Precisão das estimativas em modelos de regressão da energia metabolizável do milho para suínos em crescimento

Abstract

Trabalho selecionado durante a VI Semana de Ciências Agrárias (VI SECIAGRA), realizada de 01 a 03/10/2012

 

The proposed objective of this study was to evaluate the accuracy of estimates of the parameters of equations to predict the metabolizable energy (ME) of corn to pigs  in function of chemical composition, obtained by the least squares method (LSM) and the simulation method p-bootstrap with different numbers of re-samples. We developed a database containing information about metabolism trials, which used the method of total collection of feces and urine

in order to determine the chemical composition and ME value of corn for pigs with body weight between 10 and 65 kg. It was adjusted a multiple regression model of ME of corn in function of crude protein, ether extract, crude fiber, ash and digestible energy, using the LSM. Then, were adjusted two regression models by bootstrap simulation, in which if used 1000 and10000 re-sampling. The regression model estimated by LSM presented the crude protein and the digestible energy as significant regressive, being watched smaller amplitude us confidence intervals (CI) compared the estimates using the p-bootstrap method with different sizes of re-samples. The amplitudes of CI of parameters were: intercept (1104.31, 1320.59 and 1348.63), crude protein (40.1, 44.34 and 43.79) and digestible energy (0.264, 0.327 and 0.323) for the LSM, bootstrap₁₀₀₀ and bootstrap₁₀₀₀₀, respectively. In the conditions observed, the parameter estimates of the regression model the ME of corn for pigs in function of crude protein and digestible energy are more accurate when used the LSM.