Doktorandseminarium: Skewness of random variables and a (not so) silly definition of convexity

  • Datum:
  • Plats: 64119
  • Föreläsare: Tilo Wiklund
  • Kontaktperson: Volodymyr Mazorchuk
  • Seminarium

Abstract:

Various definitions of skewness have been suggested throughout the history of statistics. It has been proposed that any reasonable such notion should be monotone with respect to a partial order (on

distributions) first suggested by van Zwet. Proving that one distribution precedes another in this order often involves proving convexity of functions that are not obviously convex. Here the theory of variation diminishing kernels can come to the rescue. We will have a look at this and, as time permits, some other application of counting sign changes and the theory of variation diminishing kernels in statistics.