Re: [igraph] clustering coefficients for bipartite networks

[Tamás Nepusz]

Hi Simone,

To speed up things, I'd probably create the list of neighbor sets in advance:

neisets = [set(g.neighbors(i)) for i in xrange(g.vcount())]

After that, you can do this:

vs = g.vs(type=0)
[len(neisets[u] & neisets[v]) / len(neisets[u] | neisets[v]) for u in vs for v 
in vs]

Of course make sure that you do float division and not integer division; either 
run from __future__ import division beforehand or convert one of the lengths to 
float.

-- 
Tamas

On 1 Feb 2011, at 17:40, Simone Gabbriellini wrote:

> I guess I have to refine this description: I need to find, for a node u, the 
> average of clustering values that u has with all the neighbors v which u 
> share some neighbors with, where the clustering value is computed as the 
> number of same neighbors between u and v divided by the total number of 
> unique neighbors of u and v.
> 
> is this something appreciable in that direction?
> 
> [ ( len( set( g.neighbors(u) ) & set( g.neighbors(v) ) ) / len( list( set( 
> g.neighbors(u) + g.neighbors(v) ) ) ) ) for u in g.vs(type=0) for v in 
> g.vs(type=0)]
> 
> best,
> simone
> 
> Il giorno 01/feb/2011, alle ore 16.08, Simone Gabbriellini ha scritto:
> 
>> I am really facing difficulties in translating the code from R... Maybe, it 
>> would be better to start from scratch in python.
>> 
>> I am trying to find, for every pair (u,v) of nodes of the same set, how many 
>> neighbors of the other sets they have in common divided by the  sum of 
>> unique neighbors of u and v. 
>> 
>> is there a simple way to accomplish this task?
>> 
>> If I am right, this should produce, for every node, a list of values as long 
>> as the number of pairs of nodes in a set. I then can use a running mean to 
>> calculate the average value of the list.
>> 
>> best,
>> Simone
>> 
>> Il giorno 29/gen/2011, alle ore 11.50, Tamás Nepusz ha scritto:
>> 
>>>> I don't think I can define functions on the fly like in R (but maybe I am 
>>>> wrong)
>>> You can, see the lambda keyword:
>>> 
>>> http://diveintopython.org/power_of_introspection/lambda_functions.html
>>> 
>>> You can also define functions within functions, so if your auxiliary 
>>> function is longer (but you won't use it from anywhere else), you can 
>>> define it inside your original function.
>>> 
>>> I don't know exactly what the difference is between the different *apply 
>>> functions in R (lapply, sapply, apply), but I believe that Python's map() 
>>> function does something similar. But even better, use list comprehensions 
>>> if possible:
>>> 
>>> http://docs.python.org/tutorial/datastructures.html#list-comprehensions
>>> 
>>> -- 
>>> T.
>>> 
>>> 
>>> _______________________________________________
>>> igraph-help mailing list
>>> address@hidden
>>> http://lists.nongnu.org/mailman/listinfo/igraph-help
>> 
> 
> 
> _______________________________________________
> igraph-help mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/igraph-help
>