You are in the accessibility menu

Please use this identifier to cite or link to this item: http://acervodigital.unesp.br/handle/10400.2/4987
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorOliveira, Amílcar-
dc.contributor.authorMonteiro, Adriano Arcanjo-
dc.date.accessioned2016-03-10T17:06:49Z-
dc.date.accessioned2017-12-14T17:40:42Z-
dc.date.available2017-12-14T17:40:42Z-
dc.date.issued2015-
dc.identifier.citationMonteiro, Adriano Arcanjo - Análise conjunta de regressões lineares [Em linha] : revisão e desenvolvimento computacionais em R. Lisboa : [s.n.], 2015. 172 p.pt_PT
dc.identifier.urihttp://hdl.handle.net/10400.2/4987-
dc.identifier.urihttp://acervodigital.unesp.br/handle/10400.2/4987-
dc.descriptionDissertação de Mestrado em Matemática, Estatística e Computação apresentada à Universidade Abertapt_PT
dc.description.abstractO R é em simultâneo uma linguagem de programação e um ambiente para computação estatística e gráficos. Estaremos interessados em compreender a funcionalidade dos packages do R, criando um novo, capaz de executar automaticamente a Análise Conjunta de Regressões, ACR. A ACR é uma técnica muito utilizada para interpretar redes de ensaios no domínio da avaliação de cultivares. Nesta técnica encontramos um conjunto de procedimentos que nos permite na sua utilização fazer a comparação e seleção de cultivares. Numa primeira fase recorremos ao algoritmo zigzag no ajustamento de regressões lineares, uma por cultivar, onde nas regressões a variável controlada é o índice ambiental que mede a produtividade dos vários ambientes. Os valores dos índices ambientais e dos coeficientes das regressões serão ajustados simultaneamente pelo referido algoritmo, que segundo Oliveira, A. (2007) se baseia num processo iterativo onde as minimizações alternam entre os coeficientes de regressão e os índices ambientais. Para avaliarmos a qualidade do ajustamento obtida pelo algoritmo calcularemos o coeficiente de determinação R2 através dum modelo linear para os resíduos. Contrariamente ao que é usual, neste caso estamos interessados em obter valores baixos de R2, pois um valor próximo a zero indica um bom ajuste, enquanto valores maiores indicam que o algoritmo zigzag não foi capaz de extrair toda a informação relevante dos dados. Através do Método do Contorno Superior exemplificaremos, quais os cultivares integram esse contorno e posteriormente, através do teste t e dos métodos de comparação múltipla de Scheffé, Bonferroni e Tukey, averiguaremos quais os cultivares, ao nível de significância de 5%, são significativamente dominantes ou dominados, nos determinados intervalos de dominância. Neste trabalho, após apresentação dos aspetos teóricos, no Capitulo 5 desenvolveremos um conjunto de funções e rotinas, criando um “package” em linguagem R que nos permitirá, analisar no Capitulo 6, um conjunto de dados simulados para verificação da aplicabilidade do mesmo. Através da utilização de dados simulados referentes ao ensaio de cultivares, mostraremos como utilizar a ACR em planos de melhoramento com auxílio ao package „jra‟, desenvolvido para cumprir um dos objetivos desta dissertação.pt_PT
dc.description.abstractR is simultaneously a programming language and an environment for statistical computation and graphics. We will be interested in understanding the functionality of the packages of R, creating a new one, capable to execute the Joint Regression Analysis, ACR, automatically. The JRA is a technique very used to interpret networks in the domain of the evaluation of cultivars. In this technique we find a group of procedures that allow us to do the comparison and selection of cultivars. In a first phase we resorted upon the zigzag algorithm in the adjustment of linear regressions, one for cultivar, where in the regressions the controlled variable is the environmental index that measures the productivity of the several environments. The values of the environmental indexes and of the coefficients of the regressions, will be adjusted simultaneously by the referred algorithm, that according to Oliveira, A. (2007) is based on a iterative process where the minimizations alternate between the regression coefficients and the environmental indexes. In order to evaluate the quality of the adjustment obtained by the algorithm we will calculate the determination coefficient R2 using a linear model for the residues. Contrarily to what is usual in this case we are interested in obtaining low values of R2, because a close value to zero indicates a good adjustment, while larger value indicate that the zigzag algorithm was not capable to extract all the relevant information from the data. Through the Upper Contour Method, we will exemplify, which cultivars integrate that Contour and later, observed through t test and the methods of multiple comparisons of Scheffé, Bonferroni and Tukey, which cultivars, at level of significance of 5%, are significantly dominant or dominated, in the determined intervals of dominance. In this work, after presenting the theoretical aspects, in Chapter 5 we will develop a set of functions and routines, creating a "package" in language R that will allow us, to analyze in Chapter 6, a set of simulated data for verification of the applicability of the same.pt_PT
dc.language.isoporpt_PT
dc.rightsembargoedAccesspt_PT
dc.subjectMatemáticapt_PT
dc.subjectEstatísticapt_PT
dc.subjectMatemática computacionalpt_PT
dc.subjectLinguagens de programaçãopt_PT
dc.subjectR languagept_PT
dc.subjectJoint Regression Analysispt_PT
dc.subjectUpper contour methodpt_PT
dc.subjectT testpt_PT
dc.subjectMethods of multiple comparison of Scheffépt_PT
dc.subjectBonferroni and Tukeypt_PT
dc.titleAnálise conjunta de regressões lineares : revisão e desenvolvimento computacionais em Rpt_PT
dc.typeoutropt_PT
dc.identifier.tid201136457-
dc.date.embargo2019-03-10-
Appears in Collections:Dissertações de Mestrado - Universidade Aberta de Portugal

There are no files associated with this item.
 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.