dplyr：我如何根据其他列中的值计算组内的倍数变化

Question

My current data roughly has the following pattern:

Tree   Fertilized   Region   Fruits

apple  lightly      sunny    100
apple  lightly      dark     50
apple  heavily      sunny    300
apple  heavily      dark     200
pear   lightly      sunny    150
pear   lightly      dark     200
pear   heavily      sunny    300
pear   heavily      dark     150

Here I want to calculate (as part of a bigger function) the fold-change of placing the tree in a sunny place compared to a dark one within each combination of fertilization amount and type of tree(eg a 2-fold change for lightly fertilized apple trees):

df%<>%
  group_by(Tree,Fertilized) %>% 
  summarise(!!paste0("fold_change_", quote(Fruits)) := .[Region == "sunny","Fruits"]/.[type == "dark","Fruits"])

However, I get an error saying that the "Fruits" column doesn't exist. Does anyone have a suggestion on how to get this working? I guess the solution is some minor syntax tweak but I can"t seem to find it myself or online.

The actual dataset has many more tree types and parameters like "Fruits", hence I picked the pipe structure and dynamic labelling of columns (",:paste0()", ".="), which may be relevant or irrelevant for solving this issue.

Thanks in advance to anyone trying to help!

Cheers, Rob

Answer 1

I would use a group-by operation:

library(data.table)
library(dplyr)


f <- tempfile()
writeLines("
Tree,  Fertilized,  Region,  Fruits,
apple, lightly, sunny, 100,
apple, lightly, dark, 50,
apple, heavily, sunny, 300,
apple, heavily, dark, 200,
pear, lightly, sunny, 150,
pear, lightly, dark, 200,
pear, heavily, sunny, 300,
pear, heavily, dark, 150
", f)
dat <- read.csv(f)

data.table

dat <- data.table(dat)

dat[order(Region), .(fold_change = Fruits[2] / Fruits[1]), by=.(Tree, Fertilized)]
#>     Tree Fertilized fold_change
#> 1: apple    lightly        2.00
#> 2: apple    heavily        1.50
#> 3:  pear    lightly        0.75
#> 4:  pear    heavily        2.00

tidyverse

dat %>% 
  arrange(Region) %>%
  group_by(Tree, Fertilized)  %>%
        summarize(fold_change = Fruits[2] / Fruits[1])
#> `summarise()` regrouping output by 'Tree' (override with `.groups` argument)
#> # A tibble: 4 x 3
#> # Groups:   Tree [2]
#>   Tree  Fertilized fold_change
#>   <chr> <chr>            <dbl>
#> 1 apple " heavily"        1.5 
#> 2 apple " lightly"        2   
#> 3 pear  " heavily"        2   
#> 4 pear  " lightly"        0.75