From geometric quantiles to halfspace depths: A geometric approach for extremal behaviour
share
We investigate the asymptotics for two geometric measures, geometric quantiles and halfspace depths. While much literature is known on the population side, we fill out some gaps there to obtain a full picture, before turning to the sample versions, where the questions on asymptotics become crucial in view of applications. This is the core of the paper: We provide rates of convergence for the sample versions and address the extremal behaviour of the geometric measures according to the type of underlying distribution.
READ FULL TEXT