Quantile Function for a Vector of Dates

Question

I noticed that the base R quantile function does not support date arguments.

I appreciate that defining quantiles for dates needs care in the definitions (i.e. if you have 6 dates and ask for the 25th percentile, you need to define the suitable rounding).

Is there an efficient implementation of such a quantile function, either as a part of base or another package.

The following sample function achieves essentially what I am interested in (with some tweaking to handle the case of the 0'th percentile), but I imagine that more efficient implementations are possible.

#Date quantile function.
dquantile <- function(x, probs){

  sx <- sort(x)

  pos <- round( probs * length(x) )

  return( sx[pos] )
}

# Example.
dates <- as.Date("01/01/1900", "%d/%m/%Y") + floor( 36500 * runif(100000) )

dquantile(dates, c(0.001, 0.025, 0.975, 0.999) )

Answer 1

The quantile function does support dates, you just need to specify the type argument. Your problem can be solved with:

dates <- as.Date("01/01/1900", "%d/%m/%Y") + floor( 36500 * runif(100000) )

quantile(dates, probs = c(0.001, 0.025, 0.975, 0.999), type = 1)

        0.1%         2.5%        97.5%        99.9% 
"1900-02-04" "1902-06-23" "1997-06-10" "1999-10-30"