Trabalhos Aprovados

 

Atenção: As comunicações foram provisoriamente aceitas para apresentação no 18o SINAPE. Apenas os trabalhos em que pelo menos um dos autores tiver, até 16/06, pago o boleto de inscrição do congresso terão a aceitação definitiva, os demais não poderão ser apresentados no evento. Verifique se o seu trabalho satisfaz esse requisito.

 

Filtrar por categoria/formato/palavra:


 
Formato Título
Oral
SKEW SCALE MIXTURE OF NORMAL DISTRIBUTIONS: PROPERTIES AND ESTIMATION
Modelos de Regressão
Autor 1: Clécio da Silva Ferreira (UNICAMP)
clecio@unicamp.br
Autor 2: Heleno Bolfarine (IME/USP)
hbolfar@ime.usp.br
Autor 3: Víctor Hugo Lachos (IMECC/UNICAMP)
hlachos@ime.unicamp.br
Abstract:
Scale mixture of normal distributions are often used as a challenging family for statistical procedures of symmetrical data. In this article, we have defined a skewed version of these distributions and we have derived several of its probabilistic and inferential properties. The main virtue of the members of this family of distributions is that they are easy to simulate from and they also supply genuine EM algorithms for maximum likelihood estimation. For univariate skewed responses, the EM-type algorithm has been discussed with emphasis on the skew-t, skew-slash, skew-contaminated normal and skew-exponential power distributions. Some simplifying and unifying results are also noted with the Fisher informating matrix, which is derived in closed form for some distributions in the family. Results obtained from simulated and real data sets are reported illustrating the usefulness of the proposed methodology. The main conclusion in reanalyzing a data set previously studied is that the models so far entertained are clearly not the most adequate ones.
Pôster
Skew-Normal/Independent Distribution in Linear Mixed Models for Bivariate Response
Modelos de Regressão
Autor 1: Lourdes Contreras Montenegro (UFMG)
lourdes@est.ufmg.br
Autor 2: Victor Hugo Lachos (UNICAMP)
hlachos@ime.unicamp.br
Autor 3: Pulak Ghosh (Georgia State Univer)
pghosh@mathstat.gsu.edu
Abstract:
Correlated data arise in longitudinal studies from epidemiological and clinical research. Generally in the longitudinal data setting is assumed that the random effects and errors are normally distributed. In this work, we developed a bivariate linear mixed model and relax the normality assumption by using a skew-normal/independent distribution. We illustrate the procedure using a real data.
Oral
Smoothed Design-adapted Haar Wavelet Regression with Correlated Errors
Séries Temporais
Autor 1: Rogerio de Faria Porto (IME-USP)
rdporto@ime.usp.br
Autor 2: Pedro Alberto Morettin (IME-USP)
pam@ime.usp.br
Autor 3: Donald B Percival (U.Washington,USA)
dbp@uwashington.edu
Abstract:
We present some results on non-parametric regression models in the presence of autocorrelated errors, using design-adapted Haar wavelets smoothed through a lifting scheme. The autocovariances are sup- posed to be absolutely bounded by the variance of the errors. We also comment how these results can be used to show that a soft-threshold estimator almost achieves a ideal risk in the wavelet domain. Keywords: autocorrelation; non-parametric estimation; smoothing; denoising; lifting; design-adapted wavelets. 2000 Mathematics Subject Classifications: 62M10;
Oral
State Space Markov Switching Models Using Wavelets
Séries Temporais
Autor 1: Airlane Pereira Alencar (IME-USP)
lane@ime.usp.br
Autor 2: Pedro Alberto Morettin (IME-USP)
pam@ime.usp.br
Autor 3: Clelia M. Castro Toloi (IME-USO)
clelia@ime.usp
Abstract:
