Halfspace depth and floating body (Q2002525)

The halfspace depth is one of the most promising tools for the efficient classifiers design for multivariate and functional data. The main result of this paper is the link between nonparametric statistics and convex and affine geometry via connecting the concepts of the halfspace depth and the convex floating body. Starting with comprehensive historical review of the halfspace depth, authors deal with the convex bodies approximation by polytopes using the log-concave measures. It is shown that convex floating body corresponds to the upper level sets of the halfspace depth. Then conditions under which no two probability measures can have the same depth are formulated. One of the main results of this paper is the bound of the halfspace depth formulated in terms of the Mahalanobis depth. Finally, the distribution-by-depth characterization problem is attacked.