helio migon - Academia.edu (original) (raw)

Papers by helio migon

Procceedings of the 19th Brazilian Congress of Thermal Sciences and Engineering

Proceeding Series of the Brazilian Society of Computational and Applied Mathematics

Source identification methodologies use inverse problems techniques combined with a dispersion mo... more Source identification methodologies use inverse problems techniques combined with a dispersion model and observational data to estimate relevant source parameters. This work proposes a time-dependent model to estimate source parameters of multiple point releases. The forward problem or dispersion model accounts for the time variation of the wind field using a Fourier series that best fits the wind field time series of the experimental data. The source parameters are estimated by an adaptive Monte Carlo Markov Chain algorithm.

arXiv (Cornell University), Apr 7, 2021

This paper proposes a generalization of Gaussian mixture models, where the mixture weight is allo... more This paper proposes a generalization of Gaussian mixture models, where the mixture weight is allowed to behave as an unknown function of time. This model is capable of successfully capturing the features of the data, as demonstrated by simulated and real datasets. It can be useful in studies such as clustering, change-point and process control. In order to estimate the mixture weight function, we propose two new Bayesian nonlinear dynamic approaches for polynomial models, that can be extended to other problems involving polynomial nonlinear dynamic models. One of the methods, called here component-wise Metropolis-Hastings, apply the Metropolis-Hastings algorithm to each local level component of the state equation. It is more general and can be used in any situation where the observation and state equations are nonlinearly connected. The other method tends to be faster, but is applied specifically to binary data (using the probit link function). The performance of these methods of estimation, in the context of the proposed dynamic Gaussian mixture model, is evaluated through simulated datasets. Also, an application to an array Comparative Genomic Hybridization (aCGH) dataset from glioblastoma cancer illustrates our proposal, highlighting the ability of the method to detect chromosome aberrations.

Dynamic generalized linear models may be seen simultaneously as an extension to dynamic linear mo... more Dynamic generalized linear models may be seen simultaneously as an extension to dynamic linear models and to generalized linear models, formally treating serial auto-correlation inherent to responses observed through time. The present work revisits inference methods for this class, proposing an approach based on information geometry, focusing on the kkk- parametric exponential family. Among others, the proposed method accommodates multinomial and can be adapted to accommodate compositional responses on k=d+1k=d+1k=d+1 categories, while preserving the sequential aspect of the Bayesian inferential procedure, producing real-time inference. The updating scheme benefits from the conjugate structure in the exponential family, assuring computational efficiency. Concepts such as Kullback-Leibler divergence and the projection theorem are used in the development of the method, placing it close to recent approaches on variational inference. Applications to real data are presented, demonstrating the co...

Anais do(a) Encontro nacional de modelagem computacional e encontro de ciência e tecnologia de materiais, 2021

Proceedings of the 26th International Congress of Mechanical Engineering, 2021

Abstract: Bayesian inference in factor analytic models has received re-newed attention in recent ... more Abstract: Bayesian inference in factor analytic models has received re-newed attention in recent years, partly due to computational advances but also partly to applied focuses generating factor structures as exemplified by recent work in financial time series modeling. The focus of our current work is to investigate the commonly overlooked problem of prior specification and sensitivity in factor models. We accomplish that by implementing Pérez and Berger’s (1999). Expected Posterior (EP) prior distributions. As opposed to alternative objective priors, such as Jeffreys ’ prior and Bernardo’s prior, EP prior has several important theoretical and practical properties, with its straightforward computation through MCMC methods and coherence when comparing multiple models perhaps the most important ones. Key words: Bayes ’ factors; Bayesian inference; expected posterior prior; latent factor models; model selection criteria; model uncertainty. 1

Computational Statistics & Data Analysis, 2016

We develop a Bayesian framework for estimation and prediction of dynamic models for observations ... more We develop a Bayesian framework for estimation and prediction of dynamic models for observations from the two-parameter exponential family. Different link functions are introduced to model both the mean and the precision in the exponential family allowing the introduction of covariates and time series components. We explore conjugacy and analytical approximations under the class of partial specified models to keep the computation fast. The algorithm of West et al. (1985) is extended to cope with the twoparameter exponential family models. The methodological novelties are illustrated with two applications to real data. The first, considers unemployment rates in Brazil and the second some macroeconomic variables for the United Kingdom.

Applied Stochastic Models in Business and Industry, 2010

The modified mixture model with Markov switching volatility specification is introduced to analyz... more The modified mixture model with Markov switching volatility specification is introduced to analyze the relationship between stock return volatility and trading volume. We propose to construct an algorithm based on Markov chain Monte Carlo simulation methods to estimate all the parameters in the model using a Bayesian approach. The series of returns and trading volume of the British Petroleum stock will be analyzed. Copyright © 2009 John Wiley & Sons, Ltd.

Brazilian Review of Econometrics, 2021

This work investigates the effects of using the independent Jeffreys prior for the degrees of fre... more This work investigates the effects of using the independent Jeffreys prior for the degrees of freedom parameter of a t-student model in the asymmetric generalised autoregressive conditional heteroskedasticity (GARCH) model. To capture asymmetry in the reaction to past shocks, smooth transition models are assumed for the variance. We adopt the fully Bayesian approach for inference, prediction and model selection We discuss problems related to the estimation of degrees of freedom in the Student-t model and propose a solution based on independent Jeffreys priors which correct problems in the likelihood function. A simulated study is presented to investigate how the estimation of model parameters in the t-student GARCH model are affected by small sample sizes, prior distributions and misspecification regarding the sampling distribution. An application to the Dow Jones stock market data illustrates the usefulness of the asymmetric GARCH model with t-student errors.

