Re: [igraph] Correlation based on the attributes of vertices

[Gábor Csárdi]

I think what you need is a contingency table, exactly as you say, and
then a chi-square or Fisher-test, or maybe logistic regression.

Best,
Gabor

On Mon, Oct 10, 2011 at 7:38 AM, S.M. Ali Abbas <address@hidden> wrote:
> Thank you very much! It does work.
> I was thinking that perhaps this correlation on attribute list won't be
> relevant to my problem, as, for instance, dorm values are mere integer
> labels of random nature. Corr(x,y) does not take the variation in y values
> but also of x. If I get it right, it won't work for labels then. I should
> rather make a matrix for all the unique dorms m as (m x m matrix), and then
> populate each cell with the number of connections between them. After that,
> I should calculate correlation over it.
>
>
> On Wed, Oct 5, 2011 at 5:00 PM, Gábor Csárdi <address@hidden> wrote:
>>
>> Hi,
>>
>> On Mon, Oct 3, 2011 at 1:34 PM, S.M. Ali Abbas <address@hidden>
>> wrote:
>> > Hi,
>> > I have a directed graph which has some vertices attributes (like dorm,
>> > year
>> > etc). Edges and the graph itself do not have any attributes. Based on
>> > the
>> > attributes of the vertices, I'd like to calculate correlation among the
>> > edges (e.g. how likely people of the same dorm are connected?) for the
>> > whole
>> > graph. Also, I'd like to calculate inter-attributes correlation for the
>> > whole graph (how correlated dorm and year attributes are?)
>> > Could you kindly tell how to go about it?
>>
>> el <- get.edgelist(g, names=FALSE)
>> dorm <- matrix(nc=2, get.vertex.attribute(g, "dorm", el))
>>
>> gives you the "dorm" for the vertices along each edge.
>>
>> > I thought of populating a list just like the graph edge list, and then
>> > replacing each source and destination by its attribute value. For
>> > instance,
>> > instead of the edge (0->1), I will replace it by (dorm_valueOf(0) ->
>> > dorm_valueOf(1)) and then run the function cor over it. It does not seem
>> > like a nice solution.
>>
>> Why not? Basically this is what I did above. You cannot run cor() on
>> this, though, cor() is for numeric data.
>>
>> > On a side note, how does one get source and destination out of an edge
>> > list
>> > by the Edge Iterator?
>>
>> See get.edge()
>>
>> Cheers,
>> Gabor
>>
>> > For example, I'd like to know the source and the
>> > target vertices' index of the first edge, E(graph)[0]. How can I extract
>> > this information?
>> >
>> > Thanks!
>> >
>> > Cheers,
>> > Ali
>> >
>> >
>> >
>> > _______________________________________________
>> > igraph-help mailing list
>> > address@hidden
>> > https://lists.nongnu.org/mailman/listinfo/igraph-help
>> >
>> >
>>
>>
>>
>> --
>> Gabor Csardi <address@hidden>     MTA KFKI RMKI
>>
>> _______________________________________________
>> 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