从另一个中减去因子水平

Question

Given the following data frame:

set.seed(42)
test <- data.frame(tb = c(rep("top", 5), rep("bottom", 5)),
                   name = rep(c("q1", "q2", "q3", "q4", "q5"), 2),
                   values = rnorm(10, 1))

I'd like to subtract each "name" (bottom) level from its corresponding (top) value. For example: top q1 - bottom q1 = 1.4770829, top q2 - bottom q2 = -2.0762202, etc.

How can I produce the following results? In reality, "name" will have many more levels than five.

q1  1.4770829
q2  -2.0762202
q3  0.4577874
q4  -1.3855611
q5  0.4669824

Answer 1

You can reshape the data and subtract the two columns.

library(dplyr)
library(tidyr)

test %>%
  pivot_wider(names_from = tb, values_from = values) %>%
  mutate(diff = top - bottom)

#  name    top bottom   diff
#  <chr> <dbl>  <dbl>  <dbl>
#1 q1    2.37   0.894  1.48 
#2 q2    0.435  2.51  -2.08 
#3 q3    1.36   0.905  0.458
#4 q4    1.63   3.02  -1.39 
#5 q5    1.40   0.937  0.467

Another way would be -

test %>%
  arrange(name, tb) %>%
  group_by(name) %>%
  summarise(values = diff(values))

#  name  values
#  <chr>  <dbl>
#1 q1     1.48 
#2 q2    -2.08 
#3 q3     0.458
#4 q4    -1.39 
#5 q5     0.467