{"id":2294,"date":"2025-06-29T12:38:02","date_gmt":"2025-06-29T15:38:02","guid":{"rendered":"https:\/\/www.ufrgs.br\/ppgest\/?page_id=2294"},"modified":"2025-12-23T15:19:59","modified_gmt":"2025-12-23T18:19:59","slug":"2025-2","status":"publish","type":"page","link":"https:\/\/www.ufrgs.br\/ppgest\/2025-2\/","title":{"rendered":"2025"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-page\" data-elementor-id=\"2294\" class=\"elementor elementor-2294\">\n\t\t\t\t\t\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-d8fd15f elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"d8fd15f\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-175a695\" data-id=\"175a695\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-20eca3d elementor-widget elementor-widget-heading\" data-id=\"20eca3d\" data-element_type=\"widget\" data-widget_type=\"heading.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.6.2 - 04-04-2022 *\/\n.elementor-heading-title{padding:0;margin:0;line-height:1}.elementor-widget-heading .elementor-heading-title[class*=elementor-size-]>a{color:inherit;font-size:inherit;line-height:inherit}.elementor-widget-heading .elementor-heading-title.elementor-size-small{font-size:15px}.elementor-widget-heading .elementor-heading-title.elementor-size-medium{font-size:19px}.elementor-widget-heading .elementor-heading-title.elementor-size-large{font-size:29px}.elementor-widget-heading .elementor-heading-title.elementor-size-xl{font-size:39px}.elementor-widget-heading .elementor-heading-title.elementor-size-xxl{font-size:59px}<\/style><h2 class=\"elementor-heading-title elementor-size-default\">2025<\/h2>\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-b30f441 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"b30f441\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-06ba395\" data-id=\"06ba395\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap elementor-element-populated\">\n\t\t\t\t\t\t\t\t<div class=\"elementor-element elementor-element-72969d9 elementor-widget elementor-widget-text-editor\" data-id=\"72969d9\" data-element_type=\"widget\" data-widget_type=\"text-editor.default\">\n\t\t\t\t<div class=\"elementor-widget-container\">\n\t\t\t<style>\/*! elementor - v3.6.2 - 04-04-2022 *\/\n.elementor-widget-text-editor.elementor-drop-cap-view-stacked .elementor-drop-cap{background-color:#818a91;color:#fff}.elementor-widget-text-editor.elementor-drop-cap-view-framed .elementor-drop-cap{color:#818a91;border:3px solid;background-color:transparent}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap{margin-top:8px}.elementor-widget-text-editor:not(.elementor-drop-cap-view-default) .elementor-drop-cap-letter{width:1em;height:1em}.elementor-widget-text-editor .elementor-drop-cap{float:left;text-align:center;line-height:1;font-size:50px}.elementor-widget-text-editor .elementor-drop-cap-letter{display:inline-block}<\/style>\t\t\t\t<p><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700973\"  tabindex=\"0\" title=\"Semin\u00e1rio 03\/06\/2025. Profa. Dra. Denise Britz do Nascimento Silva - CI\u00caNCIAS\"    >Semin\u00e1rio 03\/06\/2025. Profa. Dra. Denise Britz do Nascimento Silva &#8211; CI\u00caNCIAS<\/span><div id=\"target-id6ae5cdf700973\" class=\"collapseomatic_content \"><\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> Produ\u00e7\u00e3o de indicadores do mercado de trabalho com modelos de s\u00e9ries temporais para repetidas pesquisas.<\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> V\u00e1rias das s\u00e9ries temporais utilizadas por diversos setores da sociedade s\u00e3o provenientes de pesquisas repetidas no tempo. Tais pesquisas, que em geral investigam as mesmas unidades em diversos benef\u00edcios, s\u00e3o direcionadas \u00e0 produ\u00e7\u00e3o de informa\u00e7\u00f5es que permitem o estudo da evolu\u00e7\u00e3o de uma popula\u00e7\u00e3o ou de uma frequ\u00eancia ao longo do tempo. Diante dos desafios de manuten\u00e7\u00e3o e amplia\u00e7\u00e3o do sistema de estat\u00edsticas oficiais, investiga-se a produ\u00e7\u00e3o de indicadores do mercado de trabalho sob uma abordagem de investiga\u00e7\u00e3o baseada em modelos de s\u00e9ries temporais para a Pesquisa Nacional por Amostra de Domic\u00edlios Cont\u00ednua do IBGE. A metodologia compreende os modelos estruturais de s\u00e9ries temporais no formato de espa\u00e7o de estados em suas formula\u00e7\u00f5es univariadas e multivariadas. Esses s\u00e3o empregados para estimar componentes n\u00e3o observ\u00e1veis, como tend\u00eancia e sazonalidade, considerando a presen\u00e7a de erros amostrais e a estrutura de brilho impl\u00edcita no desenho da pesquisa. Adicionalmente, os modelos propostos lidam com a integra\u00e7\u00e3o de dados de fontes alternativas ao incorporar s\u00e9ries do Google Trends de palavras relacionadas \u00e0 busca de emprego (big data).<\/span><\/span><\/span><\/span><\/p>\n<p><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">&nbsp;O trabalho \u00e9 resultado da tese de doutorado de Caio Gon\u00e7alves (Funda\u00e7\u00e3o Jo\u00e3o Pinheiro) no Programa de P\u00f3s-Gradua\u00e7\u00e3o da ENCE, no \u00e2mbito de um projeto de pesquisa em parceria com Jan van den Brakel (Statistics Netherlands e Maastrich University) e Luna Hidalgo (Consultora, mestre pela ENCE).<\/span><\/span><\/span><\/span><\/p>\n<p><\/div><\/p>\n<p><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span class=\"collapseomatic \" id=\"id6ae5cdf7009ad\"  tabindex=\"0\" title=\"Semin\u00e1rio 24\/06\/2025. Prof. Dr. Alu\u00edsio de Souza Pinheiro - UNICAMP\"    >Semin\u00e1rio 24\/06\/2025. Prof. Dr. Alu\u00edsio de Souza Pinheiro &#8211; UNICAMP<\/span><div id=\"target-id6ae5cdf7009ad\" class=\"collapseomatic_content \"><\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> Triagem de caracter\u00edsticas wavelet em problemas de regress\u00e3o de alta dimens\u00e3o.