Bayesian model averaging over tree-based dependence structures for multivariate extremes

Vettori, Sabrina [King Abdullah University of Science and Technology (KAUST) Thuwal, Saudi Arabia] Huser, Raphaël [King Abdullah University of Science and Technology (KAUST) Thuwal, Saudi Arabia] Segers, Johan [UCL] Genton, Marc G. [King Abdullah University of Science and Technology (KAUST) Thuwal, Saudi Arabia]

Describing the complex dependence structure of extreme phenomena is particularly challenging. To tackle this issue we develop a novel statistical method that describes extremal dependence taking advantage of the inherent tree-based dependence structure of the max-stable nested logistic distribution, and that identifies possible clusters of extreme variables using reversible jump Markov chain Monte Carlo techniques. Parsimonious representations are achieved when clusters of extreme variables are found to be completely independent. Moreover, we significantly decrease the computational complexity of full likelihood inference by deriving a recursive formula for the likelihood function of the nested logistic model. The method’s performance is verified through extensive simulation experiments which also compare different likelihood procedures. The new methodology is used to investigate the dependence relationships between extreme concentrations of multiple pollutants in California and how these concentrations are related to extreme weather conditions. Overall, we show that our approach allows for the representation of complex extremal dependence structures and has valid applications in multivariate data analysis, such as air pollution monitoring, where it can guide policymaking

• metadata
• relations

Document type	Article de périodique (Journal article) – Article de recherche
Access type	Accès libre
Publication date	2019
Language	Anglais
Journal information	"Journal of Computational and Graphical Statistics" - Vol. 29, no. 1, p. 174-190 (2020)
Peer reviewed	yes
Publisher	Informa UK Limited
issn	1061-8600
e-issn	1537-2715
Publication status	Publié
Affiliation	UCL - SSH/LIDAM/ISBA - Institut de Statistique, Biostatistique et Sciences Actuarielles
Keywords	Air pollution ; Extreme event ; Fast likelihood inference ; Nested logistic model ; Reversible jump Markov chain Monte Carlo
Links	https://doi.org/10.1080/10618600.2019.1647847[DOI] http://hdl.handle.net/2078.1/218809[Handle]

See also
	[boreal:185485]	Bayesian Clustering and Dimension Reduction in Multivariate Extremes