Modified Regression for Estimating Total Lactation from Part-Lactation Yields

Abstract

A modified regression method for extending part records eliminates population or herd means of the partial lactations. Errors of prediction by the method were compared to those of extended records by ratio factor in two samples of data, one from Dairy Herd Improvement Association and one from the Beltsville herd. Errors of estimates by ratio factor were large, especially early in lactation. For projecting total yield from the first test, the average absolute difference from actual 305-day yield was 500 kg smaller for the modified regression procedure as compared to the ratio estimate.