Search code examples
rdplyrgroup-by

How to get the unique combined groups in R?

I am doing grouping for some data, like entity data. I have found the groups based on some entity attributes, like this:

df <- data.frame(uniq_index.x = c(1426, 1426, 1426, 1426, 7796, 7796, 7796, 7796, 
                                  7159, 7159, 7159, 7159, 7857, 7857, 7857, 7857,
                                  7158, 7158, 7158, 7158, 5440, 9861, 1641, 8685,
                                  1644, 7525, 6030, 5672), 
                 uniq_index.y = c(7796, 7159, 7857, 7158, 1426, 7159, 7857, 7158,
                                  1426, 7796, 7857, 7158, 1426, 7796, 7159, 7158,
                                  1426, 7796, 7159, 7857, 9861, 5440, 8685, 1641,
                                  7525, 1644, 5673, 6030)
                 )

# grouping
a <- df %>% 
  group_by(uniq_index.x) %>% 
  group_split

From the above data, "1426", "7796", "7159", "7877" and "7158" should be in the same group; 5672, 5673 and 6030 should be in another group. I can use group_by and group_split to get groups.

However, giving there are duplicated groups, I used the following code to get the unique groups:

# initial an empty dataframe
b <- data.frame(V1 = character())
# loop through a (which is obtained from group_split)
for (i in 1:length(a)) {
  x <- a[[i]][,1]
  y <- a[[i]][,2]
  x <- x %>% 
    mutate(uniq_index = uniq_index.x) %>% 
    select(uniq_index)
  y <- y %>% 
    mutate(uniq_index = uniq_index.y) %>%
    select(uniq_index)
  z <- unique(x) %>% 
    rbind(y) %>% 
    arrange(uniq_index)
  b <- b %>% 
    rbind(paste(z))
}
  
# unique groups
b <- b %>% 
  unique() %>% 
  mutate(
    uniq_agency_id = 100000 + 1:nrow(unique(b))
  )

Then, I noticed this issue:

enter image description here

Similar to the sample data that (6030, 5672) and (5673, 6030) are two separate groups. These two groups should be in one big group.

I am struggling to think of a logic to obtained the combine unique groups.

Solution

  • The solution to this is everywhere in this website. Here is one way using igraph:

    igraph::components(igraph::graph_from_data_frame(df))$membership
1426 7796 7159 7857 7158 5440 9861 1641 8685 1644 7525 6030 5672 5673 
   1    1    1    1    1    2    2    3    3    4    4    5    5    5