1 paper
Peter M. Aronow, Donald K. K. Lee
Given samples x1,⋯,xn, it is well known that any sample median value (not necessarily unique) minimizes the absolute loss ∑i=1n∣q−xi∣. Interestingly, we show th…