在 R 中循环数据帧删除过程

Question

I wonder how to loop my code below to make it more functional and generalizable for other data (the current data is just a toy):

FIRST , I select a study from data using sample() and then filter() rows of it whose outcome == outcome_to_remove . This gives datat output.

SECOND , I select a study from datat using sample() and then filter() rows of it whose outcome == outcome_to_remove2 . This gives the final output.

Can we possibly loop this process?

EDIT: The only conditional I would like to add to my code is that the length(unique(data$study)) before and after the looping should always remain the same. That is, it shouldn't be possible that a study looses its outcome == "A" in the FIRST step, and outcome == "B" at the SECOND step, thus the whole study gets deleted.

(data <- expand_grid(study = 1:5, group = 1:2, outcome = c("A", "B")))

n = 1
#====-------------------- FIRST:  
studies_to_remove = sample(unique(data$study), size = n)
outcome_to_remove = c("A")
             
datat <- data %>%
  filter(
    !(    study %in% studies_to_remove &
        outcome %in% outcome_to_remove
    ))

#====------------------- SECOND:
studies_to_remove2 = sample(unique(datat$study), size = n)
outcome_to_remove2 = c("B")

datat %>%
  filter(
    !(    study %in% studies_to_remove2 &
        outcome %in% outcome_to_remove2
    ))

Answer 1

With the help of for loop -

data <- tidyr::expand_grid(study = 1:5, group = 1:2, outcome = c("A", "B"))

n = 1
set.seed(9873)
outcome_to_remove <- unique(data$outcome)
unique_study <- unique(data$study)

for(i in outcome_to_remove) {
  studies_to_remove = sample(unique_study, size = n)
  outcome_to_remove = i
  unique_study <- setdiff(unique_study, studies_to_remove)
  cat('\nDropping study ', studies_to_remove, 'and outcome ', outcome_to_remove)
  data <- data %>%
    filter(
      !( study %in% studies_to_remove &
         outcome %in% outcome_to_remove
      ))
}

#Dropping study  3 and outcome  A
#Dropping study  1 and outcome  B

data
#   study group outcome
#   <int> <int> <chr>  
# 1     1     1 A      
# 2     1     2 A      
# 3     2     1 A      
# 4     2     1 B      
# 5     2     2 A      
# 6     2     2 B      
# 7     3     1 B      
# 8     3     2 B      
# 9     4     1 A      
#10     4     1 B      
#11     4     2 A      
#12     4     2 B      
#13     5     1 A      
#14     5     1 B      
#15     5     2 A      
#16     5     2 B