<\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> Na an\u00e1lise de dados modernos, o n\u00famero p de vari\u00e1veis explicativas \u00e9 comumente muito maior que o tamanho da amostra n. Alguns m\u00e9todos cl\u00e1ssicos para lidar com esse tipo de problema s\u00e3o covari\u00e1veis de subamostragem, regress\u00e3o de Ridge e Lasso (Friedman et al., 2009). Al\u00e9m do alto custo computacional, um fator muito maior que tamb\u00e9m pode afetar o desempenho dos m\u00e9todos de sele\u00e7\u00e3o de vari\u00e1veis. Isso motiva um rastreamento de covari\u00e1veis.&nbsp;<\/span><\/span><\/span><\/span><span style=\"font-style: inherit; font-weight: inherit;\">Portanto, um rastreamento inicial de covari\u00e1veis relevantes \u00e9 comum em modelos de alta dimensionalidade. Uma quest\u00e3o importante aqui \u00e9 que muitas caracter\u00edsticas importantes podem ter uma rela\u00e7\u00e3o relevante, embora altamente n\u00e3o linear, com a resposta. Propomos um m\u00e9todo de triagem de caracter\u00edsticas wavelet que pode ser facilmente implementado e provamos que, sob condi\u00e7\u00f5es adequadas, ele captura como verdadeiras covari\u00e1veis com alta probabilidade. Os resultados da simula\u00e7\u00e3o tamb\u00e9m mostram que nossa abordagem supera outros m\u00e9todos de triagem em modelos altamente n\u00e3o lineares. Aplicamos o rastreamento de caracter\u00edsticas a dois conjuntos de dados sobre concentra\u00e7\u00e3o de oz\u00f4nio e epilepsia. Em ambas as aplica\u00e7\u00f5es, o m\u00e9todo proposto seleciona caracter\u00edsticas encontradas nas literaturas de seus respectivos campos de pesquisa, ilustrando a aplicabilidade do rastreamento de caracter\u00edsticas.<\/span><\/p>\n<p><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Este \u00e9 um trabalho conjunto com Rodney Fonseca e Pedro Morettin.<\/span><\/span><\/span><\/span><\/p>\n<p><\/div><\/p>\n<p><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span class=\"collapseomatic \" id=\"id6ae5cdf7009c3\"  tabindex=\"0\" title=\"Semin\u00e1rio 01\/07\/2025. Profa. Dra. Taiane Schaedler Prass - PPGEST UFRGS\"    >Semin\u00e1rio 01\/07\/2025. Profa. Dra. Taiane Schaedler Prass &#8211; PPGEST UFRGS<\/span><div id=\"target-id6ae5cdf7009c3\" class=\"collapseomatic_content \"><\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> Integrando o tratamento de dados ausentes em \u00e1rvores de regress\u00e3o probabil\u00edstica: o pacote R PRTree.<\/span><\/span><\/span><\/span><\/p>\n<p><strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><\/span><\/span><\/strong><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"> \u00c1rvores de Regress\u00e3o Probabil\u00edstica (PRTrees) estendem \u00e1rvores de decis\u00e3o tradicionais ao incorporar fun\u00e7\u00f5es de probabilidade que atribuem pontos de dados a m\u00faltiplas regi\u00f5es da \u00e1rvore, permitindo limites de decis\u00e3o mais suaves e abrangentes. Neste trabalho, propomos uma adapta\u00e7\u00e3o de PRTrees que considere explicitamente os valores ausentes nas covari\u00e1veis, aprimorando sua aplicabilidade a conjuntos de dados incompletos do mundo real. Implementamos essa metodologia no pacote R PRTree, combinando R com rotinas FORTRAN e C de baixo n\u00edvel para obter ganhos substanciais em efici\u00eancia computacional. Como resultado, o pacote oferece desempenho significativamente superior \u00e0s implementa\u00e7\u00f5es em R puro. Nesta, delineamos os fundamentos te\u00f3ricos do m\u00e9todo, descrevemos os detalhes da implementa\u00e7\u00e3o e ilustramos seu desempenho por meio de estudos de simula\u00e7\u00e3o.<\/span><\/span><\/span><\/span><\/p>\n<p><\/div><\/p>\n<p><span class=\"collapseomatic \" id=\"id6ae5cdf7009d5\"  tabindex=\"0\" title=\"Semin\u00e1rio 08\/07\/2025. Prof. Lucas S. Sibemberg - PPGMAp \u2013 UFRGS,&nbsp; Chief Artificial Intelligence Office\"    >Semin\u00e1rio 08\/07\/2025. Prof. Lucas S. Sibemberg &#8211; PPGMAp \u2013 UFRGS,&nbsp; Chief Artificial Intelligence Office<\/span><div id=\"target-id6ae5cdf7009d5\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Clusteriza\u00e7\u00e3o Espectral em Grafos Geom\u00e9tricos<\/p>\n<p style=\"font-size: 16px; font-weight: 400;\"><span style=\"font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><span dir=\"auto\" style=\"vertical-align: inherit;\">&nbsp;Este trabalho estuda o problema de clusteriza\u00e7\u00e3o no <i>Soft Geometric Block Model<\/i> com n\u00famero arbitr\u00e1rio de clusters. Nesse modelo, propomos um algoritmo espectral para a recupera\u00e7\u00e3o dos clusters. Al\u00e9m disso, estabelecemos resultados de converg\u00eancia em probabilidade que garantem a acur\u00e1cia do algoritmo. Por fim, apresentamos experimentos num\u00e9ricos que demonstram sua acur\u00e1cia mesmo em grafos pequenos.<\/span><\/p>\n<p><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf7009e4\"  tabindex=\"0\" title=\"Semin\u00e1rio 22\/07\/2025. &nbsp;Prof. Dr. Diego Ramos Canterle&nbsp; \u2013 UFRGS\"    >Semin\u00e1rio 22\/07\/2025. &nbsp;Prof. Dr. Diego Ramos Canterle&nbsp; \u2013 UFRGS<\/span><div id=\"target-id6ae5cdf7009e4\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Robust estimation in a multivariate normal regression model with general parameterization<\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><span dir=\"auto\" style=\"font-size: 16px; font-style: normal; font-weight: 400; vertical-align: inherit;\">&nbsp;<\/span>This work explores the use of a non-extensive q-entropy measure for parameter estimation in a multivariate normal regression model with general parameterization. This model encompasses a large class of regression models that includes mixed models, errors-in-variables models, nonlinear models, among others. The resulting estimator, called the maximum L_q-likelihood estimator, depends on a tuning parameter q that is chosen to improve inferential accuracy in the presence of outliers. We derive the quantities required to obtain the parameter estimates and the asymptotic variance and covariance matrix of the maximum L_q-likelihood estimator under this general class of models. Some of these quantities are explained in detail for three specific cases, namely, mixed models, errors-in-variables models, and nonlinear models. Although the maximum L_q-likelihood estimator may not be consistent, we obtain a new estimator that satisfies this property through a calibration function.