Speakers

• Abeer Hasan — North Carolina State University

Abstract

Data in biostatistics, finance, environmental science, and related fields often exhibit asymmetry, heavy tails, and latent structural shifts that are not well captured by classical multivariate models. Although skew distributions offer greater flexibility, many existing formulations represent departures from symmetry through mechanisms that can behave like location adjustments rather than structural changes in distributional structure.

This talk introduces a non-central formulation of the multivariate skew distribution, in which the location parameter enters the model before the scale-mixture transformation. This construction provides a distinct mechanism for representing non-centrality, allowing latent mean shifts to influence shape, tail behavior, and dependence structure rather than simply translating the distribution.

We will outline the model's main theoretical features and illustrate its performance through simulations and an application to tumor shape data. The results suggest that the proposed non-central skew framework can improve modeling of complex multivariate data when asymmetry and heavy tails arise from mechanisms not adequately represented by conventional normal, skew-normal, or skew models.

Looking for more in the field?

Explore more events from Journal of Statistical Theory and Practice Webinars and Springer Nature's Mathematics & Statistics.