Comparison of methods for estimation of lactation yield, from weekly, biweekly, monthly and bimonthly controls

Abstract

Three statistical methods for estimating the total milk yield of gir cows were tested using 1,085 lactations from weekly controls, 1,084 from biweekly, 1,036 from monthly and 630 from bimonthly controls. Results obtained with the methods above were compared with actual yield, by means of the error distribution analysis and by the regression line Y = A + BX, where Y is the estimated yield and X is the actual yield. Only estimates obtained by a logarithimic quadratic lactation curve using weekly and biweekly controls were consistent with actual yield, and their error were symmetrically distributed even with monthly control. Some of the methods tested can be corrected simply by adding or subtracting a constant and therefore can be used for weekly and biweekly controls. Yield estimates obtained with the use of bimonthly controls were not consistent with actual yield for any of the methods studied.