Title: Bayesian robustness in product partition models
Authors: Rosangela Loschi - Universidade Federal de Minas Gerais (Brazil) [presenting]
Jacqueline Alves Ferreira - Universidade Federal de Minas Gerais (Brazil)
Fabrizio Ruggeri - CNR - IMATI (Italy)
Abstract: The focus is on the robustness analysis of non-exchangeable product partition models (PPM), which are widely used for multiple change points detection. Bayesian robustness is usually concerned with the impact of perturbations in the prior distribution of the parameter of interest on its posterior inference. We consider multiplicative perturbations in the data distribution, as well as in the prior distribution of its parameters. As a novelty in the robust Bayesian and PPM literature, we introduce some sensitivity measures to examine how those perturbations are affecting the posterior inference about the number of clusters and their position, as well as the product estimates. We focus our analysis on the skew-normal class of distributions, thus building a PPM under skew-normality. We apply the proposed PPM to analyze a Brazilian tomato price data set.