Adjacency Matrix from a dataframe

user_7472392 0
Viewed 185

I am trying to convert an edgelist to an adjacent matrix.

Below is the sample data

#Sample Data
User<-c("1","1","2","3","4")  
v1 <- c("b", "b", "a", "d", "c")
v2 <- c("c", "d", "c", "a", "a")
v3 <- c(0, 0, "d", 0, "b")
v4 <- c(0, 0, 0, 0, 0)
v5 <- c(0, 0, 0, 0, 0)

my_data<-data.frame(User, v1, v2, v3, v4, v5)
my_data

If you run this code you will get the below as output,

  User v1 v2 v3 v4 v5
    1  b  c  0  0  0
    1  b  d  0  0  0
    2  a  c  d  0  0
    3  d  a  0  0  0
    4  c  a  b  0  0

Using the data, I want to create an adjacent matrix that looks like follows:

   a  b  c  d
a  0  0  2  2        
b  0  0  1  1
c  2  1  0  1  
d  2  1  1  0

Basically, the desired output diplays the count how many times each pair appeared in column v1~v5 in the sample data frame.

I have tried to use AdjacencyFromEdgelist function from dils library, also tried to create a matrix shell with NAs and fill out the matrix by looping through the dataframe.

However, I could not get neither way to work.

adjacency-matrix edge-list igraph r
3 Answers

I think this may be close to what you have in mind. In the rows where there are more than 2 vertices, I considered every existing pairs:

library(igraph)

do.call(rbind, my_data[-1] |>
          apply(1, \(x) x[x != 0]) |>
          lapply(\(x) t(combn(x, m = 2)))) |>
  graph_from_edgelist(directed = FALSE) %>%
  as_adjacency_matrix()

4 x 4 sparse Matrix of class "dgCMatrix"
  b c d a
b . 2 1 1
c 2 . 1 2
d 1 1 . 2
a 1 2 2 .

Or without the pip operator in base R:

tmp <- apply(my_data[-1], 1, function(x) x[x != 0])
tmp <- do.call(rbind, lapply(tmp, function(x) t(combn(x, m = 2))))

my_graph <- graph_from_edgelist(tmp, directed = FALSE)
adj_mat <- as_adjacency_matrix(my_graph)
adj_mat
user_143145200

Another attempt, minus the need to calculate all the combinations with combn

sel <- my_data[-1] != 0
dat <- data.frame(row=row(my_data[-1])[sel], value = my_data[-1][sel])
out <- crossprod(table(dat))
diag(out) <- 0
out

#     value
#value a b c d
#    a 0 1 2 2
#    b 1 0 2 1
#    c 2 2 0 1
#    d 2 1 1 0

Matches the result from @AnoushiravanR:

adj_mat[c("a","b","c","d"), c("a","b","c","d")]
#4 x 4 sparse Matrix of class "dgCMatrix"
#  a b c d
#a . 1 2 2
#b 1 . 2 1
#c 2 2 . 1
#d 2 1 1 .
user_4968030

Another igraph option

do.call(
  rbind,
  combn(df, 2, setNames, nm = c("from", "to"), simplify = FALSE)
) %>%
  filter(from > 0 & to > 0) %>%
  arrange(from) %>%
  graph_from_data_frame(directed = FALSE) %>%
  get.adjacency(sparse = FALSE)

gives

  a b c d
a 0 1 2 2
b 1 0 2 1
c 2 2 0 1
d 2 1 1 0
user_121587570
Related