<\/p>\n<p>Additionally, we also obtain another robust and consistent estimator that minimizes an empirical version of the density power divergence, called the minimum density power divergence estimator.&nbsp;<span style=\"font-style: inherit; font-weight: inherit;\">To evaluate the performance of our proposal, we conduct Monte Carlo simulations in the absence and presence of outliers under finite samples. Finally, we apply our method to actual data.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf7009f3\"  tabindex=\"0\" title=\"Semin\u00e1rio 29\/07\/2025. &nbsp;Prof. Dr. Valderio A. Reisen \u2013 Centrale Sup\u00e9lec-Universit\u00e9 Paris-Sud (Fran\u00e7a), Unicamp (SP, Brasil), e UFES (ES, Brasil)\"    >Semin\u00e1rio 29\/07\/2025. &nbsp;Prof. Dr. Valderio A. Reisen \u2013 Centrale Sup\u00e9lec-Universit\u00e9 Paris-Sud (Fran\u00e7a), Unicamp (SP, Brasil), e UFES (ES, Brasil)<\/span><div id=\"target-id6ae5cdf7009f3\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>M-robust approach to estimate Seasonal INAR processes and outlier detection based on Wavelets and the application to the daily air&nbsp;<span style=\"font-style: inherit;\">quality index data for the pollutant Carbon.<\/span><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><span dir=\"auto\" style=\"vertical-align: inherit;\">&nbsp;<\/span><\/span>This study investigates the detection of outliers in INAR(1)s models using an approach based on wavelet transforms and robust parameter estimation. Initially, the fundamental concepts of the INAR(1)s model are reviewed, followed by the proposal of a new detection method that combines the Discrete Wavelet Transform (DWT) and the Maximal Overlap Discrete Wavelet Transform (MODWT) with a robust version of conditional least squares (CLS) based on M-estimators. The Seasonal Discrete Wavelet Transform (SDWT) is introduced as an extension of the DWT, adapted to handle seasonality in count time series. The proposed method uses thresholds based on the empirical distribution of the&nbsp;<span style=\"font-style: inherit; font-weight: inherit;\">detail coefficients, enhancing the identification of outlier values. The effectiveness of the procedure was&nbsp;<\/span><span style=\"font-style: inherit; font-weight: inherit;\">evaluated through Monte Carlo simulations, comparing the percentage of correctly detected outliers. The practical application, exemplified by air quality index data, demonstrates the relevance and applicability of the method, offering a robust and efficient approach for outlier detection in seasonal daily air&nbsp;<\/span><span style=\"font-style: inherit; font-weight: inherit;\">quality index data for the pollutant Carbon.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a01\"  tabindex=\"0\" title=\"Semin\u00e1rio 05\/08\/2025. &nbsp;Prof. Dr. Danilo Marcondes Filho - PPGEst UFRGS\"    >Semin\u00e1rio 05\/08\/2025. &nbsp;Prof. Dr. Danilo Marcondes Filho &#8211; PPGEst UFRGS<\/span><div id=\"target-id6ae5cdf700a01\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Cartas de Controle para Dados Autocorrelacionados: Abordagens via colagem de S\u00e9ries Temporais e Bootstrapping Param\u00e9trico.<\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Boa parte das estrat\u00e9gias para controle estat\u00edstico de processos industriais com dados que apresentam depend\u00eancia temporal baseia-se no monitoramento de res\u00edduos gerados por modelos ajustados. Embora essa abordagem seja eficaz na&nbsp;<span style=\"font-style: inherit; font-weight: inherit;\">detec\u00e7\u00e3o de dist\u00farbios que afetam a tend\u00eancia ou a m\u00e9dia do processo (como drifts), ela se mostra pouco sens\u00edvel a altera\u00e7\u00f5es na pr\u00f3pria estrutura de depend\u00eancia dos dados. Este trabalho prop\u00f5e abordagens para o monitoramento de observa\u00e7\u00f5es futuras, a L passos a frente, baseada na compara\u00e7\u00e3o com simula\u00e7\u00f5es do processo sob a hip\u00f3tese de que ele permanece sob controle a L passos a frente. Seguindo a ideia da colagem de s\u00e9ries temporais e do bootstrapping param\u00e9trico, s\u00e3o obtidas distribui\u00e7\u00f5es a L passos a frente para dois indicadores: (i) as estimativas dos coeficientes do modelo ajustado e (ii) a estat\u00edstica Q de Box-Ljung (avalia correla\u00e7\u00e3o serial nos res\u00edduos). Novas observa\u00e7\u00f5es s\u00e3o ent\u00e3o monitoradas comparando-se suas estat\u00edsticas \u00e0s distribui\u00e7\u00f5es emp\u00edricas geradas. Simula\u00e7\u00f5es preliminares, baseadas em dados com estrutura de depend\u00eancia gerada por modelos da classe ARMA, indicam resultados promissores quanto a sensibilidade da abordagem na detec\u00e7\u00e3o de altera\u00e7\u00f5es na correla\u00e7\u00e3o serial. Esses resultados ser\u00e3o apresentados e discutidos.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a0e\"  tabindex=\"0\" title=\"Semin\u00e1rio 12\/08\/2025. &nbsp;Profa. Dra. Gabriela Bettella Cybis\"    >Semin\u00e1rio 12\/08\/2025. &nbsp;Profa. Dra. Gabriela Bettella Cybis<\/span><div id=\"target-id6ae5cdf700a0e\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Filodin\u00e2mica e corre\u00e7\u00e3o de vis\u00e3o amostral para estudos de dispers\u00e3o viral baseados em biosequ\u00eancias.