在 r 中将 dataframe 从宽转换为长时，如何创建两个组？

Question

I have a wide dataset which I would like to convert to a long format and split the columns into two groups; in the summary example below, variable score n.1 would be group 1 and variable score n.2. How do I convert a wide format data frame into a long format data frame and assign a group to the respective variables?

   age.1<- c(23,34,52,12,23)
score1.1 <- c(44,23,62,1,0)
score2.1<- c(3,4,2,1,3)
score3.1<- c(230,304,502,102,203)
score1.2<- c(2343,4534,5652,1642,2233)
score1.2<- c(2233,32324,5232,1232,2233)
score2.2<- c(12323,12334,1352,1312,1323)
score3.2<- c(21233,33454,53452,12452,23532523)

df<- data.frame(age.1, score1.1,score2.1, score3.1, score1.2, score2.2, score3.2)

Answer 1

We can use stringi::stri_extract_last_regex to extract the last numbers from a string:

library(dplyr)
library(tidyr)
df %>%
  pivot_longer(-age.1) %>%
  mutate(group = stringi::stri_extract_last_regex(name, "[0-9]+"))
# # A tibble: 30 × 4
#    age.1 name     value group
#    <dbl> <chr>    <dbl> <chr>
#  1    23 score1.1    44 1    
#  2    23 score2.1     3 1    
#  3    23 score3.1   230 1    
#  4    23 score1.2  2233 2    
#  5    23 score2.2 12323 2    
#  6    23 score3.2 21233 2    
#  7    34 score1.1    23 1    
#  8    34 score2.1     4 1    
#  9    34 score3.1   304 1    
# 10    34 score1.2 32324 2    
# # … with 20 more rows

Answer 2

If you would like to split the number after the period into a new column while pivoting, then we can use names_pattern :

library(tidyverse)

df %>% pivot_longer(
  cols = -age.1,
  names_to = c("name", "group"),
  names_pattern = "(.+).(.)",
  values_to = "Value"
)

Output

# A tibble: 30 × 4
   age.1 name   group Value
   <dbl> <chr>  <chr> <dbl>
 1    23 score1 1        44
 2    23 score2 1         3
 3    23 score3 1       230
 4    23 score1 2      2233
 5    23 score2 2     12323
 6    23 score3 2     21233
 7    34 score1 1        23
 8    34 score2 1         4
 9    34 score3 1       304
10    34 score1 2     32324
# … with 20 more rows

However, if you need to retain those numbers in the name column, then we can just get the value after pivoting:

df %>%
  pivot_longer(-age.1) %>%
  mutate(group = str_replace(name, '.*\\.', ""))

Output

# A tibble: 30 × 4
   age.1 name     value group
   <dbl> <chr>    <dbl> <chr>
 1    23 score1.1    44 1    
 2    23 score2.1     3 1    
 3    23 score3.1   230 1    
 4    23 score1.2  2233 2    
 5    23 score2.2 12323 2    
 6    23 score3.2 21233 2    
 7    34 score1.1    23 1    
 8    34 score2.1     4 1    
 9    34 score3.1   304 1    
10    34 score1.2 32324 2    
# … with 20 more rows