We propose a state space model with Markov switching, whose regimes are associated with the model parameters and regime transition probabilities are time-dependent. The estimation is based on the maximum likelihood method using the EM algorithm and a bootstrap method is proposed in order to obtain the distribution of the maximum likelihood estimators. To evaluate the state variables and regime probabilities, the Kalman filter and a probability filter procedure conditional on each possible regime, at each instant, are used. These procedures are illustrated with simulated data and the United States monthly industrial production index from January 1965 to August 2007.
Pôster
STOPPING AND SELECTION CRITERIA FOR\r\nSTOPPING AND SELECTION CRITERIA FOR EDGEWORTH EXPANSION
Estatística Computacional
Autor 1: Silvio Rodrigues de Faria Junior (IME-USP)
silvio@ime.usp.br
Autor 2: Julio Michael Stern (IME-USP)
jstern@ime.usp.br
Abstract:
The normal distribution plays a important role in various fields of science. This\r\nis quite natural since the sum of ramdon variables tends to a normal distribution\r\nwhen the quite general conditions of the central limit theorem are satisfied. The\r\nasymptotic Edgeworth expansion, a serie of Chebyshev-Hermite polinomials,is used\r\nfor weakly non-normal distributions. This work describes stopping and selection\r\ncriteria for Edgeworth type expansions.
Oral
Structural heteroscedastic measurement error models: modeling quasar X-ray emission versus Black hole mass
Modelos de Regressão
Autor 1: Alexandre Galvão Patriota (IME-USP)
patriota@ime.usp.br
Autor 2: Heleno Bolfarine (IME-USP)
hbolfar@ime.usp.br
Abstract:
Recently, linear measurement error models have been applied to relate functions of the quasar X-ray emission and black hole masses. However, the linear assumption is not always tenable. In this work, we present a polynomial model in which we embrace on the corrected score approach to estimate the model parameters. An application to a real dataset obtained from the Chandra Observatory is reported. We also simulate the rejection rates for the Wald statistic in order to study test size and power for small and moderate samples, indicating that the test behaves satisfactorily in those situations.
Pôster
Study of Simulation to Assess the Bias Correction in Generalized Nonlinear Models with Dispersion
Inferência Estatística
Autor 1: Margareth Cizuka Udo (UEM)
mctudo@uem.br
Autor 2: Rosangela Getirana Santana (UEM)
rgsantana@uem.br
Autor 3: Isolde Santos Previdelli (UEM)
iprevidelli@uem.br
Autor 4: Gauss Moutinho Cordeiro (UFRE)
gauss@ufrpe.br
Abstract:
We derive general formulae for the second-order biases of maximum likelihood estimates of the parameters in generalized nonlinear models with dispersion co- variates. This result generalizes previous work by Botter and Cordeiro (1998) and Cordeiro and McCullagh (1991). The practical use of such bias corrections and their assessment is illustrated in a simulation study. Keywords: Bias correction; Dispersion parameter; Generalized nonlinear model; Maximum likelihood; Precision parameter. 1
Pôster
SYMMETRY STUDIES OF CHIRALITY IN SHORT SYMBOLIC SEQUENCES
Estatística em Agronomia e Biologia
Autor 1: DEVANIL JAQUES DE SOUZA (UFLA-Lavras-MG)
djs@oi.com.br
Autor 2: LUCAS MONTEIRO CHAVES (UFLA-Lavras-MG)
Lucas@ufla.br
Autor 3: MARLOS VIANA (UNIV ILLINOIS-USA)
marlos.viana@gmail.com
Abstract:
In this report we apply the methods of symmetry studies to describe the behavior of certain reference DNA sequences. The genomes are studied with basis on the dihedral (left) orbits associated with selected DNA words in length of three and four and are described in terms of the canonical invariant characteristic of the chiral properties of the genomes.
Pôster
TAXA DE DESEMPREGO PARA O SEXO FEMININO NA CIDADE DE RECIFE