<\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Uma combina\u00e7\u00e3o filodin\u00e2mica de dados sequenciais gen\u00e9ticos virais com informa\u00e7\u00f5es geogr\u00e1ficas e temporais para fazer infer\u00eancia sobre processos de interesse epidemiol\u00f3gico. Quando o foco \u00e9 geogr\u00e1fico, buscamos inferir taxas de migra\u00e7\u00e3o e padr\u00f5es de dispers\u00e3o espacial ao longo da hist\u00f3ria evolutiva dos organismos. Esses modelos s\u00e3o amplamente utilizados na investiga\u00e7\u00e3o de epidemias virais, como COVID-19, influenza e HIV, possibilitando reconstruir a origem geogr\u00e1fica dos surtos, reconstruir sua propaga\u00e7\u00e3o espacial e identificar fatores determinantes da migra\u00e7\u00e3o viral. Entretanto, \u00e9 amplamente reconhecido que modelos discretos de difus\u00e3o filogeogr\u00e1fica bayesiana s\u00e3o sens\u00edveis a vis\u00f5es amostrais, especialmente em configura\u00e7\u00f5es onde limita\u00e7\u00f5es t\u00e9cnicas e de recursos exigem a representatividade espacial das amostras em rela\u00e7\u00e3o \u00e0 preval\u00eancia real. Uma das abordagens mais comuns para minimizar esse impacto \u00e9 realizar a subamostragem dos dados observados. Para contornar esse problema, propomos o Bias-correcting Subsampling Trait Model (BSTM), uma nova abordagem para o c\u00e1lculo da verossimilhan\u00e7a em processos com estados discretos sobre \u00e1rvores filogen\u00e9ticas. A BSTM busca corrigir a vis\u00e3o amostral sem a necessidade de descartar dados, utilizando um modelo de mistura baseado em m\u00faltiplas \u00e1rvores subamostradas, ponderadas por informa\u00e7\u00f5es externas sobre frequ\u00eancias populacionais reais. Al\u00e9m disso, derivamos uma express\u00e3o anal\u00edtica para a verossimilhan\u00e7a do modelo, resultando em ganhos significativos de efici\u00eancia computacional. Simula\u00e7\u00f5es de Monte Carlo s\u00e3o propostas para avaliar o modelo.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a1c\"  tabindex=\"0\" title=\"Semin\u00e1rio 19\/08\/2025.&nbsp; Prof. Dr. Cleber Bisognin\"    >Semin\u00e1rio 19\/08\/2025.&nbsp; Prof. Dr. Cleber Bisognin<\/span><div id=\"target-id6ae5cdf700a1c\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Distribui\u00e7\u00e3o Marshall-Olkin Cosseno Fr\u00e9chet: Distribui\u00e7\u00e3o e Modelo de Regress\u00e3o.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>A distribui\u00e7\u00e3o Fr\u00e9chet foi introduzida no contexto da an\u00e1lise de valores extremos. Tal distribui\u00e7\u00e3o possui um \u00fanico par\u00e2metro e suporte nos reais positivos. \u00c9 amplamente aplicada em dados de desastres naturais, como precipita\u00e7\u00f5es m\u00e1ximas anuais e polui\u00e7\u00e3o do ar, e na an\u00e1lise de risco na \u00e1rea de finan\u00e7as. O presente trabalho prop\u00f5e uma nova reparametriza\u00e7\u00e3o da distribui\u00e7\u00e3o Fr\u00e9chet por meio do gerador Marshall-Olkin Cosseno, denotada por MOCF(\u00b5, \u03c3), para \u00b5, \u03c3, y &gt; 0. Al\u00e9m disso, define-se um novo modelo de regress\u00e3o quant\u00edlica baseado na distribui\u00e7\u00e3o MOCF. Para definir o novo modelo de regress\u00e3o, reparametriza-se a fun\u00e7\u00e3o densidade de probabilidade da distribui\u00e7\u00e3o MOCF em termos do seu quantil \u00b5. A estrutura de regress\u00e3o \u00e9 inclu\u00edda atrav\u00e9s da rela\u00e7\u00e3o g(\u00b5t) = x\u22a4t \u03b2, para t = 1, . . . , n, em que g(\u00b7) = log(\u00b7) \u00e9 a fun\u00e7\u00e3o de liga\u00e7\u00e3o. Para estima\u00e7\u00e3o do vetor de par\u00e2metros dos modelos propostos, utilizou-se o m\u00e9todo de m\u00e1xima verossimilhan\u00e7a (EMV). O desempenho do EMV foi avaliado via simula\u00e7\u00f5es de Monte Carlo, para dferentes cen\u00e1rios. As medidas utilizadas na avalia\u00e7\u00e3o do EMV foram a m\u00e9dia, vi\u00e9s, erro padr\u00e3o, erro quadr\u00e1tico m\u00e9dio (EQM), assimetria e curtose. Atrav\u00e9s da an\u00e1lise dos resultados das simula\u00e7\u00f5es, verificou-se que o estimador apresenta ind\u00edcios de que s\u00e3o assintoticamente n\u00e3o viesados, consistentes, al\u00e9m de apresentar distribui\u00e7\u00e3o assintoticamente normal. Medidas de diagn\u00f3stico para o modelo e aplica\u00e7\u00f5es a dados reais est\u00e3o sendo realizados.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a28\"  tabindex=\"0\" title=\"Semin\u00e1rios 26\/08\/2025 - Discentes\"    >Semin\u00e1rios 26\/08\/2025 &#8211; Discentes<\/span><div id=\"target-id6ae5cdf700a28\" class=\"collapseomatic_content \"><\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\">Palestrantes:<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><ol>\n<li style=\"font-size: 16px; font-style: normal;\">Renato Augusto Esp\u00edndola Susin&nbsp;\u2013 \u201cAvalia\u00e7\u00e3o de Testes de Martingale Difference para amostras pequenas.\u201d<\/li>\n<li>Rafaela Oliveira da Silva \u2013 \u201cCensura Dependente e Fra\u00e7\u00e3o de Cura utilizando C\u00f3pula de Frank e Distribui\u00e7\u00e3o Defectiva Gompertz.\u201d<\/li>\n<li style=\"font-size: 16px; font-style: normal; font-weight: 400;\">Maicon Michael Fridrich Gottselig \u2013 \u201cModelo de Cura para Riscos Competitivos Correlacionados e N\u00e3o Proporcionais.\u201d<\/li>\n<\/ol>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span dir=\"auto\" style=\"vertical-align: inherit;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a34\"  tabindex=\"0\" title=\"Semin\u00e1rios 09\/09\/2025 - Discentes\"    >Semin\u00e1rios 09\/09\/2025 &#8211; Discentes<\/span><div id=\"target-id6ae5cdf700a34\" class=\"collapseomatic_content \"><\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\">Palestrantes:<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><ol>\n<li style=\"font-size: 16px; font-style: normal; font-weight: 400;\">Andressa de Oliveira Dorneles \u2013 \u201cMultistep machine learning for time series forecasting.