Comparison of Fitting Performance of Random Regression Models to Test Day Milk Yields in Holstein Friesians

In this study, random regression models with Ali-Schaeffer functions, Wilmink functions and orthogonal Legendre polynomials were compared for fitting performance to test day milk yields. Legendre polynomials with orders from two to six for additive genetic and permanent environmental effects were fitted under homogeneous error variance assumption throughout lactation. The analyzes were applied to 5918 first lactation test day milk yields of 612 Holstein Friesian cows calving from 1987 to 1993 in Dalaman, Tahirova, Sarimsakli and Turkgeldi State Farms. To compare the models, residual variances, -2LogL value, Akaike's information criterion, Bayesian information criterion and eigenvalues for additive genetic and permanent environmental random regression (co)variance matrix were used. Among 27 models, the L(6,2), L(6,5) and L(6,6) were chosen as better models.