Séries Temporais
Autor 1: Rejane Cristina dos Santos Silva (UFMG)
rejaneufmg@yahoo.com.br
Autor 2: Natália Karoline Silva (UFMG)
naty_karol@yahoo.com.br
Autor 3: Ela Mercedes Toscano (UFMG)
mercedes@est.ufmg.br
Abstract:
Uma série temporal é uma seqüência de dados obtidos em intervalos regulares de tempo durante um período específico. Este conjunto pode ser obtido através de observações periódicas do evento de interesse, podendo ser trabalhado com diversas técnicas estatísticas. Este trabalho é dedicado à análise da série Taxa de desemprego para o sexo feminino na cidade de Recife, no período de janeiro de 1990 a dezembro de 1995 com periodicidade mensal. Este banco de dados foi obtido no site do Instituto Brasileiro de Geografia e Estatística (IBGE) e tem como finalidade a aplicação e comparação das diversas técnicas de modelagem de séries temporais. Os modelos ajustados à série foram Regressão, Médias Móveis, Alisamento Exponencial, Decomposição e ARIMA. Para a escolha do melhor modelo foram utilizadas as estatísticas alternativas (AIC e Shwarz) e as estatísticas dos erros de previsão (MAPE, MAD e MSD). Os resultados mostraram ser os modelos ARIMA(0,1,0)(0,1,1)12 e Alisamento Exponencial Multiplicativo de Winter os melhores ajustados. Porém, para o cálculo das previsões, o melhor modelo obtido foi o ARIMA(1,0,0)(1,0,0). Portanto, podemos concluir que obter um modelo que se ajusta satisfatoriamente aos dados de uma série temporal, nem sempre significa encontrar um modelo que produz as melhores previsões. A escolha do modelo a ser utilizado depende do foco em estudo. Toda a análise foi realizada com o auxílio do Software Minitab 14. Palavras–chave: Séries temporais, Modelo ARIMA, Modelo de Alisamento Exponencial, Previsão.
Pôster
Teoria das Evidências Aplicada na Inteligência de um Jogo Educacional do Tipo RPG
Estatística Computacional
Autor 1: José Cavalcanti Netto (UFPB)
netto.cavalcanti@gmail.com
Autor 2: Liliane dos Santos Machado (UFPB)
liliane@di.ufpb.br
Autor 3: Ronei Marcos Moraes (UFPB)
ronei@de.ufpb.br
Abstract:
Jogos eletrônicos podem ser vistos como uma ferramenta para o entretenimento do usuário e possuem técnicas e tecnologias que envolvem um caráter multidisciplinar dos seus desenvolvedores. A utilização de jogos eletrônicos para o ensino deram origem aos Jogos Educacionais (JEs), onde a idéia de aprendizado sobrepuja a de entretenimento. A Inteligência Artificial, através dos diversos modelos existentes, pode ser utilizada em JEs, proporcionando ao aluno uma forma lúdica de aprendizado, além de melhorar as suas habilidades. A Teoria de Dempster-Shaffer, ou Teoria das Evidências usa a distribuição de Crenças que por sua vez usa uma estrutura de discernimento, definida por um conjunto exaustivo de eventos mutuamente exclusivos. Três funções são utilizadas nesta teoria: a chamada massa de probabilidade, a Crença e a Plausibilidade. A partir da função massa de probabilidade, os valores superior (Plausibilidade) e inferior (Crença) de um intervalo podem ser definidos, onde está contida a verdadeira probabilidade do subconjunto em questão. A utilização de uma sistema inteligente (SI) baseado na Teoria das Evidências em um JE, permite que as tomadas de decisões tornem-se dinâmicas evitando a monotonia, além de permitir avaliar o grau de Crença do aprendizado do aluno sobre os conceitos abordados. Outra função do SI é a manipulação do cenário do jogo, permitindo reforçar conceitos fracamente assimilados. Um JE para o aprendizado de conceitos de geografia para alunos da 4ª e 5ª série do ensino fundamental foi implementado para testar o sistema inteligente concebido, com sucesso.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54