CODEWITHSUNDEEP
rnetwork-programming
user1021713
user1021713

Reputation: 2203

clustering of nodes of network in r without knowing their distances

I have a network. In which the nodes are connected like this :

A-B
B-C
A-C
B-D
E-F
F-G
H-G
H-E

I want them to be clustered like ABCD and EFGH. Is there a way to do it in R??

Upvotes: 0

Views: 985

Answers (2)

Tim P
Tim P

Reputation: 1383

Thought I'd bang together a direct solution to provide a useful comparison (and practice my coding - all pointers welcome...)

## initial link data frame (with start and end points)
link=data.frame(st=c("A","B","A","B","E","F","H","H"),
                end=c("B","C","C","D","F","G","G","E"),
                stringsAsFactors=FALSE)

## form blank list where clusters will build up (works up 
## to n=26!)
n=nrow(link)
L=c(); for (j in seq_len(n)) {L=c(L,list(NULL))}
names(L)=LETTERS[seq_len(n)]

## for each link in turn, pull together a collection of
## everything in the same cluster as either end, and update
## all relevant entries in L
for (j in seq_len(n)) {
    clus=sort(unique(c(link$st[j],link$end[j],
                       L[[link$st[j]]],L[[link$end[j]]])))
    for (k in clus) {L[[k]]=clus}
}
print(L)

The output is, as expected:

$A
[1] "A" "B" "C" "D"

$B
[1] "A" "B" "C" "D"

$C
[1] "A" "B" "C" "D"

$D
[1] "A" "B" "C" "D"

$E
[1] "E" "F" "G" "H"

$F
[1] "E" "F" "G" "H"

$G
[1] "E" "F" "G" "H"

$H
[1] "E" "F" "G" "H"

Upvotes: 1

Thomas Jungblut
Thomas Jungblut

Reputation: 20969

That is called "Connected Components" of a graph and is present in the igraph package.

Here is the doc:

http://igraph.sourceforge.net/doc/R/clusters.html

Upvotes: 3

Related Questions