Análise de correspondência simples com novos escores e o uso da análise de correspondência múltipla em dados composicionais de granulometria de grãos de café

• Ano de defesa: 2016
• Autor(a) principal: Costa, André Luís Alves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Programa de Pós-Graduação em Estatística e Experimentação Agropecuária; UFLA; brasil; Departamento de Ciências Exatas
• Palavras-chave em Português: Binomial correlacionada; Hipótese de independência; Transformações logarítmicas; Dados composicionais; Related binomial; Independence hypothesis; Logarithmic transformations; Compositional data; Probabilidade e Estatística Aplicadas; Agronomia
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/11811

Three topics are presented in the composition of this work. The first one contains the theoretical basis of this study on the methods of correspondence analysis used for its development. The second topic contains a scientific article where a new approach on residuals incorporation is proposed to calculate the coordinates of the simple correspondence analysis by contingency tables in which categories have different levels of correlation, using Monte Carlo simulation in generation of frequencies from the correlated binomial distribution BC (n, π, ρ). The first article led to the conclusion that in all scenarios this approach is promising in the sense that the subjects were better discriminated when compared to the conventional approach. The second scientific article, which discusses the application of multiple correlation analysis of compositional data for a comparative study of logarithmic transformation effects performed on the original data to a study on the granulometry of the coffee beans, is presented in the third and last point. The use of logarithmic transformation was found suitable for compositional data analysis using multiple correspondence analyses. Among the transformations used in this study, the isometric logarithmic was the one able to discriminate most coffee samples in relation to the categories of the components.