Inferential methods based on pseudo-likelihoods

Cluster

Methodological and Experimental Statistics

Description

The complexity of statistical models used in applications poses a challenge to statistics. This research group aims to respond to the emerging needs in the various application fields. The common trait that unifies the different research topics of the group is the search for computationally easy solutions to solve problems that cannot be tackled with the usual statistical tools. In particular, approximations of the proper likelihood are sought in cases where the reference model is too complex, and then the objective is to obtain simple tools for inference and prediction. Aspects of particular interest for the group are the following: the study and definition of pseudo-likelihoods that allow the application of statistical procedures even in complex models or with large amounts of data; the study and definition of forecasting procedures based on pseudo-likelihoods, the definition of advanced computational methods and the specification of suitable combination of models.

Research subjects

Definizione di pseudo verosimiglianze e studio delle proprietà delle associate procedure inferenziali
Procedure predittive basate su pseudo verosimiglianze e su metodi computazionali avanzati
Combinazione di modelli