Reassessing Wiggans’ method for estimating daily yields in 3× milking systems over four decades: what have we learned?

Authors: WU, XIAO-LIN - Council On Dairy Cattle Breeding; WIGGANS, GEORGE - Council On Dairy Cattle Breeding; CAPUTO, MALIA - Council On Dairy Cattle Breeding; Van Tassell, Curtis; Baldwin, Ransom; SIEVERT, STEVEN - Collaborator; MATTISON, JAY - Collaborator; COLE, JOHN - Council On Dairy Cattle Breeding; BURCHARD, JAVIER - Council On Dairy Cattle Breeding; DURR, JOAO - Council On Dairy Cattle Breeding

Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 3/20/2026
Publication Date: N/A
Citation: N/A

Interpretive Summary:

Technical Abstract: More than forty years have passed since Wiggans (1986) proposed a least-squares approach followed by an a posteriori scaling of the intercepts to derive yield correction factors from proportional yields in cows milked three times daily. In the present study, we re-evaluated these correction factors using 3× milking data collected between 2023 and 2025 from four Holstein dairy farms across three U.S. states. The objectives were twofold: to assess the validity of the existing yield factors and to examine the robustness of the associated modeling strategies. The accuracy of estimated daily yields was evaluated using out-of-sample validation, with approximately 50% of the records allocated to training and the remaining 50% to testing. Accuracy measures included the correlation between observed and estimated 24-h yields, mean absolute error (MAE), and mean squared error (MSE). We observed clear shifts in model parameters over the past four decades, indicating greater dynamic responsiveness of proportional yield to interval duration, while baseline section-specific effects have declined in relative importance. Nevertheless, improvements in the accuracy of estimated daily yields were modest. Substantial variability was observed among farms, suggesting that herd-level factors—such as management practices, milking system precision, and data consistency—contribute more to predictive performance than temporal changes in model parameters. Expanding the training dataset, particularly by including more representative herds and cow-day records, may enhance predictive stability and generalizability. Although imposing a 24-h normalization constraint improves model interpretability and ensures the coherence of proportional estimates, it does not necessarily result in meaningful gains in predictive accuracy. Overall, this study reaffirms the continued relevance of proportional-yield modeling and existing correction factors as practical benchmarks, while providing updated empirical guidance for estimating daily yields in three-times-daily milking systems.