Re: [igraph] igraph-help Digest, Vol 74, Issue 22

Send igraph-help mailing list submissions to
    address@hidden

To subscribe or unsubscribe via the World Wide Web, visit
    https://lists.nongnu.org/mailman/listinfo/igraph-help
or, via email, send a message with subject or body 'help' to
    address@hidden

You can reach the person managing the list at
    address@hidden

When replying, please edit your Subject line so it is more specific
than "Re: Contents of igraph-help digest..."

Today's Topics:

   1. Re: degree according to vertex attributes (Tam?s Nepusz)
   2. Re: Choosing between different methods of detecting
communities (??)
   3. Re: degree according to vertex attributes (Dominik Santner (CRIE))
   4. Re: Choosing between different methods of detecting
communities (G?bor Cs?rdi)
   5. Re: Choosing between different methods of detecting
communities (??)
   6. Re: Choosing between different methods of detecting
communities (G?bor Cs?rdi)

----------------------------------------------------------------------

Message: 1
Date: Fri, 28 Sep 2012 23:44:19 +0200
From: Tam?s Nepusz <address@hidden>
To: address@hidden, Help for igraph users
    <address@hidden>
Subject: Re: [igraph] degree according to vertex attributes
Message-ID: <address@hidden>
Content-Type: text/plain; charset="utf-8"

Hello,

Suppose that the city is encoded in the "city" vertex attribute of the graph; for instance:

> g <- grg.game(100, 0.2)
> V(g)$city <- LETTERS[(0:99 %/% 10) + 1]

First, we get the edge list of the graph:

> el <- get.edgelist(g)

Next, we select those edges (i.e. rows) from the edge list where the endpoints are in the same city:

> internal.edges <- which(V(g)$city[el[,1]] == V(g)$city[el[,2]])

We then create a subgraph which consists of these edges only:

> internal.graph <- subgraph.edges(g, internal.edges)

degree(internal.graph) will then give you how many friends each person has in the same city, and degree(g)-degree(internal.graph) will give you how many friends each person has in different cities.

Best,
--
T.

On Friday, 28 September 2012 at 15:21, Dominik Santner (CRIE) wrote:

> Hello,
>
> I'm using igraph with R and I have an undirected graph including vertex
> attributes (for example the city of residence). I want to calculate the
> degree separately for connections to actors living in the same city and
> those living in any other city.
>
> Example: John lives in Berlin and has 10 friends (degree of 10). 3 of
> his friends live in Berlin, 3 in Hamburg, 2 in New York and 2 in Munich.
> I need a result telling me that John has 3 friends in the same city and
> 7 in any other city.
>
> Thanks
>
> Dominik
>
> _______________________________________________
> igraph-help mailing list
> address@hidden (mailto:address@hidden)
> https://lists.nongnu.org/mailman/listinfo/igraph-help

------------------------------

Message: 2
Date: Sat, 29 Sep 2012 16:23:46 +0800
From: ?? <address@hidden>
To: Help for igraph users <address@hidden>
Subject: Re: [igraph] Choosing between different methods of detecting
    communities
Message-ID:
    <CAKSeTcE6NGbd30cMTQT6Cy6R53S+address@hidden>
Content-Type: text/plain; charset="utf-8"

Hi Gabor,

Thanks for your reply.
I know that the edge weights are used to calculate the shortest paths for
edge betweenness.
While I don't know that what the edge weights represent in the library.
Specifically, is it the higher the weight, the longer the edge, or inverse?
Namely, the edge weight denotes the distance between the connected nodes or
the similarity between the connected nodes.

Regards,
chen

On Wed, Sep 26, 2012 at 9:07 PM, G?bor Cs?rdi <address@hidden> wrote:

