[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [igraph] Anybody ever use igraph to create regression path diagrams?

From: Gábor Csárdi
Subject: Re: [igraph] Anybody ever use igraph to create regression path diagrams?
Date: Tue, 17 Jul 2012 14:53:50 -0400

I agree. tkplot() is very much just an experiment, a not very
successful one, as I wasn't satisfied with Tk.

igraph might get proper interactive plotting in the future, but only
if someone implements a nice modern cross-platform R package for
developing GUIs.


On Mon, Jul 16, 2012 at 6:53 PM, Bernie Hogan <address@hidden> wrote:
> With all due respect to the igraph team, I think this is a pretty cumbersome 
> process for igraph. If you get your coefficients in some sort of igraph 
> object you can be a lot more flexible in NodeXL (on windows) or gephi. NodeXL 
> is probably the best way to customise a graph to taste with pointing and 
> clicking. Just write the igraph object to graphml. As for storing the 
> arrangement - save the graphml and it should preserve the xy coordinates.
> But if you are a tk hacker by all means try out tamas's suggestions!
> Bernie
> Sent from my iPhone
> On 16 Jul 2012, at 23:26, Tamás Nepusz <address@hidden> wrote:
>> Dear Prof. Johnson,
>> I'm not the developer of the R interface but I thought I'd answer anyway and 
>> let Gabor correct or extend what I say.
>>> 1. I need to have relatively complicated R objects that are
>>> represented by the igraph nodes. For example, a variable may be it
>>> 'ordinary' format X, or it may be log(x), or poly(x,d), or so forth.
>>> I'm thinking of still calling it X in the graph, but there would be
>>> some right-click magic to allow the user to ask for transformations.
>>> Is it feasible to add more variables inside a node object. Possible?
>>> Can you point me at the right starting point or example?
>> Internally, igraph treats the nodes as integers, but it allows you to attach 
>> arbitrary R objects as attributes to the nodes (and also to the edges). So, 
>> you can probably store all the extra information you need in an 
>> appropriately constructed R data type and then attach this to the node as a 
>> "data" attribute or something like that. For instance, suppose that you want 
>> to store the "type" of a node and the type is one of "ordinary", "log" or 
>> "exp". You can then do something like this:
>> V(g)$type <- "ordinary"          # assigns "ordinary" to the type of each 
>> node
>> V(g)[2]$type <- "exp"             # sets the type of node 2 to "exp"
>> V(g)$type                      # returns the types of each vertex
>> As for the right-click magic: tkplot does not allow the customization of the 
>> vertex popup menu yet, but it is probably not too difficult by taking the 
>> source code of tkplot() as a template, copying it, and then modifying it as 
>> needed (look for the lines starting with tkadd(vertex.popup.menu…)).
>>> 2. Suppose a graph exists, and I view and re-arrange it with tkrplot.
>>> Is there a way to make the re-arranged objects take notice of the
>>> changes so that they are recorded permanently.
>> Again, this can be achieved by modifying the tkplot() function. tkplot() 
>> attaches event handlers to several mouse events of the drawing canvas; for 
>> instance, the line starting with tkitembind(canvas, tobind, "<B1-Motion>", 
>> …) attaches an event handler to the "<B1-Motion>" event, which is fired when 
>> the user moves the mouse while button 1 of the mouse is pressed. This 
>> corresponds to moving the selected vertices. Adding your own code into the 
>> body of the event handler function allows you to respond to the 
>> re-arrangement of the nodes.
>> Note that tkplot() does not support adding new vertices/edges or removing 
>> existing ones (as far as I know), so if the re-arrangements you wish to 
>> perform on the graphical representation of the models includes 
>> adding/removing vertices or edges, then this definitely needs more coding.
>> All the best,
>> Tamas
>> _______________________________________________
>> igraph-help mailing list
>> address@hidden
>> https://lists.nongnu.org/mailman/listinfo/igraph-help
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help

Gabor Csardi <address@hidden>     MTA KFKI RMKI

reply via email to

[Prev in Thread] Current Thread [Next in Thread]