Prediction of annualized lactation yield from partial lactations

Extension factors for annualized lactation yields [365 (total lactation yield)/-days between calvings] were computed by generating 747,904 partial lactations from complete lactation records of 105,379 Israeli-Holstein cows with one to three lactations. Factors included in the model were last test yield, days pregnant, days in milk, farm type, calving season, and days remaining in the lactation. Both linear and quadratic effects were assumed for the first three factors. Primiparous and multiparous cows were analyzed separately for milk, fat, and economically fat-corrected milk [.67 (milk) + 10 (fat)] yields. Adjusted coefficients of determination computed for this model were .04 higher for fat and .01 higher for milk and economically fat-corrected milk than for an alternate model in which remaining yield was predicted as a linear function of last test yield and remaining days in milk. Correlations between actual and predicted lactations for partial lactations of less than 4 mo in milk computed by the first model were .03 higher for fat but were nearly identical to those computed by the second model for milk. Although last test yields was the most important factor in predicting annualized yields, inclusion of other factors increased accuracy of annualized fat yield prediction.