Skip to Main Content (Press Enter)

Logo UNISS
  • ×
  • Home
  • Degrees
  • Courses
  • Jobs
  • People
  • Outputs
  • Organizations
  • Third Mission
  • Expertise & Skills
  • Equipment

Logo UNISS

|

UNIFIND

uniss.it
  • ×
  • Home
  • Degrees
  • Courses
  • Jobs
  • People
  • Outputs
  • Organizations
  • Third Mission
  • Expertise & Skills
  • Equipment
  1. Outputs

Comparison of parametric, orthogonal, and spline functions to model individual lactation curves for milk yield in Canadian Holsteins

Academic Article
Publication Date:
2010
Short description:
Comparison of parametric, orthogonal, and spline functions to model individual lactation curves for milk yield in Canadian Holsteins / Dimauro, Corrado; Miglior, Filippo; Macciotta, Nicolò Pietro Paolo; Schaeffer, Lawrence R.. - 9:4(2010), pp. 460-464. [10.4081/ijas.2010.e87]
abstract:
Test day records for milk yield of 57,390 first lactation Canadian Holsteins were analyzed with a linear model that included the fixed effects of herd-test date and days in milk (DIM) interval nested within age and calving season.
Residuals from this model were analyzed as a new variable and fitted with a five parameter model, fourth-order Legendre polynomials, with linear, quadratic and cubic spline models with three knots. The fit of the models was
rather poor, with about 30%-40% of the curves showing an adjusted R-square lower than 0.20 across all models. Results underline a great difficulty in modelling individual deviations around the mean curve for milk yield. However, the Ali and Schaeffer (5 parameter) model and
the fourth-order Legendre polynomials were able to detect two basic shapes of individual deviations among the mean curve. Quadratic and, especially, cubic spline functions had better fitting performances but a poor predictive
ability due to their great flexibility that results
in an abrupt change of the estimated curve when data are missing. Parametric and orthogonal polynomials seem to be robust and affordable under this standpoint.
Iris type:
1.1 Articolo in rivista
Keywords:
Test day model; residuals; legendre; polynomials; splines
List of contributors:
Dimauro, Corrado; Miglior, Filippo; Macciotta, Nicolò Pietro Paolo; Schaeffer, Lawrence R.
Authors of the University:
DIMAURO Corrado
MACCIOTTA Nicolo' Pietro Paolo
Handle:
https://iris.uniss.it/handle/11388/264381
Full Text:
https://iris.uniss.it//retrieve/handle/11388/264381/195616/Macciotta_N_Comparison_of_parametric_orthogonal.pdf
  • Use of cookies

Powered by VIVO | Designed by Cineca | 26.11.0.1