Files
Abstract
For the first study a multiple-trait model (MTM), a random regression model utilizing Legendre polynomials (RRML), and a random regression model construct with linear spline functions (RRMS) were applied for analysis of national beef cattle growth data. The impact of the additional information included in the RRML and RRMS were examined through correlations of random effect predictions. Results showed decreases in correlations between MTM and both RRM when additional information was incorporated into RRM analysis. The second study focused on the modeling of fixed effects within the frame work of a RRM. Models utilizing polynomials and two-dimensional splines were evaluated via cross 2validation based on the average squared error (ASE), R, and percent bias. Due to the nonlinearity of two-dimensional splines, weighted spline extrapolation had to be used outside the two-dimensional grid. Results showed comparable performance between polynomials and two-dimensional splines.