CODEWITHSUNDEEP
rdplyrtidyversetibble
acvill
acvill

Reputation: 445

group tibble rows by unordered combination of columns

Given the following tibble

tibble(sample = c(1:6),
       string = c("ABC","ABC","CBA","FED","DEF","DEF"),
       x = c("a","a","b","e","d","d"),
       y = c("b","b","a","d","e","e"))

# A tibble: 6 × 4
  sample string x     y    
   <int> <chr>  <chr> <chr>
1      1 ABC    a     b    
2      2 ABC    a     b    
3      3 CBA    b     a    
4      4 FED    e     d    
5      5 DEF    d     e    
6      6 DEF    d     e

I would like to group rows by the unordered combination of columns x,y, then flip x ⇔ y and reverse string in instances where x,y is inverted relative to the first row in the group. The desired output:

# A tibble: 6 × 5
  sample string x     y     group
   <int> <chr>  <chr> <chr> <dbl>
1      1 ABC    a     b         1
2      2 ABC    a     b         1
3      3 ABC    a     b         1
4      4 FED    e     d         2
5      5 FED    e     d         2
6      6 FED    e     d         2

Upvotes: 2

Views: 58

Answers (1)

Ma&#235;l
Ma&#235;l

Reputation: 51894

strSort <- function(x) sapply(lapply(strsplit(x, NULL), sort), paste, collapse="")

dat %>% 
  group_by(group = data.table::rleid(strSort(string))) %>% 
  mutate(across(string:y, first))

# A tibble: 6 x 5
# Groups:   group [2]
  sample string x     y     group
   <int> <chr>  <chr> <chr> <int>
1      1 ABC    a     b         1
2      2 ABC    a     b         1
3      3 ABC    a     b         1
4      4 FED    e     d         2
5      5 FED    e     d         2
6      6 FED    e     d         2

Previous answer

Here's a way using both tidyverse and apply approaches. First, sort the rows of columns x and y, then group_by x and y, create a cur_group_id and stri_reverse if necessary.

library(tidyverse)
library(stringi)

#Sort by row
dat[, c("x", "y")] <- t(apply(dat[, c("x", "y")], 1, sort))

dat %>% 
  group_by(x, y) %>% 
  mutate(group = cur_group_id(),
         string = ifelse(str_sub(string, 1, 1) == toupper(x), string, stri_reverse(string)))

# A tibble: 6 x 5
# Groups:   x, y [2]
  sample string x     y     group
   <int> <chr>  <chr> <chr> <int>
1      1 ABC    a     b         1
2      2 ABC    a     b         1
3      3 ABC    a     b         1
4      4 DEF    d     e         2
5      5 DEF    d     e         2
6      6 DEF    d     e         2

Upvotes: 2

Related Questions