重命名R中的相关变量组

Question

注意：我不确定这个问题的好标题是什么。我很乐意修复它。

假设我有这个测试数据框：

library(tidyverse)
test.dat <- tibble(
  "User1" = c("Aaron","Aaron","Charlie"),
  "User2" = c("Ben","Ben","Aaron"),
  "User3" = c("Charlie","Charlie","Ben"),
  "first_1" = c("A","A","C"),
  "first_2" = c("B","B","A"),
  "first_3" = c("C","C","B"),
  "second_1" = c("A","A","C"),
  "second_2" = c("B","B","A"),
  "second_3" = c("C","C","B"),
  "third_1" = c("A","A","C"),
  "third_2" = c("B","B","A"),
  "third_3" = c("C","C","B")
)

# A tibble: 3 x 12
  User1   User2 User3   first_1 first_2 first_3 second_1 second_2 second_3 third_1 third_2 third_3
  <chr>   <chr> <chr>   <chr>   <chr>   <chr>   <chr>    <chr>    <chr>    <chr>   <chr>   <chr>  
1 Aaron   Ben   Charlie A       B       C       A        B        C        A       B       C      
2 Aaron   Ben   Charlie A       B       C       A        B        C        A       B       C      
3 Charlie Aaron Ben     C       A       B       C        A        B        C       A       B

我希望 User1 和带有 _1 的变量始终指代 Aaron，User2 始终是 Ben，带有 _2 的变量等等。

我可以做的是通过创建临时变量Loc 来定位每个名称在 User1:User3 中的位置，然后使用case_when：

test.try <- test.dat %>%
  rowwise() %>%
  mutate(U1Loc = case_when(User1 == "Aaron" ~ 1,
                           User2 == "Aaron" ~ 2,
                           User3 == "Aaron" ~ 3),
         U2Loc = case_when(User1 == "Ben" ~ 1,
                           User2 == "Ben" ~ 2,
                           User3 == "Ben" ~ 3),
         U3Loc = case_when(User1 == "Charlie" ~ 1,
                           User2 == "Charlie" ~ 2,
                           User3 == "Charlie" ~ 3)) %>%
  mutate(newUser_1 = case_when(U1Loc == 1 ~ User1,
                                U1Loc == 2 ~ User2,
                                U1Loc == 3 ~ User3),
         newUser_2 = case_when(U2Loc == 1 ~ User1,
                                U2Loc == 2 ~ User2,
                                U2Loc == 3 ~ User3),
         newUser_3 = case_when(U3Loc == 1 ~ User1,
                                U3Loc == 2 ~ User2,
                                U3Loc == 3 ~ User3)) %>%
  mutate(newFirst_1 = case_when(U1Loc == 1 ~ first_1,
                                U1Loc == 2 ~ first_2,
                                U1Loc == 3 ~ first_3),
         newFirst_2 = case_when(U2Loc == 1 ~ first_1,
                                U2Loc == 2 ~ first_2,
                                U2Loc == 3 ~ first_3),
         newFirst_3 = case_when(U3Loc == 1 ~ first_1,
                                U3Loc == 2 ~ first_2,
                                U3Loc == 3 ~ first_3)) %>%
  mutate(newSecond_1 = case_when(U1Loc == 1 ~ second_1,
                                 U1Loc == 2 ~ second_2,
                                 U1Loc == 3 ~ second_3),
         newSecond_2 = case_when(U2Loc == 1 ~ second_1,
                                 U2Loc == 2 ~ second_2,
                                 U2Loc == 3 ~ second_3),
         newSecond_3 = case_when(U3Loc == 1 ~ second_1,
                                 U3Loc == 2 ~ second_2,
                                 U3Loc == 3 ~ second_3)) %>%
  select(starts_with("new"))

得到

> test.try
# A tibble: 3 x 9
# Rowwise: 
  newUser_1 newUser_2 newUser_3 newFirst_1 newFirst_2 newFirst_3 newSecond_1 newSecond_2 newSecond_3
  <chr>     <chr>     <chr>     <chr>      <chr>      <chr>      <chr>       <chr>       <chr>      
1 Aaron     Ben       Charlie   A          B          C          A           B           C          
2 Aaron     Ben       Charlie   A          B          C          A           B           C          
3 Aaron     Ben       Charlie   A          B          C          A           B           C   

但是，变量越多，这个过程就越繁琐。除了 for 循环之外，有没有其他方法可以实现这一点，最好是使用 tidy 方法？我的猜测是使用across()，但我似乎无法让它像我想象的那样工作。

Answer 1

我们可以通过旋转到“长”格式轻松做到这一点，然后在替换值后将形状重新调整为宽

library(dplyr)
library(tidyr)
out <- test.dat %>% 
    mutate(rn = row_number()) %>% 
    pivot_longer(cols = -rn, names_to = c('.value', 'grp'),
         names_sep = "(?<=[a-z])_?(?=[0-9])") %>% 
    group_by(grp) %>% 
    mutate(across(User:third, first)) %>%
    pivot_wider(names_from = grp, values_from = c(User, first, second, third)) %>%
    select(-rn)

-输出

out
# A tibble: 3 x 12
#  User_1 User_2 User_3  first_1 first_2 first_3 second_1 second_2 second_3 third_1 third_2 third_3
#  <chr>  <chr>  <chr>   <chr>   <chr>   <chr>   <chr>    <chr>    <chr>    <chr>   <chr>   <chr>  
#1 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C      
#2 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C      
#3 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C      

---

如果我们想重命名以new为前缀的列

library(stringr)
out %>%
    rename_all(~ str_c('new', .))

---

或者另一种选择是创建一个键值数据集来匹配和替换列中的值是否没有顺序

keydat <- tibble(grp = as.character(1:3), UserKey = c("Aaron", "Ben", "Charlie"),
      abbr = substr(UserKey, 1, 1))


test.dat %>% 
        mutate(rn = row_number()) %>% 
        pivot_longer(cols = -rn, names_to = c('.value', 'grp'),
             names_sep = "(?<=[a-z])_?(?=[0-9])") %>%
        left_join(keydat) %>% 
        mutate(User = UserKey) %>% 
        mutate(across(first:third, ~ abbr)) %>%
        select(-UserKey, -abbr) %>% 
        pivot_wider(names_from = grp, values_from = c(User, first, second, third)) %>%
        select(-rn)
# A tibble: 3 x 12
#  User_1 User_2 User_3  first_1 first_2 first_3 second_1 second_2 second_3 third_1 third_2 third_3
#  <chr>  <chr>  <chr>   <chr>   <chr>   <chr>   <chr>    <chr>    <chr>    <chr>   <chr>   <chr>  
#1 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C      
#2 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C      
#3 Aaron  Ben    Charlie A       B       C       A        B        C        A       B       C