\u201d<\/li>\n<li style=\"font-size: 16px; font-style: normal; font-weight: 400;\">Zahri da Paz \u2013 \u201cPredi\u00e7\u00e3o de S\u00e9ries Temporais com dados faltantes por GRU-D e Fun\u00e7\u00e3o de Perda Assim\u00e9trica.\u201d<\/li>\n<li style=\"font-size: 16px; font-style: normal; font-weight: 400;\">Jo\u00e3o Manuel M\u00e1quina \u2013 \u201cCartas de Controle para Monitoramento Online de Dados Autocorrelacionados: Abordagens via Coleta de S\u00e9ries Temporais e Bootstrapping Param\u00e9trico.\u201d<\/li>\n<li>Roger Prochnow Moreira da Silva \u2013&nbsp;&#8220;Aprendizado Semi-Supervisionado com Dados Desbalanceados: An\u00e1lise e Proposi\u00e7\u00e3o de Estrat\u00e9gias.&#8221;<\/li>\n<\/ol>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a40\"  tabindex=\"0\" title=\"Semin\u00e1rio 30\/09\/2025.&nbsp; Prof. Dr. Guilherme Pumi\"    >Semin\u00e1rio 30\/09\/2025.&nbsp; Prof. Dr. Guilherme Pumi<\/span><div id=\"target-id6ae5cdf700a40\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span><span style=\"font-size: 1rem;\">Estima\u00e7\u00e3o de Par\u00e2metros em Modelos Observation-Driven com Dados Ausentes.<\/span><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Lidar com dados ausentes em s\u00e9ries temporais \u00e9 um problema complexo devido \u00e0 presen\u00e7a de depend\u00eancia temporal. M\u00e9todos de imputa\u00e7\u00e3o de uso geral, embora amplamante empregados, costumam distorcer propriedades estat\u00edsticas fundamentais dos dados, como a vari\u00e2ncia e a estrutura de depend\u00eancia, resultando em estimativadas viesadas e infer\u00eancias equivocadas. Esses problemas tornam-se ainda mais pronunciados no contexto de s\u00e9ries temporais, onde capturar a depend\u00eancia serial do processo \u00e9 fundamental para a estima\u00e7\u00e3o de par\u00e2metros. T\u00e9cnicas tradicionais de imputa\u00e7\u00e3o n\u00e3o conseguem preservar a din\u00e2nica da s\u00e9rie. Nesta apresenta\u00e7\u00e3o, discutiremos a estima\u00e7\u00e3o de modelos tipo observation-driven (ODM) na presen\u00e7a de dados ausentes. Propomos um novo m\u00e9todo de imputa\u00e7\u00e3o m\u00faltipla, especificamente desenvolvido para o contexto de ODM, que explora a natureza iterativa do componente sistem\u00e1tico desses modelos para propagar a estrutura de depend\u00eancia atrav\u00e9s dos modelos para propagar a estrutura de depend\u00eancia atrav\u00e9s dos dados faltantes, minimizando seu impacto na estima\u00e7\u00e3o. Diferentemente das t\u00e9cnicas tradicionais de imputa\u00e7\u00e3o, o m\u00e9todo proposto acomoda dados cont\u00ednuos, discretos e de tipo misto, preservando as principais propriedades de distribui\u00e7\u00e3o e depend\u00eancia. Derivados ainda suas propriedades assint\u00f3ticas. A aplica\u00e7\u00e3o do m\u00e9todo ser\u00e1 ilustrada no contexto de modelos GARMA.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a4d\"  tabindex=\"0\" title=\"Semin\u00e1rio 07\/10\/2025.&nbsp; Prof. Dr. Bruno Iochins Grisci\"    >Semin\u00e1rio 07\/10\/2025.&nbsp; Prof. Dr. Bruno Iochins Grisci<\/span><div id=\"target-id6ae5cdf700a4d\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Entre Vieses de Dados e Visualiza\u00e7\u00e3o de Import\u00e2ncia: Novos Caminhos para Sele\u00e7\u00e3o de Atributos em Alta Dimens\u00e3o.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>A sele\u00e7\u00e3o de atributos \u00e9 uma etapa fundamental em an\u00e1lises estat\u00edsticas e de aprendizado de m\u00e1quina aplicadas a dados de alta dimensionalidade, como aqueles oriundos de express\u00e3o g\u00eanica. Neste semin\u00e1rio, discutiremos dois t\u00f3picos nesse contexto: (i) uma an\u00e1lise cr\u00edtica sobre 20 anos de uso de conjuntos de dados de express\u00e3o g\u00eanica em estudos de sele\u00e7\u00e3o de atributos, evidenciando vieses hist\u00f3ricos, limita\u00e7\u00f5es de reprodutibilidade e o impacto do reuso de dados desatualizados nos resultados obtidos; e (ii) o weighted t-SNE, uma extens\u00e3o do m\u00e9todo cl\u00e1ssico de visualiza\u00e7\u00e3o voltada para incorporar a import\u00e2ncia dos atributos. Essa abordagem permite visualizar e comparar diferentes m\u00e9todos de atribui\u00e7\u00e3o de relev\u00e2ncia, auxiliando na escolha de algoritmos mais adequados e na interpretabilidade de modelos.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a59\"  tabindex=\"0\" title=\"Semin\u00e1rio 13\/10\/2025.&nbsp;Prof. Dr. Diego de Castro Fettermann\"    >Semin\u00e1rio 13\/10\/2025.&nbsp;Prof. Dr. Diego de Castro Fettermann<\/span><div id=\"target-id6ae5cdf700a59\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Acceptance of m-health by police officers and firefighters in Santa Catarina, Brazil.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Profissionais de seguran\u00e7a p\u00fablica trabalham em ambientes de alta tens\u00e3o, principalmente em pa\u00edses com altos \u00edndices de criminalidade como o Brasil. Esse ambiente de trabalho exp\u00f5e os profissionais a riscos f\u00edsicos, qu\u00edmicos, biol\u00f3gicos e psicol\u00f3gicos. As tecnologias de IoT incorporadas aos dispositivos m-health s\u00e3o uma alternativa eficiente para melhorar a qualidade de vida de policiais e bombeiros, pois permitem o monitoramento cont\u00ednuo de sinais fisiol\u00f3gicos e podem ajudar a prevenir problemas de sa\u00fade entre esses profissionais. Para isso, compreender a aceita\u00e7\u00e3o dessas tecnologias por agentes de seguran\u00e7a p\u00fablica \u00e9 crucial para uma implementa\u00e7\u00e3o bem-sucedida. Esta pesquisa foi desenvolvida junto ao Departamento de Promo\u00e7\u00e3o da Sa\u00fade e Bem-estar da Secretaria de Seguran\u00e7a P\u00fablica do estado de Santa