Abstract
We propose a new characterization of $d$-variate measures by special convex sets in $d+1$-dimensional space. The latter sets form a subclass of zonoids, which we called lift-zonoids. This mapping from $d$-variate measures to convex sets in $R^{d+1}$ is affine equivariant, continuous, additive and might be thought as a set valued expectation of a measure. This mapping enables us to use methods of convex geometry in statistics. We will present results of using such methods in problems of risk measurement and data analysis.Welcome.
Pasi Koikkalainen
Laboratory of Data Analysis