R help - Aug 2021 - Find "undirected" duplicates in a tibble

Hi!

I am working with a large network data consisting of source-target
pairs stored in a tibble. Now I need to transform the directed dataset
to an undirected network data. This means, I need to keep only one
instance for pairs with the same "nodes". In other words, if my data
has one row with A (source) and B (target) and one with B (source) and
A (target), only the pair A-B should be kept.

Here an example how I have solved this problem so far:

--- snip ---

# Create some data
x<-tibble(Source=rep(1:3,4), Target=c(rep(1,3),rep(2,3),rep(3,3),rep(4,3)))
x	# print original data

# Remove "undirected" duplicates
x<-x %>% mutate(pair=mapply(function(x,y)
paste0(sort(c(x,y)),collapse="-"), Source, Target)) %>%
distinct(pair,
.keep_all = T) %>% mutate(Source=sapply(pair, function(x)
unlist(strsplit(x, split="-"))[1]), Target=sapply(pair, function(x)
unlist(strsplit(x, split="-"))[2])) %>% select(-pair)

x	# print cleaned data

--- snip ---

The good thing with my own solution is that it allows the creation of
weighted pairs as well. One just needs to replace 'distinct(pair,
.keep_all=T)' with 'count(pair)'.

I have done a lot of searching but not found any function providing
this functionality. Does someone know an alternative, maybe a more
effective function/solution?

Best,

Kimmo Elo


-- 
Dr. Kimmo Elo
Senior researcher in European Studies
====================================================University of Turku
Centre for Parliamentary Studies
Finland
E-mail: kimmo.elo at utu.fi
=====================================================

Kimmo,

i'll be curious to see other, maybe more elegant, answers.  in the
meantime, this seems to work.

----
  x = data.frame(Source=rep(1:3,4),
Target=c(rep(1,3),rep(2,3),rep(3,3),rep(4,3)))
  y <- apply(x, 1, function(y) return (c(n=min(y), x=max(y))))
  res <- data.frame()
  for (n in unique(y["n",])) {
    unique(y["x",y["n",]==n])
    res <- rbind(res, data.frame(A=c(n),
B=unique(y["x",y["n",]==n])))
  }
  res
----

thanks for the question!

cheers, Greg

Since you are dealing with graphs you could consider using
the igraph package.  This is more involved than needed for
what you are
asking but it might be useful for other follow on calculations.
We first define a 2 column matrix of edges, then convert it to
an igraph and simplify it to remove duplicate edges giving g.
At the end we get an edgelist back.

  library(igraph)
  m <- matrix(c(1, 2, 6, 6, 4, 9, 1, 5, 2, 1, 8, 7, 5, 10, 6, 10), 8, 2)
  g <- m |>
    graph_from_edgelist(directed = FALSE) |>
    simplify()

  plot(g)

  g |>
    get.edgelist() |>
    as.data.frame()



On Fri, Aug 20, 2021 at 5:00 AM Kimmo Elo <kimmo.elo at utu.fi>
wrote:
>
> Hi!
>
> I am working with a large network data consisting of source-target
> pairs stored in a tibble. Now I need to transform the directed dataset
> to an undirected network data. This means, I need to keep only one
> instance for pairs with the same "nodes". In other words, if my
data
> has one row with A (source) and B (target) and one with B (source) and
> A (target), only the pair A-B should be kept.
>
> Here an example how I have solved this problem so far:
>
> --- snip ---
>
> # Create some data
> x<-tibble(Source=rep(1:3,4),
Target=c(rep(1,3),rep(2,3),rep(3,3),rep(4,3)))
> x       # print original data
>
> # Remove "undirected" duplicates
> x<-x %>% mutate(pair=mapply(function(x,y)
> paste0(sort(c(x,y)),collapse="-"), Source, Target)) %>%
distinct(pair,
> .keep_all = T) %>% mutate(Source=sapply(pair, function(x)
> unlist(strsplit(x, split="-"))[1]), Target=sapply(pair,
function(x)
> unlist(strsplit(x, split="-"))[2])) %>% select(-pair)
>
> x       # print cleaned data
>
> --- snip ---
>
> The good thing with my own solution is that it allows the creation of
> weighted pairs as well. One just needs to replace 'distinct(pair,
> .keep_all=T)' with 'count(pair)'.
>
> I have done a lot of searching but not found any function providing
> this functionality. Does someone know an alternative, maybe a more
> effective function/solution?
>
> Best,
>
> Kimmo Elo
>
>
> --
> Dr. Kimmo Elo
> Senior researcher in European Studies
> ====================================================> University of
Turku
> Centre for Parliamentary Studies
> Finland
> E-mail: kimmo.elo at utu.fi
> ====================================================>
______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.


-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com