Catarina\/SC. O objetivo desta pesquisa busca investigar os fatores de determinam a aceita\u00e7\u00e3o e as funcionalidades priorit\u00e1rias dos dispositivos m-health por parte dos policiais militares e bombeiros de SC. Esta pesquisa foi dividida em dois subprojetos, o primeiro busca identificar a aceita\u00e7\u00e3o dos dispositivos m-health propondo um modelo adaptado da Teoria Unificada de Aceita\u00e7\u00e3o e Uso de Tecnologia (UTAUT), sendo este modelo testado por meio da abordagem de Modelagem de Equa\u00e7\u00f5es Estruturais Multigrupo. O segundo busca priorizar as funcionalidades do dispositivo m-health e utiliza uma abordagem que considera a heterogeneidade da amostra por meio da t\u00e9cnica de Stated Preference e da estima\u00e7\u00e3o dos modelos Conditional Multinomial Logit e Latent Class Logit. A amostra utilizada em ambos projetos consiste em 687 respostas v\u00e1lidas de um question\u00e1rio, sendo 517 de policiais militares e 170 de bombeiros de SC. Os resultados encontrados apoiam a personaliza\u00e7\u00e3o de programas educacionais para informar os profissionais sobre os benef\u00edcios das tecnologias de m-health. Al\u00e9m disso, \u00e9 proposta uma plataforma modular que prioriza as funcionalidades do dispositivo m-health que considera as necessidades dos policiais e bombeiros de SC.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a67\"  tabindex=\"0\" title=\"Semin\u00e1rio 21\/10\/2025. Prof\u00aa. Dr\u00aa. Paula T\u00f3foli\"    >Semin\u00e1rio 21\/10\/2025. Prof\u00aa. Dr\u00aa. Paula T\u00f3foli<\/span><div id=\"target-id6ae5cdf700a67\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Can AI beat a naive portfolio? An experiment with anonymized data.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Realized volatility series are generally found to exhibit very slowly decaying autocorrelation functions, suggesting the use of long-memory models. The linear HAR regression model is well known for its ability to parsimoniously model the long-memory behavior of realized volatility. Besides, there is vast empirical evidence of its good forecasting performance. We propose an alternative approach to the linear HAR model, considering a dynamic vine copula based regression capable of capturing long-memory and nonlinearity in the time series dependence of realized volatility, and it also allows for the possibility of changes in the dependence structure over time. More specifically, we derive a regression from a dynamic D-vine decomposition of the joint density of the dependent and explanatory variables of the HAR model and compute point forecasts of volatility for the one-day horizon from this regression model. We evaluate the forecasting ability of the dynamic HAR-D-vine model in an empirical study. Our data sets consist of realized volatilities from the S&amp;P 500, NASDAQ 100, DOW JONES, FTSE 100, CAC and HANG SENG stock indexes. Regarding the main results, we find that both static and dynamic D-vine copula-based HAR regressions and the linear HAR model have a statistically equivalent forecast performance, while they outperform the bivariate copula model.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a73\"  tabindex=\"0\" title=\"Semin\u00e1rio 04\/11\/2025.&nbsp;Prof. Dr. Marcelo Perlin (EA\/UFRGS)\"    >Semin\u00e1rio 04\/11\/2025.&nbsp;Prof. Dr. Marcelo Perlin (EA\/UFRGS)<\/span><div id=\"target-id6ae5cdf700a73\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Forecasting Volatility via Dynamic Copula-Based Regressions<span style=\"font-size: 1rem;\">.<\/span><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Using anonymized data from the United States (U.S.) market, we evaluate the performance of Google\u2019s main LLM (Large Language Model) Gemini 1.5 Flash in making investment decisions. Unlike other studies, we query the LLM for different investment horizons (1 to 36 months) and types of financial information (financial data, price data, and a combination of both). Running a total of 30,000 simulations for 1,522 companies over 20 years of data, we find that Gemini does not consistently outperform a naive portfolio and the S&amp;P 500 index in terms of returns and Sharpe ratios. Additionally, our findings indicate a decline in risk adjusted investment performance as the investment horizon extends.<\/p>\n<p>Joint work with Cristian R. Foguesatto, Fernanda M. M\u00fcller and Marcelo B. Righi.<span style=\"font-size: 16px; font-style: normal; font-weight: 400;\">.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a84\"  tabindex=\"0\" title=\"Semin\u00e1rio 11\/11\/2025.&nbsp;Tatiane Fontana Ribeiro (UFSM)\"    >Semin\u00e1rio 11\/11\/2025.&nbsp;Tatiane Fontana Ribeiro (UFSM)<\/span><div id=\"target-id6ae5cdf700a84\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Bivariate generalized autoregressive models for forecasting bivariate non-Gaussian times series.<\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>This paper introduces a novel approach, the bivariate generalized autoregressive (BGAR) model, for modeling and forecasting bivariate time series data. The BGAR model generalizes the bivariate vector autoregressive (VAR) models by allowing data that does not necessarily follow a normal distribution. We consider a random vector of two time series and assume each belongs to the canonical exponential family, similarly to the univariate generalized autoregressive moving average (GARMA) model. We include autoregressive terms of one series into the dynamical structure of the other and vice versa. The model parameters are estimated using the conditional maximum likelihood (CML) method. We provide general closed-form expressions for the conditional score vector and conditional Fisher information matrix, encompassing all canonical exponential family distributions. We develop asymptotic confidence intervals and hypothesis tests. We discuss techniques for model selection, residual diagnostic analysis, and forecasting. We carry out Monte Carlo simulation studies to evaluate the performance of the finite sample CML inferences, including point and interval estimation. An application to real data analyzes the number of leptospirosis cases on hospitalizations due to leptospirosis in S\u00e3o Paulo state, Brazil. Competing models such as GARMA, autoregressive integrated moving average (ARIMA), and VAR models are considered for comparison purposes. The new model outperforms the competing models by providing more accurate out-of-sample forecasting and allowing quantification of the lagged effect of the case count series on hospitalizations due to leptospirosis.