Benutzer: Gast  Login
Titel:

Identifiability and estimation of recursive max-linear models

Dokumenttyp:
Zeitungsartikel
Autor(en):
Gissibl, N., Klüppelberg, C. and Lauritzen, S.
Abstract:
We address the identifiability and estimation of recursive max-linear structural equation models represented by an edge weighted directed acyclic graph (DAG). Such models are generally unidentifiable and we identify the whole class of DAGs and edge weights corresponding to a given observational distribution. For estimation, standard likelihood theory cannot be applied because the corresponding families of distributions are not dominated. Given the underlying DAG, we present an estimator for the...     »
Stichworte:
Causal inference, Bayesian network, directed acyclic graph, extreme value theory, generalized maximum likelihood estimation, graphical model, identifiability, max-linear model, structural equation model.
Dewey Dezimalklassifikation:
510 Mathematik
Zeitschriftentitel:
Scandinavian Journal of Statistics
Jahr:
2021
Band / Volume:
48
Jahr / Monat:
2021-03
Quartal:
1. Quartal
Monat:
Mar
Heft / Issue:
1
Seitenangaben Beitrag:
188-211
Sprache:
en
Volltext / DOI:
doi:10.1111/sjos.12446
Verlag / Institution:
Wiley
Hinweise:
online Veröffentlichung: 28. April 2020
Status:
Erstveröffentlichung
Publikationsdatum:
01.03.2021
TUM Einrichtung:
Lehrstuhl für Mathematische Statistik
Format:
Text
 BibTeX