Customized floor/ceiling for dates in R

Question

say I have a date range from start to end where start <- as.Date(2009-11-05), end <- as.Date(2009-12-17).

I want a function that essentially acts as a customized floor/ceiling and returns a date interval such that the lower bound is the first date of the form 'yyyy-mm-23' smaller or equal to '2009-11-05' and the upper bound is the first date greater or equal to '2009-12-17' and of the form 'yyyy-mm-22'.

In the above example, the function should return the interval ('2009-10-23, 2009-12-22).

I've tried using seq.Date and using the length function on it but it seems tedious and I wonder if there is a faster solution.

Thanks

score 2 · Answer 1 · answered Jan 10 '23 at 00:02

I don't know a function that works like that, but I would code something like this:

Code

custom_bound <- function(date, type, ref_day){
  
  obs_day <- lubridate::day(date)

  if(type == "lower"){aux <- -1}
  
  if(type == "upper"){aux <- 1}      
  
  while(obs_day != ref_day ){
    
    date <- date + days(aux)
    
    obs_day <- lubridate::day(date)
    
  }
  
  return(date)
  
}

Output

> custom_bound(date = as.Date("2009-11-05"),type = "lower",ref_day = 23)
[1] "2009-10-23"

> custom_bound(date = as.Date("2009-12-17"),type = "upper",ref_day = 22)
[1] "2009-12-22"

Ben · Accepted Answer · 2023-01-10T13:40:38.177

This might be another approach to try. Using lubridate create new start and end dates, substituting the 23rd and 22nd. Then, if the start date precedes the 23rd, subtract a month. Likewise, if the end date exceeds the 22nd, add a month.

start <- as.Date("2009-11-05")
end <- as.Date("2009-12-17")

library(lubridate)

my_fun <- function(start, end) {
  new_start <- start
  day(new_start) <- 23
  new_end <- end
  day(new_end) <- 22
  if (day(start) < 23) new_start = new_start %m-% months(1)
  if (day(end) > 22) new_end = new_end %m+% months(1)
  return(interval(new_start, new_end))
}

my_fun(start, end)

Output

[1] 2009-10-23 UTC--2009-12-22 UTC

Edit: In the comment, the reference day of the month could be greater than 28, which could result in an invalid date. To consider this possibility, one approach is to use the clock package which can handle an invalid date (e.g., Feb. 31), and then resolve to closest day.

start <- as.Date("2009-03-30")
end <- as.Date("2009-12-17")

reference <- 31

library(lubridate)
library(clock)

my_fun <- function(start, end, reference) {
  new_start <- set_day(year_month_day(year(start), month(start)), reference)
  new_end <- set_day(year_month_day(year(end), month(end)), reference)
  if (day(start) < reference) new_start = add_months(new_start, -1) 
  if (day(end) > reference) new_end = add_months(new_end, 1)
  new_start = invalid_resolve(new_start, invalid = "previous") 
  new_end = invalid_resolve(new_end, invalid = "next")
  return(c(new_start, new_end))
}

my_fun(start, end, reference)

Output

[1] "2009-02-28" "2009-12-31"

Thanks, this is exactly what I was looking for! Nice and simple, avoiding the creation of date sequences with seq.Date — Algebro1000, Jan 10 '23 at 03:53
What about the case where we have the 31st as reference day and start date is say 2009-02-28? We surely can’t set day(new_start) <- 31 then, but we want new_start to be 2009-02-28? — Algebro1000, Jan 10 '23 at 07:04
@Lollo231000 If you can have a "reference" day that is greater than 28, you can use the `clock` package and then resolve invalid dates. See second example in the edited answer above. Let me know if this is what you had in mind. — Ben, Jan 10 '23 at 13:41
what a brilliant package! Thank you so much for your excellent solution. — Algebro1000, Jan 10 '23 at 22:39

Customized floor/ceiling for dates in R

2 Answers2

Code

Output