Optimal binning of continuous traits: Is the tradition tenable?
JDS Communications · 2026-11 · 7 (6) · 847-852
DOI: 10.3168/jdsc.2026-1064 ↗Kurz gefasst
Fachartikel in der Zeitschrift JDS Communications, 2026-11. Verfasst von Xiao-Lin Wu und 4 weiteren Autor:innen. Beteiligte Einrichtung: Council on Dairy Cattle Breeding, Bowie, MD 20716. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 11 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Discretizing continuous traits is a standard practice in animal breeding. In dairy cattle, for example, conformation or linear-type traits derived from continuous measurements are commonly converted into categorical scores. Although binning procedures are often ad hoc, it remains unclear whether traditional binning conventions are statistically well justified or optimal. This study revisited and formalized the theoretical foundations of discretizing continuous variables and derived principled criteria for optimal bounded equal-interval binning, a widely used approach in the dairy industry. Within a standard normal framework, optimal bin widths (δ) and the number of bins (k) are derived either heuristically or through formally maximizing information retention. The latter is measured by the squared Pearson correlation (ρ 2) between the original value and the conditional group means, or approximated by the correlation between the original value and the discretized scores. Closed-form expressions are obtained, allowing optimal bin widths and boundaries to be determined analytically. Numerical optimization further reveals systematic and predictable relationships among bin number, coverage span, and retained information. Under the Pearson-based optimization, approximately 5, 7, 9, and 20 bins retain 90%, 95%, 97%, and 99% of the original variance, respectively. Overall, 7 to 9 optimally spaced bins are sufficient to preserve ∼95% to 97% of the original information, whereas overly coarse binning introduces substantial and quantifiable information loss. Although traditional binning practices may retain adequate information at higher resolution, careful and principled calibration becomes essential when using coarse categorical scales. These findings provide a formal and quantitative benchmark for evaluating and optimizing traditional categorical scoring 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.
- Brown Swiss Association, Beloit, WI 53511.
- 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 Science, North Carolina State University, Raleigh, NC 27695.
- Department of Animal Biosciences, University of Guelph, Guelph, ON, Canada N1G 2W1.
Zitieren
APA
Wu, X. L., Rosa, G. J. M., Horn, A., Nicolazzi, E., & Cole, J. B. (2026). Optimal binning of continuous traits: Is the tradition tenable?. JDS Communications, 7(6), 847-852. https://doi.org/10.3168/jdsc.2026-1064
Vancouver
Wu XL, Rosa GJM, Horn A, Nicolazzi E, Cole JB. Optimal binning of continuous traits: Is the tradition tenable?. JDS Communications. 2026;7(6):847-852. doi:10.3168/jdsc.2026-1064
BibTeX
@article{Wu2026,
author = {Wu, Xiao-Lin and Rosa, Guilhmere J M and Horn, Allicia and Nicolazzi, Ezequiel and Cole, John B},
title = {Optimal binning of continuous traits: Is the tradition tenable?},
journal = {JDS Communications},
year = {2026},
volume = {7},
number = {6},
pages = {847--852},
doi = {10.3168/jdsc.2026-1064}
}
EUVIMED