Conferência 9: Objective Bayesian Analysis for Generalized Normal Distribution with Application on the Height of Brazilian Eucalyptus Clones

Vera Tomazella, DEs-UFSCar

Abstract

This work presents the Objective Bayesian approach for the estimation of the parameters of the generalized normal model. Simulation studies were performed to analyze the frequentist properties of credible intervals from the reference posterior distribution. The proposed methodology is illustrated in a real data set.

Conferência 8: Lifetime model for multivariate survival data with a surviving fraction

Vicente G. Cancho, ICMC-USP

Abstract

In this paper we propose a new lifetime model for multivariate survival data with a surviving fraction. We develop this model assuming that there are m types of unobservable competing risks, where each risk is related to a time of the occurrence of an event of interest. We explore the use of Markov chain Monte Carlo (MCMC) methods to develop a Bayesian analysis for the proposed model. We also perform a simulation study in order to analyze the frequentist coverage probabilities of credible interval derived from posteriors. Our modeling is illustrated through a real data set.

Conferência 7: A Bayesian Approach to Zero-Modified Poisson Model for the Prediction of Match Outcomes: An Application to the 2012-2013 La Liga Season

Adriano K. Suzuki, ICMC-USP

Abstract

In any sports competition, strong interest is devoted to the knowledge on the team that will be champion. The result of a match, the chance of a team being either qualified for a specific tournament, or relegated, the best attack and defense are also subjects of interest. This paper presents a Bayesian methodology for modeling the number of goals scored by a team based on Zero-Modified Poisson distribution. Inference procedures and computational simulation studies are also discussed. The proposed methodology was applied to the 2012-13 La Liga and the results were compared with those of Poisson model through the De Finetti measure and percentage of correct predictions.

Conferência 6: Zero-Modified Regression Models

Katiane S. Conceição, ICMC-USP

Abstract

In this work, we present a family of distributions for count data, the so called Zero- Modified Power Series, which is an extension of the Power Series distributions family whose support starts at zero. This extension consists in modifying the probability of observing zero of each Power Series distribution, allowing the new zero-modified distribution appro- priately accommodate datasets which have any amount of zero observations (for instance, zero-inflated or zero-deflated datasets). Power Series distributions included in the Zero- Modified Power Series family are: Poisson, Generalized Poisson, Geometric, Binomial, Negative Binomial and Generalized Negative Binomial. We introduce the Zero-Modified Power Series regression models and propose a Bayesian approach considering information matrix priori. Two real datasets, corresponding to leptospirosis notifications in cities of Bahia State at Brazil, are analyzed. We emphasize that the proposed Zero-Modified Power Series family distributions and their regression versions can accommodate sets of count data without any previous knowledge about the characteristic of zero-inflation (-deflation) present in the dataset.