Evaluation of Various Approaches in Prediction of Daily and Lactation Yields of Milk and Fat Using Statistical Models in Iranian Primiparous Holstein Dairy Cows

In this research, 272977 test day records collected from 659 herds during years 2001 to 2011 by the Iranian animal breeding center were used. In the first section the ability of different models to predict daily milk yield from alternative milk recording was tested. The result showed that a complex model including noon milking time plus the effect of lactation curve of Ali and Schaeffer function is the best equation for prediction of daily milk yield. The highest correlation between true and estimated daily milk yield (0.892) and the lowest bias (2.391) were obtained using this method. Of the four models, the Ali and Schaeffer and the Wood models resulted in the best goodness of fit and gave a good description of the lactation curve (milk and fat yield) for dairy herds when test-day yield is used. Lastly, the most appropriate models for prediction of 305 d milk and fat yields were Ali and Schaeffer and Wood respectively. These models were able to predict milk and fat yields with the lowest residual mean square errors. Thus, the performance of models based on lactation curve functions were better than the test-interval method and the centering date method for prediction of 305-d milk and fat yield in Iranian primiparous Holstein cows.