应用多元多项式回归估测猪胴体瘦肉率的研究

Studies on applying multivariate polynomial regression to estimate the pig carcass lean meat percentage

应用多元多项式回归分析研究了102头“大长北”三品种杂种猪的六个胴体性状与瘦肉率之间的关系,结果表明,①与猪胴体瘦肉率关系最密切的六项因素依次为:膘厚二次幂、膘厚一次幂、腿肉率一次幂、膘厚三次幂、后腿比一次幂和腿肉率二次幂;②瘦肉率与膘厚和腿肉率呈曲线关系;③各性状间的互作对瘦肉率的影响不可忽视;④各性状的三次幂多数对瘦肉率的作用较小,为了便于生产上的应用,采用逐步回归建立了一组估测胴体瘦肉率的简易多项式回归方程,它们的复桐关系数多在0.83以上,p<0.005,比用线性回归方程估计的决定系数(R^2)提高12％左右。

Multivariate polynomial regression was used to investigate the relation between carcass lean meat percentage and 6 carcass traits of 102 'Da-Chang-Bei' three-variety cross pigs.Results revealed;(1)the chief 6 factors affecting carcass lean meat percentage were the second and first power of backfat thickness,the first power of ham lean meat percentage,the third power of backfat thickness,the first power of ham-carcass ratio and the second power of ham lean meat percentage,(2)curve relation existed between carcass lean meat percentage and backfat thickness and ham lean meat percentage,(3)the effects of interactions between each carcass trait on carcass lean meat percentage should not be neglected,(4)the effects of the third power of each carcass trait on carcass lean meat percentage were weak,By using stepwise regression,a series of multivariate polynomial regression equations which were easy to be used to estimate carcass lean meat percentage were obtained.Most of their coefficients of multiple correlation were above 0.83(p<0.005).Coefficient of determination(R2)by this method increased by about 12% comparing to that by linear regression method.