ESTUDO DA CURVA DE CRESCIMENTO DE SUÍNOS COM O USO DE MODELOS NÃO LINEARES

Abstract

The objective of this study was to analyze the use of nonlinear models in pig growth data. The data were obtained from monthly body weight measurements taken from the first day of life until 150 days of age, from thirteen male and female pigs of the industrial crossbreed Landrace x Duroc with F1 (Landrace x Large White), provided by a partner pig farming operation. The weights were collected using a precision scale and analyzed through nonlinear regression, using the von Bertalanffy, Brody, Gompertz, Logistic, Richards, and Spline models. To determine which model best fit the data, the following evaluators were used: percentage of convergence, coefficient of determination, adjusted coefficient of determination, mean squared prediction error, AIC, and BIC. For statistical analysis, the Marquardt method was used through SAS® 9.2 software. The model with the best fit was the Richards model. Other models, such as Gompertz and von Bertalanffy, also showed good fit quality. The Brody model was excluded due to a lower percentage of convergence (23.07%).