Comparing dates within groups-CodePudding

I have a data frame with data split into groups with each row in each group given a code. I want the dates of everything within the group to be compared to the date given to 'a'. So for group 1, 'a' has a date of '2022-01-01', 'b' has a date of '2022-01-03' which is after the date of 'a' and therefore 'after' is written in the comment field (which shows the expected outcomes of the data in this example). I want to write code that populates the 'comment' field based on the rule above.

group<-c(1,1,1,1,2,2,2)
code<-c("a","b","c","d","a","b","c")
date<-c('2022-01-01','2022-01-03', '2021-12-15', '2022-05-01','2021-06-01', '2022-04-04','2022-05-10')
comment<-c(NA, "after","before","after",NA,"after","after")

data<-data.frame(group, code, date, comment, stringsAsFactors = FALSE)

Can anyone suggest any code?

CodePudding user response：

Do you mean this?

library(dplyr)
data %>%
    group_by(group) %>%
    mutate(comment2 = case_when(
        first(as.Date(date)) == as.Date(date) ~ NA_character_,
        first(as.Date(date)) > as.Date(date) ~ "before",
        TRUE ~ "after")) %>%
    ungroup()
## A tibble: 7 × 5
#  group code  date       comment comment2
#  <dbl> <chr> <chr>      <chr>   <chr>   
#1     1 a     2022-01-01 NA      NA      
#2     1 b     2022-01-03 after   after   
#3     1 c     2021-12-15 before  before  
#4     1 d     2022-05-01 after   after   
#5     2 a     2021-06-01 NA      NA      
#6     2 b     2022-04-04 after   after   
#7     2 c     2022-05-10 after   after

I'm creating a new comment2 column to explicitly show that it reproduces your expected output in column comment.

This assumes that

There is always an "a" code, and
Codes are ordered "a", "b", "c" by group (making "a" the first code in every group).

If codes are not ordered (but there is always an "a" code), you can do

data %>%
    group_by(group) %>%
    mutate(comment2 = case_when(
        as.Date(date)[code == "a"] == as.Date(date) ~ NA_character_,
        as.Date(date)[code == "a"] > as.Date(date) ~ "before",
        TRUE ~ "after")) %>%
    ungroup()