> Hi,
>
> weighted edge-betweenness community detection simply means that
> weighted paths are considered for calculating the shortest paths in
> the edge betweenness calculation.
>
> Best,
> Gabor
>
> On Wed, Sep 26, 2012 at 6:54 AM, ?? <address@hidden> wrote:
> > Hi,
> >
> > Have you used the Weighted Girvan-Newman community detected algorithm in
> the
> > library to detect community?
> > I am not sure what the edge weight represent in this algorithm in Igraph,
> > similarity or distance?
> >
> > Thanks,
> > chen
> >
> > On Wed, Sep 26, 2012 at 6:48 PM, Minh Nguyen <address@hidden>
> wrote:
> >>
> >> Hi,
> >>
> >> On Wed, Sep 26, 2012 at 8:21 PM, Roey Angel <address@hidden>
> >> wrote:
> >> > I was wondering if there's a way to choose between the methods, either
> >> > by
> >> > experience (e.g. 'method x typically gives best results) or with some
> >> > statistical test.
> >>
> >> There is a general class of statistical tests called the significance
> >> of community. The relevant ideas and techniques can be found in the
> >> following papers.
> >>
> >> http://dx.doi.org/10.1371/journal.pone.0033721
> >> http://dx.doi.org/10.1103/PhysRevE.82.066106
> >> http://dx.doi.org/10.1103/PhysRevE.81.046110
> >>
> >> --
> >> Regards,
> >> Minh Van Nguyen
> >> http://bit.ly/mvngu
> >>
> >> _______________________________________________
> >> 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
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nongnu.org/archive/html/igraph-help/attachments/20120929/7570e795/attachment.html>

------------------------------

Message: 3
Date: Sat, 29 Sep 2012 11:56:15 +0200
From: "Dominik Santner (CRIE)" <address@hidden>
To: Tam?s Nepusz <address@hidden>
Cc: Help for igraph users <address@hidden>
Subject: Re: [igraph] degree according to vertex attributes
Message-ID: <address@hidden>
Content-Type: text/plain; charset=UTF-8; format=flowed

Hello,

thank you very much for your suggestion. Unfortunately there are two
problems:

first:
when using your example graph created by

> g <- grg.game(100, 0.2)
> V(g)$city <- LETTERS[(0:99 %/% 10) + 1]

I recieve a warning when doing

> degree(g)-degree(internal.graph)

telling me that the two objects are not of the same length

> length(degree(internal.graph))
[1] 96
> length(degree(g))
[1] 100

the second problem occures when applying this procedure to my data:

in your example you use V(g)$city when doing

> internal.edges <- which(V(g)$city[el[,1]] == V(g)$city[el[,2]])

whith the result:
> internal.edges
   [1]   1   5   6   7   8 ...

when checking for
> V(g)$city
   [1] "A" "A" "A" "A" "A" ...

> V(g)$city[el[,2]]
   [1] "A" "B" "B" "B" "A" "A" ...

everything is fine.

But with my dataset this happens:

> internal.edges <- which(V(xxx)$st_AMR[el[,1]] == V(xxx)$st_AMR[el[,2]])
> internal.edges
integer(0)

> V(xxx)$st_AMR
   [1] "M?nchen" "Darmstadt" "Regensburg"    "Halle" ...

> V(xxx)$st_AMR[el[,2]]
   [1] NA NA NA NA NA ...

Maybe it has something to do with my vertex labels and/or my edge list, but I have no idea what it might be:

The edge list of your example:

> head(el)
[,1] [,2]
[1,] 1 7
[2,] 1   11
[3,] 1   12
[4,] 1   13
[5,] 2 3
[6,] 2 4

The edge list of my dataset:

> head(el)
[,1]
[1,] "Company A (6411000)"
[2,] "Company A (6411000)"
[3,] "Company A (6411000)"
[4,] "Company A (6411000)"
[5,] "Company A (6411000)"
[6,] "Company A (6411000)"
[,2]
[1,] "University A (8416041)"
[2,] "University B (8416041)"
[3,] "Company B (5562024)"
[4,] "Company C (5358024)"
[5,] "Company D (5358024)"
[6,] "Company E (8215101)"

Best regards
Dominik

Am 28.09.2012 23:44, schrieb Tam?s Nepusz:
> Hello,
>
> Suppose that the city is encoded in the "city" vertex attribute of the graph; for instance:
>
>> g <- grg.game(100, 0.2)
>> V(g)$city <- LETTERS[(0:99 %/% 10) + 1]
> First, we get the edge list of the graph:
>
>> el <- get.edgelist(g)
> Next, we select those edges (i.e. rows) from the edge list where the endpoints are in the same city:
>
>> internal.edges <- which(V(g)$city[el[,1]] == V(g)$city[el[,2]])
> We then create a subgraph which consists of these edges only:
>
>> internal.graph <- subgraph.edges(g, internal.edges)
> degree(internal.graph) will then give you how many friends each person has in the same city, and degree(g)-degree(internal.graph) will give you how many friends each person has in different cities.
>
> Best,