arXiv: Statistics Theory, 2019

This paper proposes an alternative approach for constructing invariant Jeffreys prior distributio... more This paper proposes an alternative approach for constructing invariant Jeffreys prior distributions tailored for hierarchical or multilevel models. In particular, our proposal is based on a flexible decomposition of the Fisher information for hierarchical models which overcomes the marginalization step of the likelihood of model parameters. The Fisher information matrix for the hierarchical model is derived from the Hessian of the Kullback-Liebler (KL) divergence for the model in a neighborhood of the parameter value of interest. Properties of the KL divergence are used to prove the proposed decomposition. Our proposal takes advantage of the hierarchy and leads to an alternative way of computing Jeffreys priors for the hyperparameters and an upper bound for the prior information. While the Jeffreys prior gives the minimum information about parameters, the proposed bound gives an upper limit for the information put in any prior distribution. A prior with information above that limit ...

It is well-known from many works in small area that estimators based on hierarchical models perfo... more It is well-known from many works in small area that estimators based on hierarchical models perform significantly better than the ones which do not take advantage of between area variation, see for instance, Holt and Moura(1993). Recent research works, mainly in epidemiology and geography field, has been claiming the importance of explicitly modelling the spatial structure of the data, see for example Bernadinelly and Montomoli(1992). Small area data can also exhibit a spatial structure and therefore an attempt to using spatial model should be made. This paper presents a logistic hierarchical spatial model for estimation of a small area proportion . A evaluate study with real data is also presented and comparisons with logistic hierarchical models are made.

The bridge approach for regularization of coefficients in regression models uses ℓα norm, with α ... more The bridge approach for regularization of coefficients in regression models uses ℓα norm, with α ∈ (0, +∞), to define a penalization on large values of the regression coefficients. Particular cases include the lasso and ridge penalizations. In Bayesian models, the penalization is enforced by a prior distribution on the coefficients. Although MCMC approaches are available for Bayesian bridge regression, they can be very slow for large datasets, specially in high dimensions. This paper develops an implementation of Automatic Differentiation Variational Inference for Bayesian inference on semi-parametric regression models with bridge penalization. The non-parametric effects of covariates are modelled by B-splines. The proposed inference procedure allows the use of small batches of data at each iteration (due to stochastic gradient based updates), therefore drastically reducing computational time in comparison with MCMC. Full Bayesian inference is preserved so joint uncertainty estimate...

Journal of Hydrology, 2008

We present an analysis of runoff and rainfall data from Rio Grande, a basin located in the northe... more We present an analysis of runoff and rainfall data from Rio Grande, a basin located in the northeast of Brazil. The main challenges we face here are: (i) to model runoff and rainfall jointly, taking into account their different spatial units, (ii) to use stochastic models where all the parameters have physical interpretations, and (iii) to model

In this paper we introduce a new item response theory (IRT) model with a generalized Student t-li... more In this paper we introduce a new item response theory (IRT) model with a generalized Student t-link function with unknown degrees of freedom (df), named generalized t-link (GtL) IRT model. In this model we consider only the difficulty parameter in the item response function. GtL is an alternative to the two parameter logit and probit models, since the degrees of freedom (df) play a similar role to the discrimination parameter. However, the behavior of the curves of the GtL is different from those of the two parameter models, since in GtL the curve obtained from different df's can cross each other in more than one latent trait level. The GtL model has similar proprieties to the generalized linear mixed models, such as the existence of sufficient statistics and easy parameter interpretation. Also, many techniques of parameter estimation, model fit assessment and residual analysis developed for that models can be used for the GtL model. We develop fully Bayesian estimation and model fit assessment tools through a Metropolis-Hastings step within Gibbs sampling algorithm. We consider a prior sensitivity choice concerning the degrees of freedom. The simulation study indicates that the algorithm recovers all parameters properly. In addition, some Bayesian model fit assessment tools are considered. Finally, a real data set is analyzed using our approach and other usual models. The results indicate that our model fits the data better than the two parameter models.

Brazilian Journal of Probability and Statistics, 2012

In this work, we develop Bayesian analysis based on the Jeffreys prior for the hyperbolic family ... more In this work, we develop Bayesian analysis based on the Jeffreys prior for the hyperbolic family of distributions. It is usually difficult to estimate the four parameters in this class: to be reliable the maximum likelihood estimator typically requires large sample sizes of the order of thousands of observations. Moreover, improper prior distributions may lead to improper posterior distributions, whereas proper prior distributions may dominate the analysis. Here, we show through a simulation study that Bayesian methods based on Jeffreys prior provide reliable point and interval estimators. Moreover, this simulation study shows that for the absolute loss function Bayesian estimators compare favorably to maximum likelihood estimators. Finally, we illustrate with an application to real data that our methodology allows for parameter estimation with remarkable good properties even for a small sample size.

Journal of the American Statistical Association, 1985

Bayesian Forecasting MIKE WEST, P. JEFF HARRISON, and HELIO S. MIGON* Dynamic Bayesian models are... more Bayesian Forecasting MIKE WEST, P. JEFF HARRISON, and HELIO S. MIGON* Dynamic Bayesian models are developed for application in nonlinear, non-normal time series and regression problems, providing dynamic extensions of standard generalized linear models. A key feature of the analysis is the use of conjugate prior and posterior distributions for the exponential family parameters. This leads to the calculation of closed, standard-form predictive distributions for forecasting and model criticism. The structure of the models depends on the time evolution of underlying state variables, and the feedback of observational information to these variables is achieved using linear Bayesian prediction methods. Data analytic aspects of the models concerning scale parameters and outliers are discussed, and some applications are provided.