The DataFrame API includes an approximate quartile implementation. If you ask for quantile 0.5, you will get approximate median. 

On Sun, Apr 16, 2017 at 9:24 PM svjk24 <> wrote:
  Is there any interest in an efficient distributed computation of the median algorithm?
A google search pulls some stackoverflow discussion but it would be good to have one provided.

I have an implementation (that could be improved)
from the paper " Fast Computation of the Median by Successive Binning":