------------------------------

Message: 4
Date: Sat, 29 Sep 2012 08:23:51 -0400
From: G?bor Cs?rdi <address@hidden>
To: Help for igraph users <address@hidden>
Subject: Re: [igraph] Choosing between different methods of detecting
    communities
Message-ID:
    <CABtg=K=ym=4xQ4Ty=address@hidden>
Content-Type: text/plain; charset=UTF-8

Hi Chen,

On Sat, Sep 29, 2012 at 4:23 AM, ?? <address@hidden> wrote:
> Hi Gabor,
>
> Thanks for your reply.
> I know that the edge weights are used to calculate the shortest paths for
> edge betweenness.
> While I don't know that what the edge weights represent in the library.

there is no semantics forced by igraph. Just like your vertices can
represent anything, your edge weights can represent anything you want,
similarity or capacity or distance, etc.

> Specifically, is it the higher the weight, the longer the edge, or inverse?
> Namely, the edge weight denotes the distance between the connected nodes or
> the similarity between the connected nodes.

Whichever you prefer. But note that shortest.paths() (and similarly
edge betweenness calculation) really calculates shortest weighted
paths, so you might need transform your weights for these functions.

Gabor

> Regards,
> chen
>
> On Wed, Sep 26, 2012 at 9:07 PM, G?bor Cs?rdi <address@hidden> wrote:
>>
>> Hi,
>>
>> weighted edge-betweenness community detection simply means that
>> weighted paths are considered for calculating the shortest paths in
>> the edge betweenness calculation.
>>
>> Best,
>> Gabor
>>
>> On Wed, Sep 26, 2012 at 6:54 AM, ?? <address@hidden> wrote:
>> > Hi,
>> >
>> > Have you used the Weighted Girvan-Newman community detected algorithm in
>> > the
>> > library to detect community?
>> > I am not sure what the edge weight represent in this algorithm in
>> > Igraph,
>> > similarity or distance?
>> >
>> > Thanks,
>> > chen
>> >
>> > On Wed, Sep 26, 2012 at 6:48 PM, Minh Nguyen <address@hidden>
>> > wrote:
>> >>
>> >> Hi,
>> >>
>> >> On Wed, Sep 26, 2012 at 8:21 PM, Roey Angel <address@hidden>
>> >> wrote:
>> >> > I was wondering if there's a way to choose between the methods,
>> >> > either
>> >> > by
>> >> > experience (e.g. 'method x typically gives best results) or with some
>> >> > statistical test.
>> >>
>> >> There is a general class of statistical tests called the significance
>> >> of community. The relevant ideas and techniques can be found in the
>> >> following papers.
>> >>
>> >> http://dx.doi.org/10.1371/journal.pone.0033721
>> >> http://dx.doi.org/10.1103/PhysRevE.82.066106
>> >> http://dx.doi.org/10.1103/PhysRevE.81.046110
>> >>
>> >> --
>> >> Regards,
>> >> Minh Van Nguyen
>> >> http://bit.ly/mvngu
>> >>
>> >> _______________________________________________
>> >> 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
>>
>> _______________________________________________
>> 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

------------------------------

Message: 5
Date: Sat, 29 Sep 2012 20:32:12 +0800
From: ?? <address@hidden>
To: Help for igraph users <address@hidden>
Subject: Re: [igraph] Choosing between different methods of detecting
    communities
Message-ID:
    <CAKSeTcGs=address@hidden>
Content-Type: text/plain; charset="utf-8"

Hi Gabor,

>>But note that shortest.paths() (and similarly
>>edge betweenness calculation) really calculates shortest weighted
>>paths, so you might need transform your weights for these functions.

OK, so that if I use similarity of nodes as edge weights, I need to
transform the weights (i.e., inverse) for detecting community.
is that right?

