Is there an easy way to tell if an interval overlaps a weekend in R?

Question

I have a series of arrival and leaving dates. I want to know if those dates overlap a weekend. I could write a custom function to walk each of the days in the interval and see if they are a weekend. Is there a simpler way to do this that will scale better?

I am using lubridate, but I'm happy to use a different date package if that would make my job easier.

Answer 1

How about this base-R solution:

# make a set of sequences from beginning and ending dates and test with:
as.POSIXlt(x)$wday %in% c(0,6)

That would deliver a vector of TRUE/FALSE and you could determine whether any of the items in the sequence were TRUE with:

max( as.POSIXlt(x)$wday %in% c(0,6) )

Answer 2

With a_date as arrival date and d_date ansd departure date, something like this could work:

require(lubridate)
weekend_overlap <-    ifelse(wday(a_date) %in% c(1, 7) || 
   wday(d_date) %in% c(1, 7) || 
   interval(a_date,d_date)/ddays(1) > 4,TRUE,FALSE)

Answer 3

Here is a more general (vectorized) function that checks whether the interval from - to contains a certain weekday (1-7):

#' Check if a weekday is within an interval
#' 
#' @param wday Day of week (integer 1-7)
#' @param from Date. Can be a vector.
#' @param to Date. Same length as `from` and must be greater than `from`.
#' @param week_start 1 = Monday. 7 = Sunday
#' 
wday_in_interval = function(wday, from, to, week_start = 1) {
  if (wday < 1 | weekday > 7) 
    stop("wday must be an integer from 1 to 7.")
  if (week_start)
    wday = 1 + (((wday - 2) + week_start ) %% 7)  # Translate wday to week_start = 1 (ISO standard)
  if (any(from > to, na.rm = TRUE))
    stop("`from` must come before `to`")
  
  # If the interval is greater than a week, it trivially contains any weekday
  over_a_week = difftime(from, to, units = "days") >= 7
  
  # Check if weekday is both smaller/greater than "from" and "to"
  days_from = as.numeric(strftime(from, "%u"))
  days_to = as.numeric(strftime(to, "%u"))
  contains_weekday = ifelse(
    strftime(from, "%V") == strftime(to, "%V"),  # Dates are in the same week?
    yes = wday >= days_from & wday <= days_to,
    no = wday >= days_from | wday <= days_to  # 
  )
  
  return(over_a_week | contains_weekday)
}

Finding out whether an interval includes a weekend is then just a matter of checking that the interval does not include Saturdays OR Sundays:

library(dplyr)
tibble::tibble(
  timestamp = seq(as.POSIXct("2020-09-03 0:00"), as.POSIXct("2020-09-8 12: 00"), length.out = 10),
  overlaps_saturday = wday_in_interval(6, from = lag(timestamp), to = timestamp),
  overlaps_sunday = wday_in_interval(7, from = lag(timestamp), to = timestamp),
  overlaps_weekend = overlaps_saturday | overlaps_sunday
)

Result:

# A tibble: 10 x 4
   timestamp           overlaps_saturday overlaps_sunday overlaps_weekend
   <dttm>              <lgl>             <lgl>           <lgl>           
 1 2020-09-03 00:00:00 NA                NA              NA              
 2 2020-09-03 14:40:00 FALSE             FALSE           FALSE           
 3 2020-09-04 05:20:00 FALSE             FALSE           FALSE           
 4 2020-09-04 20:00:00 FALSE             FALSE           FALSE           
 5 2020-09-05 10:40:00 TRUE              FALSE           TRUE            
 6 2020-09-06 01:20:00 TRUE              TRUE            TRUE            
 7 2020-09-06 16:00:00 FALSE             TRUE            TRUE            
 8 2020-09-07 06:40:00 FALSE             TRUE            TRUE            
 9 2020-09-07 21:20:00 FALSE             FALSE           FALSE           
10 2020-09-08 12:00:00 FALSE             FALSE           FALSE