Premium
Extreme value analysis within a parametric outlier detection framework
Author(s) -
Cabras S.,
Morales J.
Publication year - 2006
Publication title -
applied stochastic models in business and industry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.413
H-Index - 40
eISSN - 1526-4025
pISSN - 1524-1904
DOI - 10.1002/asmb.660
Subject(s) - outlier , extreme value theory , parametric statistics , computer science , surprise , anomaly detection , selection (genetic algorithm) , parametric model , measure (data warehouse) , sample (material) , sample size determination , statistics , model selection , econometrics , data mining , artificial intelligence , mathematics , physics , psychology , social psychology , thermodynamics
Threshold selection is a key aspect in extreme values analysis, especially when the sample size is small. The main idea underpinning this work is that extreme observations are assumed to be outliers of a specified parametric model. We propose a threshold selection method based on outlier detection using a suitable measure of surprise. Copyright © 2006 John Wiley & Sons, Ltd.