Procedures to estimate daily yield from either morning (AM) or evening (PM) test weighings and to compare different systems of testing for accuracy of predicting 305-day milk yield using various combinations of AM and PM production records were determined. Data of 20,328 test-day records of 1,516 cows milked twice-a-day, from 96 Holstein Dairy Cattle Association herds in Minas Gerais State and recorded from 2000 to 2003. Seven procedures were used to estimate daily yields. Comparisons were made on the basis of correlations between true and estimated daily yields, average difference between the estimated and actual values and also the variances of differences. Morning weighings predicted daily yield more accurately than evening weighings. The procedure which included a regression for each milking interval class and a regression on days in milk were selected for estimation of daily yields. Four systems for predicting 305-day milk yield were compared. Alternating AM-PM testing systems on monthly frequency showed better results than testing always in the AM or always in the PM. Correlations between observed and estimated 305-day lactation yields were 0.99 and 0.97, respectively, for the first and later systems. Alternating AM-PM testing starting with an AM test was the most accurate system of estimation.
dairy cattle; milking records; prediction; lactation stage; milking interval
