Modelos não lineares: aplicação da análise bayesiana aos dados originais e isotonizados do acúmulo do nitrogênio no feijoeiro cv. Jalo

• Ano de defesa: 2019
• Autor(a) principal: Silva, Laís Mesquita
• Tipo de documento: Tese
• 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 Estatística
• Palavras-chave em Português: Regressão não linear; Modelo logístico; Curva de crescimento; Nonlinear regression; Logistic model; Growth curve; Estatística
• Link de acesso: http://repositorio.ufla.br/jspui/handle/1/34302

The mass accumulation data has an ordering characteristic, so the application of the isotonic transformation smoothes the data so that the efficiency of the adjustment is increased. Based on the nonlinear regression models, which allow to synthesize information in a few parameters, facilitating and aiding in the explanation of the processes involved in plant growth, logistic and Gompertz nonlinear models were used in the original and isotonized data to describe the accumulation of nitrogen of common bean cv. Jalo; using the Akaike information criterion (AIC) and adjusted coefficient of determination as measures of adjustment qualities. Estimates of maximum asymptotic weight, growth rate and inflection point were obtained, which varied according to the planting system and sowing density. The methodology on the Bayesian nonlinear modeling of growth in nitrogen accumulation allowed us to compare, through the logistic model, the types of management for both the original data and the isotonized data. Therefore, the use of isotonic regression was efficient for the reduction of experimental precision. The logistic nonlinear model presents better adjustment quality for the description of nitrogen accumulation in which its accumulation increased during the crop cycle. No-tillage presented higher nitrogen accumulation than conventional tillage. The Bayesian methodology was efficient when using data isotonia, as there was a reduction of the standard deviation of the estimates for most of the parameters, implying a decrease in the amplitude of the confidence intervals.