应用非线性回归活体估测猪胴体瘦肉率的研究

以四川省养猪研究所1982—1984年166头荣昌猪屠宰试验资料为研究对象,探讨进行多元非线性回归的一般方法和步骤,建立荣昌猪瘦肉率估测的各种非线性回归方程。结果表明:以二次函数的效果为佳,所有二次回归方程均优于相应的线性回归方程;用二次回归能提高估测的准确性,尤其在线性回归效果较差的情况下。在线性回归时,单用体型性状和肥育性状估测瘦肉率(量)的准确性较低,将其加入到用膘厚估测瘦肉率的线性回归方程中,对提高估测的准确性意义不大;而在二次回归时,却能增加其准确性。

Study of Estimating Percentage of Carcass Lean form Live Pigs by Non-linear Regression Analysis

Carcass data for a total 166 Rongchong Pigs from the year 1982 to 1984 of Sichuan Pig Science Institute were used to discuss the general method and procedure of non-linear multiple regression analysis. The percentage or weight of lean in carcass was estimated by different non-linear regression eguations. The all quadratic regression equations were advantage over the linear ones. The quadratic regression can improve the precision of estimation for carcass lean percentage or weight, especially when the linear regession had poor precision, It was lower precision that the body measurements and live weight were used to estimate carcass lean percentage or weight in linear regression analysing. There was little significence when they were added to carcass backfat thickness, but Quadratic regression can increase the precision.