Cari colleghi,
vi annuncio che martedì 31 marzo, alle ore 14 precise, nell'aula Beltrami del Dipartimento di Matematica dell'Universita' di Pavia,
Bernardo Nipoti (Dipartimento di Economia e Statistica - Università di Torino)
terrà un seminario dal titolo
Rediscovery of Good-Turing estimators via Bayesian Nonparametrics
Sunto:
The problem of estimating discovery probabilities originated in the context of statistical ecology, and in recent years it has become popular due to its frequent appearance in challenging applications arising from biology, bioinformatics, genetics, linguistics, etc. A full range of statistical approaches, parametric and nonparametric as well as frequentist and Bayesian, have been proposed for estimating discovery probabilities. We investigate the relationship between the celebrated Good-Turing approach, which is a frequentist nonparametric approach developed in the 1940s, and the Bayesian nonparametric approach recently introduced in the literature. Specifically, under the assumption of a two parameter Poisson-Dirichlet prior, we show that Bayesian nonparametric estimators of discovery probabilities are asymptotically equivalent to suitably smoothed Good-Turing estimators. As a by-product of this result, we present and discuss a methodology for deriving asymptotic credible intervals to be associated to the Bayesian nonparametric estimators of discovery probabilities.
Cordiali saluti, Raffaella Carbone