I want to write a case_when code in a dplyr pipeline. However, I am trying to add multiple cases within it.

For example: If a have the following data frame

I want to summarize the output so that case when the first 2 observations per each id are needs and case when the observation "none" is not put in consideration then put yes in a new column. If there's no need or want for a given id then none, else no

The output should be the following: |id|output| |--|---------| |a|no| |b|no| |c|yes| |d|none|

My code

actions %>% group_by (id) %>% arrange(id) 
%>% summarise(output = case_when(first(purchases) == "need" & nth(purchases,2) =="need"~ "yes", "no"

I know the code is a bit messy, as I don't know who to add up the second condition of neglecting none observations when the cases would result in a yes or no

CodePudding user response：

I've tried to place your logic in a small function f(), which can then be applied to purchases, by id

f <- function(p) {
  if(p[1]==p[2] & (p[1] %in% c("need", "want"))) return("yes")
  ifelse(all(p=="none"), "none", "no")
}
df %>% group_by(id) %>% summarize(output=f(purchases))

Output

  id    output
  <chr> <chr> 
1 a     no    
2 b     no    
3 c     yes   
4 d     none 

The function checks if the first and second value of purchases are equal, and if they are either need or want; if so return "yes". Otherwise if all of purchases values are "none", return "none", else return "no".

CodePudding user response：

Try this using case_when

actions %>% group_by(id) %>%
   summarise(output = 
   case_when(isTRUE(intersect(purchases[[1]] , purchases[[2]]) == "none") ~ "none" ,
             isTRUE(intersect(purchases[[1]] , purchases[[2]]) %in% c("need" , "want")) ~ "yes",
             TRUE ~ "no"))