IRIS Uni Torinohttps://iris.unito.itIl sistema di repository digitale IRIS acquisisce, archivia, indicizza, conserva e rende accessibili prodotti digitali della ricerca.Thu, 20 Jan 2022 21:11:30 GMT2022-01-20T21:11:30Z10501Tests for Normality in Classes of Skew-Normal and Skew-t Alternativeshttp://hdl.handle.net/2318/61506Titolo: Tests for Normality in Classes of Skew-Normal and Skew-t Alternatives
Thu, 01 Jan 2009 00:00:00 GMThttp://hdl.handle.net/2318/615062009-01-01T00:00:00ZA note on Bayesian hypothesis testing for the scalar skew-normal distributionhttp://hdl.handle.net/2318/87878Titolo: A note on Bayesian hypothesis testing for the scalar skew-normal distribution
Fri, 01 Jan 2010 00:00:00 GMThttp://hdl.handle.net/2318/878782010-01-01T00:00:00ZDisclosure risk estimation via nonparametric log-linear modelshttp://hdl.handle.net/2318/122033Titolo: Disclosure risk estimation via nonparametric log-linear models
Abstract: A major concern in releasing microdata sets is protecting the privacy of individuals in the sample. Consider a data set in the form of a high-dimensional contingency table. If an individual belongs to a cell with small frequency, an intruder with certain knowledge about the individual may identify him and learn sensitive information about him in the data. To estimate the risk of such breach of confidentiality we introduce several nonparametric models which represent progressive extensions of the one adopted by Skinner and Holmes (1998). The latter is a Poisson model with rates modeled through a mixed effects log-linear model with normal random effects. In the first extension, we assume Dirichlet process random effects and, mimicking Skinner and Holmes (1998), we keep the fixed effects constant. Next, we relax the latter assumption and consider a model all effects of which are unknown. In both extended models the total mass parameter of the Dirichlet process is also unknown. The MCMC methods used for inference are extensively discussed. An application to real data concludes the article.
Sun, 01 Jan 2012 00:00:00 GMThttp://hdl.handle.net/2318/1220332012-01-01T00:00:00ZDivergence-based tests of statistical modelshttp://hdl.handle.net/2318/17729Titolo: Divergence-based tests of statistical models
Abstract: We recall the main characteristics of the Cressie and Read power divergence family and quickly sketch some extensions based on more general measures of divergence and disparity. Then a new family of Bayesian power divergence diagnostics is introduced.
Sun, 01 Jan 2006 00:00:00 GMThttp://hdl.handle.net/2318/177292006-01-01T00:00:00ZOn LBI invariant tests for normality in the class of exponential power distributionshttp://hdl.handle.net/2318/17730Titolo: On LBI invariant tests for normality in the class of exponential power distributions
Abstract: Questo lavoro presenta una derivazione alternativa del test invariante localmente ottimo (LBI) per la normalità nella classe delle normali di ordine r , piu' note come exponential power distributions. Rispetto ad altri metodi ricorrenti in letteratura, la strada qui seguita offre una prospettiva computazionale piu' semplice ed una interpretazione unitaria dei risultati ottenuti; si apre inoltre ad uno sviluppo bayesiano dell'analisi del problema della curtosi.
Sun, 01 Jan 2006 00:00:00 GMThttp://hdl.handle.net/2318/177302006-01-01T00:00:00ZA Dirichlet process elaboration diagnostic for binomial goodness of fithttp://hdl.handle.net/2318/60009Titolo: A Dirichlet process elaboration diagnostic for binomial goodness of fit
Abstract: Useful model checking tools can be constructed by measuring the distance between a prior distribution that concentrates most of its mass around a model of interest, and the resulting posterior distribution. In this paper we use this approach to construct a diagnostic measure for detecting lack of fit in discrete data, with special focus on binomial data. We begin by constructing a suitable probability model “around” the model of interest, via a Dirichlet Process elaboration. We derive the resulting diagnostic and show that, approximately, it is the sum of two terms: the first is the logarithm of the Bayes factor and the second is proportional to the Pearson chi-square statistics. We give details of a simulation algorithm for computing the diagnostic and illustrate its use in an application to biomedical data.
