The Department of Statistics and Actuarial Science Fall Colloquium Series presents:

Reetam Majumder, Assistant Professor of Statistics, Department of Mathematical Sciences, University of Arkansas

"Bayesian inference for bivariate extremes using semi-parametric representations of limit sets"

Abstract: Compound extremes, namely simultaneous, concurrent, or coincident extreme events, may have broader impacts on human society and the environment than any individual extreme alone. The study of geometric extremes, where extremal dependence properties are inferred from the deterministic limiting shapes of scaled sample clouds, provides an exciting approach to modelling the extremes of multivariate data. These shapes, termed limit sets, link together several popular extremal dependence modelling frameworks. We propose a flexible, semi-parametric model for joint-tail inference for a bivariate random vector using limit sets. The data is represented in polar coordinates; its angular component is estimated using periodic M-splines that have a natural representation as density functions, whereas its radial component, conditional on the angles, is used to estimate a limit set. In particular, we model the limit set using Bezier splines, which constitutes a parsimonious representation of the joint tail dependence without making any parametric copula assumptions. The model is fitted in a Bayesian framework, providing uncertainty quantification that is desirable for extreme events, and can be used to estimate probabilities associated with failure sets. We demonstrate the efficacy of our approach through a range of numerical studies and a fire weather dataset.

This virtual presentation begins at 3:15 pm.

Topic: Colloquium 2026-2027

Time: Sept 17, 2026 03:15 PM Central Time (US and Canada)

Join Zoom Meeting

Meeting ID: 915 1899 4105