In animal breeding, many traits admit repeated measurements or test-day records over time. In order to model the expectation and covariance of these characters as a function of time, the interest in using test-day models (TDM) in the context of the Gaussian mixed linear model has increased in recent years. Even though the advantages of TDM are well known, a problem that frequently arises is the choice of suitable number terms in the linear function. Jensen (2001) discussed various strategies about model choice and, within the Bayesian framework, he suggested Bayesian model averaging to consider the uncertainty around the model for the prediction of breeding values. In this work we implement a simple and flexible Bayesian approach proposed by Kuo and Mallick (1998) to subset a pre-specified set of covariates that best describe the trait of interest in a random coefficient regression model (RRM). The posterior probability of each regressor entering the model is computed using the Gibbs sampling algorithm. The method is illustrated with a simple example where the variable selection strategy is limited to the fixed effects.

J. P Steibel, F. Grignola

Proceedings of the World Congress on Genetics Applied to Livestock Production, Volume 2002. Session 16, , 16.1, 2002
Download Full PDF BibTEX Citation Endnote Citation Search the Proceedings

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.