• Bayesian unmasking in linear m...

Result: Bayesian unmasking in linear models

Title:
Bayesian unmasking in linear models
Authors:
JUSTEL, Ana, PENA, Daniel
Source:
Computational statistics & data analysis. 36(1):69-84
Publisher Information:
Amsterdam: Elsevier Science, 2001.
Publication Year:
2001
Physical Description:
print, 19 ref
Original Material:
INIST-CNRS
Subject Terms:
Mathematics, Mathématiques, Sciences exactes et technologie, Exact sciences and technology, Sciences et techniques communes, Sciences and techniques of general use, Mathematiques, Mathematics, Probabilités et statistiques, Probability and statistics, Théorie des probabilités et processus stochastiques, Probability theory and stochastic processes, Analyse stochastique, Stochastic analysis, Processus de markov, Markov processes, Statistiques, Statistics, Théorie de la décision, Decision theory, Algorithme, Algorithm, Algoritmo, Apprentissage, Learning, Aprendizaje, Condition initiale, Initial condition, Condición inicial, Convergence, Convergencia, Diagnostic, Diagnosis, Diagnóstico, Décision Bayes, Bayes decision, Decisión Bayes, Décision statistique, Statistical decision, Decisión estadística, Echantillonnage Gibbs, Gibbs sampling, Muestreo Gibbs, Estimation Bayes, Bayes estimation, Estimación Bayes, Hétérogénéité, Heterogeneity, Heterogeneidad, Itération, Iteration, Iteracción, Loi a posteriori, Posterior distribution, Ley a posteriori, Masquage, Masking, Enmascaramiento, Matrice covariance, Covariance matrix, Matriz covariancia, Modèle linéaire, Linear model, Modelo lineal, Modèle régression, Regression model, Modelo regresión, Méthode adaptative, Adaptive method, Método adaptativo, Méthode itérative, Iterative method, Método iterativo, Méthode séquentielle, Sequential method, Método secuencial, Observation aberrante, Outlier, Observación aberrante, Performance, Rendimiento, Probabilité a posteriori, Posterior probability, Probabilidad a posteriori, Régression linéaire, Linear regression, Regresión lineal, Théorie décision, Decision theory, Teoría decisión, Théorie statistique, Statistical theory, Teoría estadística, Valeur propre, Eigenvalue, Valor propio, AGSA, Adaptive Gibbs sampling algorithm, Donnée type Rousseeuw, Rousseeuw type data, GSA, Gibbs sampling algorithm
Document Type:
Academic journal Article
File Description:
text
Language:
English
Author Affiliations:
Department of Mathematics, Universidad Autónoma de Madrid, Spain
Department of Statistics and Econometrics, Universidad Carlos III de Madrid, Spain
ISSN:
0167-9473
Access URL:
http://pascal-francis.inist.fr/vibad/index.php?action=search&terms=937246
Rights:
Copyright 2001 INIST-CNRS
CC BY 4.0
Sauf mention contraire ci-dessus, le contenu de cette notice bibliographique peut être utilisé dans le cadre d’une licence CC BY 4.0 Inist-CNRS / Unless otherwise stated above, the content of this bibliographic record may be used under a CC BY 4.0 licence by Inist-CNRS / A menos que se haya señalado antes, el contenido de este registro bibliográfico puede ser utilizado al amparo de una licencia CC BY 4.0 Inist-CNRS
Notes:
Mathematics
Accession Number:
edscal.937246
Database:
PASCAL Archive

Further Information

We propose a Bayesian procedure for multiple outlier detection in linear models which avoids the masking problem. The posterior probabilities of each data point being an outlier are estimated by using an adaptive learning Gibbs sampling method. The idea is to modify the initial conditions of the Gibbs sampler in order to visit the posterior distribution space in a reasonable number of iterations. To find an appropriate vector of initial values we consider the information extracted from the eigenstructure of the covariance matrix of a vector of latent variables. These variables are introduced in the model to capture the heterogeneity in the data. This procedure also overcomes the false convergence of the Gibbs sampling in problems with strong masking. Our proposal is illustrated with some of the examples most frequently used in the literature.