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700a90\"  tabindex=\"0\" title=\"Semin\u00e1rio 18\/11\/2025. Prof\u00aa Dr\u00aa Cira Etheowalda Guevara Otiniano (UnB)\"    >Semin\u00e1rio 18\/11\/2025. Prof\u00aa Dr\u00aa Cira Etheowalda Guevara Otiniano (UnB)<\/span><div id=\"target-id6ae5cdf700a90\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span><span style=\"font-size: 1rem;\">Modeling Extreme Data with Bimodal Structure Through Regression.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span><span style=\"text-align: justify; font-size: 1rem;\">We introduce a new regression model based on an extension of the Generalized Extreme Value (GEV) distribution \u2014 called the bimodal GEV distribution. Unlike the classical GEV, this model can represent data with two modes and heavy tails, offering much greater flexibility. We built a regression framework linked to the median of the distribution, making interpretation more intuitive. Parameters are estimated using maximum likelihood, and performance is evaluated through Monte Carlo simulations. Finally, applying the model to real climate data from Brazil shows that it accurately captures extreme and bimodal patterns, providing a robust and efficient alternative to traditional mixture models.<\/span><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700aa0\"  tabindex=\"0\" title=\"Semin\u00e1rio 25\/11\/2025.&nbsp;Prof. Dr. Paulo Henrique Ferreira da Silva (UFBA)\"    >Semin\u00e1rio 25\/11\/2025.&nbsp;Prof. Dr. Paulo Henrique Ferreira da Silva (UFBA)<\/span><div id=\"target-id6ae5cdf700aa0\" class=\"collapseomatic_content \"><\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Sensor Monitoring Towards Industry 4.0: Bridging Concepts on the cn Multivariate Control Chart.<\/p>\n<p><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>The biggest challenge in monitoring multiple sensors is related to the high dimensionality that a process can easily obtain. Additionally, Statistical Process Control (SPC) techniques have difficulty in simultaneously detecting change of quality characteristics when extrapolated from the normality (or Gaussian) assumption. That is, the most known SPC multivariate control chart is based on Hotelling\u2019s T2 statistic. Moreover, it is assumed that the process data follow a multivariate normal distribution (which in practice rarely occurs) and marginals are supposed to be uncorrelated. This study focused on the SPC multivariate case (dimension d \u2265 2), regardless of the data dependence or marginal asymmetry. Thus, we proposed a new control chart for embedding change detection in the multivariate process, called the cn-chart, which is analogous to the tolerance region based on the density level set from a copula function. Furthermore, this methodology quantifies the correlation (not necessarily linear) between the response variables. Two study cases were adopted: one investigated the water quality control task, and the other the brass-steel thermostats monitoring relationship. Our results show that both association and anomaly detection are captured adopting the proposed methodology, which also allows the inclusion of the variable selection as conditional statistical modeling.<\/p>\n<p>Joint work with Diego Carvalho do Nascimento (NEOMA Business School-Fran\u00e7a), Francisco Louzada (ICMC-USP) and Ana Claudia da Silva Batista (DEst-UFBA).<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700ad0\"  tabindex=\"0\" title=\"Semin\u00e1rio 02\/12\/2025. Prof. Dr. Fl\u00e1vio Bambirra Gon\u00e7alves (UFMG)\"    >Semin\u00e1rio 02\/12\/2025. Prof. Dr. Fl\u00e1vio Bambirra Gon\u00e7alves (UFMG)<\/span><div id=\"target-id6ae5cdf700ad0\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>The Poisson-Gaussian Mixture Process: A Flexible and Robust Approach for Non-Gaussian Geostatistical Modeling.<\/p>\n<p style=\"\"><span style=\"font-size: 16px; font-style: normal; font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>This work introduces a novel family of geostatistical models designed to capture complex features beyond the reach of traditional Gaussian processes. The proposed family, termed the Poisson-Gaussian Mixture Process (POGAMP), is hierarchically specified, combining the infinite-dimensional dynamics of Gaussian processes with any multivariate continuous distribution. This combination is stochastically defined by a latent Poisson process, allowing the POGAMP to define valid processes with finite-dimensional distributions that can approximate any continuous distribution. Unlike other non-Gaussian geostatistical models that may fail to ensure validity of the processes by assigning arbitrary finite-dimensional distributions, the POGAMP preserves essential probabilistic properties crucial for both modeling and inference.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700adf\"  tabindex=\"0\" title=\"Semin\u00e1rio 09\/12\/2025. Prof. Dr. Fernando Arturo Pe\u00f1a Ramirez (UFSM)\"    >Semin\u00e1rio 09\/12\/2025. Prof. Dr. Fernando Arturo Pe\u00f1a Ramirez (UFSM)<\/span><div id=\"target-id6ae5cdf700adf\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Um Modelo de S\u00e9ries Temporais Maxwell com Parametriza\u00e7\u00e3o na M\u00e9dia para Dados Positivos.