Make a named list of pairwise locations of a dataframe rows in R-CodePudding

In my data below, I was wondering how to make a named list of pairwise locations (i.e., row number) of a dataframe rows in R?

Specifically, I want to do pairwise comparison of every 3 rows. For example, in my Desired_output, you'll see that I have manually done so for the first 6 rows.

For the first 3 rows in my Desired_output, I compare third row - first row whose row numbers are c(3, - 1). Then, I compare third row - second row whose row numbers are c(3, - 2). And then, I compare second row - first row whose row numbers are c(2, - 1).

Can we possibly automate this to obtain my desired output below?

Desired_output =

list("4.100 - 1.700 (c2s,complex,t1)"=c(3,-1),
     "4.100 - 2.900 (c2s,complex,t1)"=c(3,-2),
     "2.900 - 1.700 (c2s,complex,t1)"=c(2,-1),
     
     "4.100 - 1.700 (s2c,complex,t2)"=c(6,-4),
     "4.100 - 2.900 (s2c,complex,t2)"=c(6,-5),
     "2.900 - 1.700 (s2c,complex,t2)"=c(5,-4),
     
     .
     .
     .
)

data = read.table(h=TRUE, text = "
     Measure    motiv   t   ord task_df 
1   ac_EFC/C    1.700   1   c2s complex 
2   ac_EFC/C    2.900   1   c2s complex 
3   ac_EFC/C    4.100   1   c2s complex 

4   ac_EFC/C    1.700   2   s2c complex 
5   ac_EFC/C    2.900   2   s2c complex 
6   ac_EFC/C    4.100   2   s2c complex 

7   ac_EFC/C    1.700   3   c2s complex 
8   ac_EFC/C    2.900   3   c2s complex 
9   ac_EFC/C    4.100   3   c2s complex 

10  ac_EFC/C    1.700   4   s2c complex 
11  ac_EFC/C    2.900   4   s2c complex 
12  ac_EFC/C    4.100   4   s2c complex 

13  ac_EFC/C    1.700   1   s2c simple  
14  ac_EFC/C    2.900   1   s2c simple  
15  ac_EFC/C    4.100   1   s2c simple  
    
16  ac_EFC/C    1.700   2   c2s simple  
17  ac_EFC/C    2.900   2   c2s simple  
18  ac_EFC/C    4.100   2   c2s simple  

19  ac_EFC/C    1.700   3   s2c simple  
20  ac_EFC/C    2.900   3   s2c simple  
21  ac_EFC/C    4.100   3   s2c simple  

22  ac_EFC/C    1.700   4   c2s simple  
23  ac_EFC/C    2.900   4   c2s simple  
24  ac_EFC/C    4.100   4   c2s simple")

CodePudding user response：

We may use combn after grouping - create a sequence column rn, then concatenate the 'ord', 'task_df', 't' columns to create a grouping column 'grp', grouped by the column, apply combn on the 'motiv' to return a pairwise combination of elements, paste (str_c) them, as well as create a pairwise index from the 'rn', created, unite the 'val', 'grp' columns and the set the names of the list column 'ind' with 'val'

library(dplyr)
library(stringr)
library(tidyr)
data %>% 
  mutate(rn = row_number()) %>% 
  mutate(grp = sprintf('(%s,%s,t%s)', ord, task_df, t)) %>%
  group_by(grp = factor(grp, levels = unique(grp)) ) %>% 
  # or use summarise, but reframe will be preferred when the number
  # of rows are expanded
  reframe(val = combn(motiv, 2, 
    FUN = \(x) str_c(x[2], ' - ', x[1])),
       ind = combn(rn, 2, \(x) c(x[2], -x[1]), simplify = FALSE)) %>% 
  unite(val, val, grp, sep = " ") %>%
  with(., setNames(ind, val))

-output

`2.9 - 1.7 (c2s,complex,t1)`
[1]  2 -1

$`4.1 - 1.7 (c2s,complex,t1)`
[1]  3 -1

$`4.1 - 2.9 (c2s,complex,t1)`
[1]  3 -2

$`2.9 - 1.7 (s2c,complex,t2)`
[1]  5 -4

$`4.1 - 1.7 (s2c,complex,t2)`
[1]  6 -4

$`4.1 - 2.9 (s2c,complex,t2)`
[1]  6 -5

$`2.9 - 1.7 (c2s,complex,t3)`
[1]  8 -7

$`4.1 - 1.7 (c2s,complex,t3)`
[1]  9 -7

$`4.1 - 2.9 (c2s,complex,t3)`
[1]  9 -8

$`2.9 - 1.7 (s2c,complex,t4)`
[1]  11 -10

$`4.1 - 1.7 (s2c,complex,t4)`
[1]  12 -10

$`4.1 - 2.9 (s2c,complex,t4)`
[1]  12 -11

$`2.9 - 1.7 (s2c,simple,t1)`
[1]  14 -13

$`4.1 - 1.7 (s2c,simple,t1)`
[1]  15 -13

$`4.1 - 2.9 (s2c,simple,t1)`
[1]  15 -14

$`2.9 - 1.7 (c2s,simple,t2)`
[1]  17 -16

$`4.1 - 1.7 (c2s,simple,t2)`
[1]  18 -16

$`4.1 - 2.9 (c2s,simple,t2)`
[1]  18 -17

$`2.9 - 1.7 (s2c,simple,t3)`
[1]  20 -19

$`4.1 - 1.7 (s2c,simple,t3)`
[1]  21 -19

$`4.1 - 2.9 (s2c,simple,t3)`
[1]  21 -20

$`2.9 - 1.7 (c2s,simple,t4)`
[1]  23 -22

$`4.1 - 1.7 (c2s,simple,t4)`
[1]  24 -22

$`4.1 - 2.9 (c2s,simple,t4)`
[1]  24 -23