Show simple item record

dc.contributor.authorGneiting, Tilmann*
dc.contributor.authorBalabdaoui, Fadoua*
dc.contributor.authorRaftery, Adrian E.*
dc.date.accessioned2010-02-15T15:37:40Z
dc.date.available2010-02-15T15:37:40Z
dc.date.issued2007
dc.identifier.urihttps://basepub.dauphine.fr/handle/123456789/3466
dc.language.isoenen
dc.subjectProper scoring ruleen
dc.subjectProbability integral transformen
dc.subjectPrequential principleen
dc.subjectPredictive distributionen
dc.subjectPosterior predictive assessmenten
dc.subjectModel diagnosticsen
dc.subjectForecast verificationen
dc.subjectEx post evaluationen
dc.subjectEnsemble prediction systemen
dc.subjectDensity forecasten
dc.subjectCross-validationen
dc.subject.ddc519en
dc.titleProbabilistic forecasts, calibration and sharpnessen
dc.typeArticle accepté pour publication ou publié
dc.description.abstractenProbabilistic forecasts of a continuous variable take the form of predictive densities or predictive cumulative distribution functions. We propose a diagnostic approach to the evaluation of predictive performance that is based on the paradigm of {\sl maximizing the sharpness of the predictive distributions subject to calibration}. Calibration refers to the statistical consistency between the distributional forecasts and the observations and is a joint property of the predictions and the events that materialize. Sharpness refers to the concentration of the predictive distributions and is a property of the forecasts only. A simple theoretical framework phrased in terms of a game between nature and forecaster allows us to distinguish probabilistic calibration, exceedance calibration and marginal calibration. We propose and study tools for checking calibration and sharpness, among them the probability integral transform (PIT) histogram, marginal calibration plots, the sharpness diagram and proper scoring rules. The diagnostic approach is illustrated by an assessment and ranking of probabilistic forecasts of wind speed at the Stateline wind energy center in the U.S.~Pacific Northwest. In combination with cross-validation or in the time series context, our proposal provides very general, nonparametric alternatives to the use of information criteria for model diagnostics and model selection.en
dc.relation.isversionofjnlnameJournal of the Royal Statistical Society. Series B, Statistical Methodology
dc.relation.isversionofjnlvol69en
dc.relation.isversionofjnlissue2en
dc.relation.isversionofjnldate2007-04
dc.relation.isversionofjnlpages243-268en
dc.relation.isversionofdoihttp://dx.doi.org/10.1111/j.1467-9868.2007.00587.xen
dc.identifier.urlsitehttp://hal.archives-ouvertes.fr/hal-00363242/en/en
dc.description.sponsorshipprivateouien
dc.relation.isversionofjnlpublisherBlackwell Publishingen
dc.subject.ddclabelProbabilités et mathématiques appliquéesen
hal.person.labIds*
hal.person.labIds*
hal.person.labIds*


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record