Thanks & Regards
chen

On Sat, Sep 29, 2012 at 8:23 PM, G?bor Cs?rdi <address@hidden> wrote:

> Hi Chen,
>
> On Sat, Sep 29, 2012 at 4:23 AM, ?? <address@hidden> wrote:
> > Hi Gabor,
> >
> > Thanks for your reply.
> > I know that the edge weights are used to calculate the shortest paths for
> > edge betweenness.
> > While I don't know that what the edge weights represent in the library.
>
> there is no semantics forced by igraph. Just like your vertices can
> represent anything, your edge weights can represent anything you want,
> similarity or capacity or distance, etc.
>
> > Specifically, is it the higher the weight, the longer the edge, or
> inverse?
> > Namely, the edge weight denotes the distance between the connected nodes
> or
> > the similarity between the connected nodes.
>
> Whichever you prefer. But note that shortest.paths() (and similarly
> edge betweenness calculation) really calculates shortest weighted
> paths, so you might need transform your weights for these functions.
>
> Gabor
>
> > Regards,
> > chen
> >
> > On Wed, Sep 26, 2012 at 9:07 PM, G?bor Cs?rdi <address@hidden>
> wrote:
> >>
> >> Hi,
> >>
> >> weighted edge-betweenness community detection simply means that
> >> weighted paths are considered for calculating the shortest paths in
> >> the edge betweenness calculation.
> >>
> >> Best,
> >> Gabor
> >>
> >> On Wed, Sep 26, 2012 at 6:54 AM, ?? <address@hidden> wrote:
> >> > Hi,
> >> >
> >> > Have you used the Weighted Girvan-Newman community detected algorithm
> in
> >> > the
> >> > library to detect community?
> >> > I am not sure what the edge weight represent in this algorithm in
> >> > Igraph,
> >> > similarity or distance?
> >> >
> >> > Thanks,
> >> > chen
> >> >
> >> > On Wed, Sep 26, 2012 at 6:48 PM, Minh Nguyen <address@hidden>
> >> > wrote:
> >> >>
> >> >> Hi,
> >> >>
> >> >> On Wed, Sep 26, 2012 at 8:21 PM, Roey Angel <
> address@hidden>
> >> >> wrote:
> >> >> > I was wondering if there's a way to choose between the methods,
> >> >> > either
> >> >> > by
> >> >> > experience (e.g. 'method x typically gives best results) or with
> some
> >> >> > statistical test.
> >> >>
> >> >> There is a general class of statistical tests called the significance
> >> >> of community. The relevant ideas and techniques can be found in the
> >> >> following papers.
> >> >>
> >> >> http://dx.doi.org/10.1371/journal.pone.0033721
> >> >> http://dx.doi.org/10.1103/PhysRevE.82.066106
> >> >> http://dx.doi.org/10.1103/PhysRevE.81.046110
> >> >>
> >> >> --
> >> >> Regards,
> >> >> Minh Van Nguyen
> >> >> http://bit.ly/mvngu
> >> >>
> >> >> _______________________________________________
> >> >> 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
> >>
> >> _______________________________________________
> >> 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
>
> _______________________________________________
> igraph-help mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/igraph-help
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nongnu.org/archive/html/igraph-help/attachments/20120929/3367f0ed/attachment.html>

------------------------------

Message: 6
Date: Sat, 29 Sep 2012 09:06:02 -0400
From: G?bor Cs?rdi <address@hidden>
To: Help for igraph users <address@hidden>
Subject: Re: [igraph] Choosing between different methods of detecting
    communities
Message-ID:
    <CABtg=K=mH=0fOnspc=address@hidden>
Content-Type: text/plain; charset=UTF-8

On Sat, Sep 29, 2012 at 8:32 AM, ?? <address@hidden> wrote:
> OK, so that if I use similarity of nodes as edge weights, I need to
> transform the weights (i.e., inverse) for detecting community.
> is that right?

I think so, because when you calculate shortest paths, you want edge
weights to represent distances.

Gabor

[...]

------------------------------

_______________________________________________
igraph-help mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/igraph-help

End of igraph-help Digest, Vol 74, Issue 22
*******************************************

From:	leandro mattos
Subject:	Re: [igraph] igraph-help Digest, Vol 74, Issue 22
Date:	Sat, 29 Sep 2012 19:35:34 -0700 (PDT)