Thu, 01 Jan 1998 00:00:00 GMThttp://hdl.handle.net/2318/600091998-01-01T00:00:00ZRobust Bayesian Analysis given Priors on Partitions Setshttp://hdl.handle.net/2318/59588Titolo: Robust Bayesian Analysis given Priors on Partitions Sets
Abstract: In a Bayesian analysis, suppose that probability measures may be specified over the subsets partitioning the parameter space Ω, in such a way that they can be combined to form a unique prior measure, defined over all Ω, according to some weights. Should the weights be uncertain, then the class Г of all the probability measures compatible with such uncertainty is specified instead. Situations in which such a class Г is justified are presented and, in these cases, established and quite recent techniques in the field of robust Bayesian analysis are applied to Г. Bounds on posterior expectations are computed, as the prior measure varies in Г, whilst concentration functions and coefficients of divergence are considered when interest lies with comparing functional forms of measures in Г.
Sat, 01 Jan 1994 00:00:00 GMThttp://hdl.handle.net/2318/595881994-01-01T00:00:00ZStatistical models for species richness in the Ross Seahttp://hdl.handle.net/2318/1521392Titolo: Statistical models for species richness in the Ross Sea
Abstract: In recent years, a large international effort has been placed in compiling a complete list of Antarctic mollusc distributional records based both on historical occurrences, dating back to 1899, and on newly collected data. Such dataset is highly asymmetrical in the quality of contained information, due to the variety of sampling gears used and the amount of information recorded at each sampling station (e.g. sampling gear used, sieve mesh size used, etc.). This dataset stimulates to deploy all statistical potential in terms of data representation, estimation, clusterization and prediction.
In this paper we aim at selecting an appropriate statistical model for this dataset in order to explain species richness (i.e. the number of observed species) as a function of several covariates, such as gear used, latitude, etc.. Given the nature of data, we preliminary implement a Poisson regression model and we extend it with a Negative Binomial regression to manage over-dispersion. Generalized linear mixed models (GLMM) and generalized additive models (GAM) are also explored to capture a possible extra explicative power of the covariates. However, preliminary results under them suggest that more sophisticated models are needed. Therefore, we introduce a hierarchical Bayesian model, involving a nonparametric approach through the assumption of random effects with a Dirichlet Process prior.
Thu, 01 Jan 2015 00:00:00 GMThttp://hdl.handle.net/2318/15213922015-01-01T00:00:00ZA Bayesian test for normality against skewed alternatives.http://hdl.handle.net/2318/59868Titolo: A Bayesian test for normality against skewed alternatives.
Abstract: Under a suitable reparameterization we show that the locally best invariant test for normality in the class of skew-normal alternatives (Azzalini,1985), provided by Salvan (1986), represents an upper bound for the Jeffreys divergenge between prior and posterior distributions of the skewness parameter.
Tue, 01 Jan 2008 00:00:00 GMThttp://hdl.handle.net/2318/598682008-01-01T00:00:00ZBayesian Model Criticismhttp://hdl.handle.net/2318/114050Titolo: Bayesian Model Criticism
Abstract: We develop a framework for model criticism. Our approach is based on embedding a given model via the addition of an elaboration parameter. The value of an elaborated model is defined by the change in expected utility associated with updating the probability distribution on the elaboration parameter. This provides the experimenter with a numerical summary for model criticism. Our approach offers a unified framework for probing assumptions on which probability modelling is based. Model criticism is performed enterely within the Baysian framework. We explore relationships with other diagnostic measures, namely Fisher's score function, deviance and the Bayes factor
Fri, 01 Jan 1993 00:00:00 GMThttp://hdl.handle.net/2318/1140501993-01-01T00:00:00Z