Constructing multivariate distributions with generalized marginals and t-copulas
Sarat C. Dass, Wenmei Huang, Mohana Sundaram Muthuvalu · AIP conference proceedings · 2014
Generalized distributions are probability distributions that have both discrete and continuous components. In this paper, a method is proposed for constructing flexible multivariate distributions based on arbitrarily pre-specified generalized marginals and t-copulas. We give theoretical results establishing identifiability of the parameters of the multivariate distribution. These distributions are useful for modeling real data that show non-Gaussian characteristics such as disease trajectories (i.e., malaria and dengue) over time and space.