How do I pivot pairs of columns in R?

user_1422335 0
Viewed 85

I'm having difficulty pivotting paired columns. I understand there is the names_pattern argument and I cannot figure out how to make this work.

My data looks like this:

structure(list(Q8_1_avg = 4.72562264837839, Q8_1_avg_se = 0.0595342202500642, 
               Q8_1_unweighted = 4.90473815461347, Q8_2_avg = 4.65508494735021, 
               Q8_2_avg_se = 0.0541589332376175, Q8_2_unweighted = 4.6498753117207, 
               Q8_3_avg = 5.4756060523178, Q8_3_avg_se = 0.0534895224170486, 
               Q8_3_unweighted = 5.57506234413965), row.names = c(NA, -1L
               ), class = "data.frame")->dat

And my desired output looks like this:

df<-data.frame(
Question=c('Q8_1', 'Q8_2','Q8_3'),
#Values taken from Q8_[123]_avg
Weighted_Average=c(4.72,4.65, 5.47),
#Values taken from Q8_[123]_avg_se
Weighted_SE=c(0.05,0.05 ,0.05),
#Values taken from Q8_[123]_unweighted
Unweighted_Average=c(4.90, 4.64, 5.57)
)
df

Thank you for any assistance.

tidyr r
1 Answers

We can use pivot_longer. If needed use rename to change the column names. Specify the names_to as a vector of 'Question' (which return the prefix part of the column name) and .value returns the value in the long format. Then, in names_pattern, capture the prefix part i.e. one or more characters not a _ ([^_]+) followed by _ and some digits (\\d+) as a group ((...)) then the _ and the second capture group involving the rest of the characters ((.*))

library(dplyr)
library(tidyr)
dat %>% 
   pivot_longer(cols = everything(), names_to = c("Question", 
       ".value"), names_pattern = "^([^_]+_\\d+)_(.*)")
# A tibble: 3 × 4
  Question   avg avg_se unweighted
  <chr>    <dbl>  <dbl>      <dbl>
1 Q8_1      4.73 0.0595       4.90
2 Q8_2      4.66 0.0542       4.65
3 Q8_3      5.48 0.0535       5.58
user_37322710
Related