator' merging strategy.
Then apply partition to the new column. To do the merging, go to Data Laboratory and click on Merge Columns.

Eduardo

[Frieder]

Hi Freddy,
Here is an idea for that. You could combine (merge) the 2 or more columns into a new column using the 'Join with separator' merging strategy.
Then apply partition to the new column. To do the merging, go to Data Laboratory and click on Merge Columns.

Eduardo

Hi Eduardo,

thaks for your helpful reply. That solves part of my problem - namely coloring by a combination of attributes. But when I run modularity and aply it to the graph it "overrides" the partition by attribute. What I want is a visual representation that gives you a quick visual clue as to whether groups form according to a certain attribute or combination of attributes. Do you see my problem? I'll fiddle around some more, maybe I am just dense (always a possibility )

thanks

Freddy

[eduramiba]

Hi,
If I understand your need, maybe you could do the merge including the modularity column, but I am not sure how this will work for you.
For example join column1,column2,modularity and visually see how the coloring including the attributes and modularity in one column matches the groups.

Eduardo