将 dplyr 函数映射到 R 数据帧中变量对的每个组合

Question

我想将一个函数映射到 R 中数据帧中的每个变量组合对，返回一个带有每对函数输出的数据帧。我可以像这样手动执行此操作：

library(tidyverse)

df <- tibble(a = c(1, 2), b = c(4, 3), c = c(5, 7))

f <- function(a, b) a - b # a simple function for sake of example

df %>% transmute(a_minus_b = f(a, b),
                 a_minus_c = f(a, c),
                 b_minus_c = f(b, c),
                 b_minus_a = f(b, a),
                 c_minus_a = f(c, a),
                 c_minus_b = f(c, b))

对于具有许多变量的数据框，手动执行此操作显然是不切实际的。如何使用迭代将我的函数应用于每个变量组合对？

Answer 1

另一种使用dplyr和 的方法purrr可能如下所示：

library(tidyverse)

df <- tibble(a = c(1, 2), b = c(4, 3), c = c(5, 7))

f <- function(a, b) a - b # a simple function for sake of example

f_help <- function(x) {
  df %>% 
    transmute_at(setdiff(names(.), x), ~ f(!!sym(x), .x)) %>%
    rename_all(.funs = ~ paste0(x, "_minus_", .x))
}

map(names(df), f_help) %>% 
  bind_cols()
#> # A tibble: 2 x 6
#>   a_minus_b a_minus_c b_minus_a b_minus_c c_minus_a c_minus_b
#>       <dbl>     <dbl>     <dbl>     <dbl>     <dbl>     <dbl>
#> 1        -3        -4         3        -1         4         1
#> 2        -1        -5         1        -4         5         4

Answer 2

一种dplyr和purrr解决方案可能是：

map_dfc(.x = c(combn(rev(names(df)), 2, simplify = FALSE),
               combn(names(df), 2, simplify = FALSE)),
        ~ df %>%
         rowwise() %>%
         transmute(!!paste(.x, collapse = "_") := reduce(c_across(all_of(.x)), `-`)) %>%
         ungroup())

    c_b   c_a   b_a   a_b   a_c   b_c
  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1     1     4     3    -3    -4    -1
2     4     5     1    -1    -5    -4

或者使用指定的函数：

map_dfc(.x = c(combn(rev(names(df)), 2, simplify = FALSE),
               combn(names(df), 2, simplify = FALSE)),
        ~ df %>%
         rowwise() %>%
         transmute(!!paste(.x, collapse = "_") := reduce(c_across(all_of(.x)), f)) %>%
         ungroup())

Answer 3

使用 tidyverseset_names

library(tidyverse)
f <- function(a, b) a - b # a simple function for sake of example
c(combn(df, 2, simplify = F),
  combn(rev(df), 2, simplify = F)) %>% 
  set_names(map_chr(., ~paste(names(.), collapse = "_minus_"))) %>% 
  map(., ~f(.x[1], .x[2]) %>% pull) %>%   
  bind_cols()
 # A tibble: 2 x 6
  a_minus_b a_minus_c b_minus_c c_minus_b c_minus_a b_minus_a
      <dbl>     <dbl>     <dbl>     <dbl>     <dbl>     <dbl>
1        -3        -4        -1         1         4         3
2        -1        -5        -4         4         5         1