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Este trabalho introduz uma nova abordagem para modelar s\u00e9ries temporais positivas e cont\u00ednuas baseada na distribui\u00e7\u00e3o de Maxwell. O modelo proposto utiliza uma reparametriza\u00e7\u00e3o dessa distribui\u00e7\u00e3o, na qual seu par\u00e2metro de escala representa diretamente a m\u00e9dia. Nessa formula\u00e7\u00e3o, a m\u00e9dia condicional \u00e9 modelada por meio de uma estrutura din\u00e2mica que combina componentes autorregressivos e de m\u00e9dia m\u00f3vel, vinculados por uma fun\u00e7\u00e3o de liga\u00e7\u00e3o apropriada. A estima\u00e7\u00e3o dos par\u00e2metros \u00e9 realizada pelo m\u00e9todo da m\u00e1xima verossimilhan\u00e7a condicional, para o qual s\u00e3o derivadas express\u00f5es matriciais em forma fechada do vetor score condicional e da matriz de informa\u00e7\u00e3o de Fisher. Com base nas propriedades assint\u00f3ticas dos estimadores, s\u00e3o apresentados procedimentos para estima\u00e7\u00e3o intervalar e testes de hip\u00f3tese. Por meio de simula\u00e7\u00f5es de Monte Carlo, avalia-se o desempenho em amostras finitas e apresenta-se evid\u00eancia da converg\u00eancia dos estimadores para os valores verdadeiros dos par\u00e2metros \u00e0 medida que o tamanho amostral aumenta. Uma aplica\u00e7\u00e3o emp\u00edrica com dados de velocidade do vento em Bras\u00edlia, capital do Brasil, demonstra a relev\u00e2ncia pr\u00e1tica e a efic\u00e1cia do modelo proposto para o ajuste e a previs\u00e3o de s\u00e9ries temporais reais.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span class=\"collapseomatic \" id=\"id6ae5cdf700aea\"  tabindex=\"0\" title=\"Semin\u00e1rio 17\/12\/2025. Prof\u00aa. Dr\u00aa. Ana J\u00falia Alves C\u00e2mara (UFES)\"    >Semin\u00e1rio 17\/12\/2025. Prof\u00aa. Dr\u00aa. Ana J\u00falia Alves C\u00e2mara (UFES)<\/span><div id=\"target-id6ae5cdf700aea\" class=\"collapseomatic_content \"><\/p>\n<p style=\"font-weight: 400; font-style: normal; font-size: 16px;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">T\u00edtulo:&nbsp;<\/span><\/span>Modelagem de processos de contagem e deriva\u00e7\u00f5es: Aplica\u00e7\u00f5es em dados ambientais e epidemiol\u00f3gicos.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><span style=\"font-weight: bold;\"><span dir=\"auto\" style=\"vertical-align: inherit;\">Resumo:&nbsp;<\/span><\/span>Neste semin\u00e1rio, ser\u00e3o apresentados os resultados centrais de uma linha de pesquisa dedicada \u00e0 modelagem estat\u00edstica de s\u00e9ries temporais de contagens na epidemiologia ambiental. A investiga\u00e7\u00e3o parte da necessidade de aprimorar a an\u00e1lise de processos n\u00e3o Gaussianos e autocorrelacionados que descrevem a din\u00e2mica de morbidade e mortalidade associada \u00e0 exposi\u00e7\u00e3o a poluentes atmosf\u00e9ricos. O primeiro estudo prop\u00f5e o modelo GAM-ARMA, que combina fun\u00e7\u00f5es splines a uma estrutura autorregressiva de m\u00e9dias m\u00f3veis, permitindo capturar simultaneamente n\u00e3o linearidades e depend\u00eancia temporal. O segundo integra o modelo GLARMA a diferentes abordagens de bootstrap, aprimorando a infer\u00eancia do risco relativo em s\u00e9ries com poucas observa\u00e7\u00f5es. O terceiro introduz um modelo GLARMA robusto, baseado em M-estimadores, com o objetivo de reduzir o impacto de observa\u00e7\u00f5es at\u00edpicas em covari\u00e1veis ambientais. As aplica\u00e7\u00f5es com dados das regi\u00f5es metropolitanas de Belo Horizonte\/MG e Vit\u00f3ria\/ES demonstram ganhos expressivos em ajuste, estabilidade e validade inferencial. Em conjunto, os trabalhos ampliam o uso dos modelos GLARMA, articulando robustez, flexibilidade e rigor estat\u00edstico na an\u00e1lise de fen\u00f4menos complexos de sa\u00fade e meio ambiente.<\/p>\n<p style=\"font-size: 16px; font-style: normal; font-weight: 400;\">\n<\/p><p style=\"font-size: 16px; font-style: normal; font-weight: 400;\"><\/div><\/p>\t\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t<section class=\"elementor-section elementor-top-section elementor-element elementor-element-f5a9e54 elementor-section-boxed elementor-section-height-default elementor-section-height-default\" data-id=\"f5a9e54\" data-element_type=\"section\">\n\t\t\t\t\t\t<div class=\"elementor-container elementor-column-gap-default\">\n\t\t\t\t\t<div class=\"elementor-column elementor-col-100 elementor-top-column elementor-element elementor-element-9d8af07\" data-id=\"9d8af07\" data-element_type=\"column\">\n\t\t\t<div class=\"elementor-widget-wrap\">\n\t\t\t\t\t\t\t\t\t<\/div>\n\t\t<\/div>\n\t\t\t\t\t\t\t<\/div>\n\t\t<\/section>\n\t\t\t\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>2025<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"site-sidebar-layout":"no-sidebar","site-content-layout":"page-builder","ast-global-header-display":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"disabled","ast-breadcrumbs-content":"","ast-featured-img":"disabled","footer-sml-layout":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","footnotes":""},"_links":{"self":[{"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/pages\/2294"}],"collection":[{"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/comments?post=2294"}],"version-history":[{"count":63,"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/pages\/2294\/revisions"}],"predecessor-version":[{"id":2634,"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/pages\/2294\/revisions\/2634"}],"wp:attachment":[{"href":"https:\/\/www.ufrgs.br\/ppgest\/wp-json\/wp\/v2\/media?parent=2294"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}