Research

Causal discovery in multivariate extremes

Tail-induced asymmetry enables causal structure learning in high-dimensional multivariate extremes. We develop methods that exploit directional information encoded by tail dependence patterns to recover causal graphs in extreme regimes.

Extreme quantile treatment effect (EQTE)

Extreme quantile treatment effects (eQTEs) measure causal impacts on outcome distribution tails. We propose the TIEE framework combining information across quantile levels with extreme value models, enabling causal inference for rare, high-impact events in environmental risk, economics, and public health.

Extreme Event Attribution (EEA)

Exploring how tail assumptions influence attribution of extreme climate events, combining simulation, spatial methodology, and software development.

Wee Extremes: EVA 2023 Data Challenge

Statistical Methodology for Extreme Event Risk Assessment in the EVA 2023 Data Challenge.