EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An analytical framework for prediction accuracy of estimated daily yields in 3-times-daily milking herds

Xiao-Lin Wu, George Wiggans, Ashley Ling, Malia J Caputo, Curtis P Van Tassell, Ransom L Baldwin, John B Cole, Javier Burchard, João Dürr

JDS Communications · 2026-11 · 7 (6) · 853-858

DOI: 10.3168/jdsc.2026-1072 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JDS Communications, 2026-11. Verfasst von Xiao-Lin Wu und 8 weiteren Autor:innen. Beteiligte Einrichtung: Council on Dairy Cattle Breeding, Bowie, MD 20716. Die Arbeit nennt 6 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate estimation of daily milk, fat, and protein yields from 3-times-daily (3×) milking data relies on multiplicative correction factors derived from proportional-yield modeling. Although this method has been used for more than 40 yr in the United States, the determinants of prediction accuracy and the practical limits of further improvement remain poorly understood. The objective of this study was to develop an analytical framework to quantify and decompose the sources of prediction error in estimated daily yields and to clarify how accuracy can be improved. Building on a reparameterized proportional-yield model centered on an 8-h milking interval, we derived an approximation for the expected correlation between observed and estimated daily yields. The total prediction error variance was decomposed into intrinsic stochastic variability and parameter estimation-induced uncertainty. The method was evaluated through rotational cross-validation using 3× milking data collected from 4 US Holstein herds during 2023 to 2025. Analytical accuracies closely matched empirical correlations between observed and estimated yields for milk, fat, and protein. Across herds and traits, intrinsic stochastic variability accounted for nearly all prediction error variance (e.g., >99%), whereas estimation-induced variance was negligible. These results demonstrate that attainable accuracy is fundamentally limited by biological and sampling variability in yields rather than by estimation uncertainty in regression parameters once sufficient data were available for training. Consequently, meaningful improvements in prediction accuracy will require refinement of the statistical model or the sampling procedure, for example, using a multiple-day average of records to effectively reduce intrinsic stochastic variance. The proposed analytical framework provides a practical and effective tool for diagnosing the underlying sources of prediction error and informing strategies to improve daily yield estimation in 3× milking systems.

Autor:innen und Institutionen

  • Council on Dairy Cattle Breeding, Bowie, MD 20716.
  • Department of Animal and Dairy Sciences, University of Wisconsin, Madison, WI 53706.
  • USDA Animal Genomics and Improvement Laboratory, Beltsville, MD 20705.
  • Department of Animal Science, North Carolina State University, Raleigh, NC 27695.
  • Department of Animal Sciences, Donald Henry Barron Reproductive and Perinatal Biology Research Program, and the Genetics Institute, University of Florida, Gainesville, FL 32611.
  • Department of Animal Biosciences, University of Guelph, ON, Canada N1G 2W1.

Zitieren

APA

Wu, X. L., Wiggans, G., Ling, A., Caputo, M. J., Van Tassell, C. P., Baldwin, R. L., Cole, J. B., Burchard, J., & Dürr, J. (2026). An analytical framework for prediction accuracy of estimated daily yields in 3-times-daily milking herds. JDS Communications, 7(6), 853-858. https://doi.org/10.3168/jdsc.2026-1072

Vancouver

Wu XL, Wiggans G, Ling A, Caputo MJ, Van Tassell CP, Baldwin RL, et al. An analytical framework for prediction accuracy of estimated daily yields in 3-times-daily milking herds. JDS Communications. 2026;7(6):853-858. doi:10.3168/jdsc.2026-1072

BibTeX
@article{Wu2026,
  author = {Wu, Xiao-Lin and Wiggans, George and Ling, Ashley and Caputo, Malia J and Van Tassell, Curtis P and Baldwin, Ransom L and Cole, John B and Burchard, Javier and Dürr, João},
  title = {An analytical framework for prediction accuracy of estimated daily yields in 3-times-daily milking herds},
  journal = {JDS Communications},
  year = {2026},
  volume = {7},
  number = {6},
  pages = {853--858},
  doi = {10.3168/jdsc.2026-1072}
}
Kostenlos starten →