Statistical inference and modeling with the S distribution

Abstract

We consider the problem of statistical inference for the S distribution and introduce new minimum distance estimators for the four parameters of the S distribution using Kolmogorov-Smirnov, Cramer-von Mises and related distance metrics. Approximate goodness-of-fit and confidence intervals for parameters are calculated using bootstrap methods. We discuss further how the S distribution can be used to solve various problems of statistical modeling associated with parameter inference, including goodness-of-fit tests, Monte Carlo simulations and modeling trends in the distributions.