EcoSta 2024: Start Registration
View Submission - EcoSta2024
A1075
Title: On order selection for multivariate extremes via clustering Authors:  He Tang - University of Georgia (United States) [presenting]
Shuyang Bai - University of Georgia (United States)
Shiyuan Deng - University of Georgia (United States)
Abstract: The estimation of multivariate extreme models with a discrete spectral measure is investigated using clustering techniques. The primary innovation involves devising a method for selecting the appropriate order that not only consistently identifies the true order in theory but also has a straightforward and easy implementation in practice. Specifically, an extra penalty term is introduced to the well-known simplified average silhouette width, which penalizes small cluster sizes and minimal dissimilarities between cluster centers. Consequently, a consistent method is provided for the order of a max-linear factor model, where a typical information-based approach is not viable due to the absence of likelihood.