A clustering algorithm is presented in order to group time series of maxima for pair of random variables recorded at different stations. The method is based on two kinds of dissimilarity measure. The first one involves the distance of the marginal distributions; the second one involves the distance between the respective copulas. In both cases, a Wasserstein metric is used. For the estimation, we rely on tools for extreme-value distributions and copulas.
Regionalization Methods for Compound Extremes Based on the Wasserstein Distance
Regina Castrovilli;Fabrizio Durante;Daniela Gallo;Gianfausto Salvadori
2025-01-01
Abstract
A clustering algorithm is presented in order to group time series of maxima for pair of random variables recorded at different stations. The method is based on two kinds of dissimilarity measure. The first one involves the distance of the marginal distributions; the second one involves the distance between the respective copulas. In both cases, a Wasserstein metric is used. For the estimation, we rely on tools for extreme-value distributions and copulas.File in questo prodotto:
| File | Dimensione | Formato | |
|---|---|---|---|
|
25_Castrovilli_et_al_SIS2024.pdf
non disponibili
Tipologia:
Versione editoriale
Licenza:
Copyright dell'editore
Dimensione
1.36 MB
Formato
Adobe PDF
|
1.36 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
