Evaluation of models for predicting the preweaning body weight in Holstein calves

Abstract

This study compared six non-linear equations [Exponential growth (4 parameters), Exponential growth (Stirling), Polynomial (Cubic), Quadratic, Brody, and Sinusoidal] for prediction of pre-weaning body weights at different ages in Holstein calves. Thirty-two calves (16 males and 16 females) were randomly divided into two treatment groups and fed with starter diets containing either corn or barley as the grain source. Starter feeding began on the third day of life, and high quality alfalfa hay and fresh cow milk were fed according to the farm schedule. The calves were weighed at birth and weekly thereafter until weaning. In this manner, ten weight records, including the birth and weaning weights, constituted the data set. The results of experiment revealed the fact that all functions mentioned earlier showed good fitness to predict weight gain in relation to age in all groups of calves. However, based on the goodness of the fit of various criteria and the statistical performance, the polynomial (cubic) function was considerably superior to other functions for predicting the calf live weight. The flexible growth functions (more parameters) very often give a closer fit to data points and a smaller residual sum of square (RSS) value than the simpler functions such as the Brody functions.