R : know table(s), how to calc quantile(s)

Question

s is a large array, just save table(s) in database

> table_s
s
1       2 3  4                   5
3000000 1 1  999999999999999999  34

how to calc quantile(s) with table_s in R ?

thanks

Answer 1

You can use the quantile function from the Hmisc package, which allows weights.

Hmisc::wtd.quantile(as.numeric(names(table_s)),weights = table_s)

Answer 2

The simplest (but computationally expensive) way I can think of is to re-expand your table into a vector of observations and use the quantile function:

s <- c(3000000,1,1,999999999999999999,34)
names(s) <- 1:5    
quantile(rep.int(as.integer(names(s)),times=s))
#  0%  25%  50%  75% 100% 
#   1    4    4    4    5 

If you are looking for something faster, then you might need to write your own function.

EDIT: As Matthew Lundberg states in the comments, the code above doesn't work. It will run only if sum(s) is less than the maximum possible length of a vector, which is currently 2^31-1 < 10^10.