利用回归分析与最小二乘法对影响产奶量的因素进行校正的比较研究

A Comparision Study on the Adjusting Coefficients Estimated by Regression Analysis and the Least Square Method for Milk Yield

为了校正产犊月份和胎次对产奶量的影响，提高选种的准确性。以河北双娃奶牛场815头中国荷斯坦奶牛DHI测定结果为材料，利用回归分析和最小二乘法，制订了不同产犊月份和胎次对产奶量的校正系数。并且对两种研究方法的优劣进行了比较研究。两种方法对305天产奶量估计值与实际测定值之间均无显著差异，但利用最小二乘法得出的第1胎305天产奶量估计值与利用回归分析法得出的估计值之间差异显著（P=-0．04〈0．05），且1-4胎的最小二乘法估计值与实际测定值更为接近。最小二乘法在对影响产奶量因素校正上准确性明显优于回归分析，且应用起来更方便。

To improve the accuracy of breeding, the effects of calving monthes and lactations must be considered. This study is based on the milk yield records of 815 Chinese Holstein Dairy cattle from Hebei Shuangwa dairy farm. The adjusting coefficients of milk yield with calving monthes and lactations were investigated with the simple statistical polynomial regression analysis method and the least square method. Besides, a comparison study on the two different methods was carried out. Results shows that there was no significant difference between the milk yield estimated by the adjusting coefficients and the actual calculated 305 d milk yield value. But there was a significant difference between the two methods on milk yield estimated at parity I （P=-0.04〈0.05）, and milk yield estimated by the adjusting coefficients of the least square method were much closer to the actual values in all the parities from I to IV. The adjusting coefficients estimated by the least square method were much more accurate than that by regression analysis, and much more convenient in application for milk yield.