Análise dos resíduos em modelos de regressão não linear ajustados aos dados de Crotalária juncea por diferentes métodos de estimação

Abstract

The objective of this work was to verify the quality of the residues in the nonlinear regression model Gompertz and Logistic estimated by different methods of parameter estimation. Two uniformity were performed. After the emergence of the seeds of Sunn hemp, they were collected randomly four plants. The traits: mass of leaf fresh matter (MFF), mass of stem fresh matter (MFC), mass of root fresh matter (MFR), mass shoot fresh matter (MFPA), the total fresh matter mass (MFT), mass of dry matter leaf (MSF), mass of dry matter stem (MSC), mass of dry matter root (MSR), mass of dry matter shoot (MSPA), and total dry matter mass (MST) depending on the days after sowing. The residual assumptions of the characters studied were verified by test Shapiro-Wilk, test Breusch-Pagan and test DurbinWatson. After, the different estimation methods for the parameters of the models were analyzed: the least squares method and the maximum likelihood method. The methods of estimating the parameters of the nonlinear regression models did not determine.