Predicting individual dry matter intake in lactating Girolando cows using an integrative approach and machine learning: model performance metrics.

SILVA, C. S. da; SGUIZATTO, A. L. L.; MACHADO, A. F.; GRAÇAS, A. V. das; LOPES, F. C. F.; SILVA, A. S.; CAMPOS, M. M.; MORENZ, M. J. F.

Predicting individual dry matter intake in lactating Girolando cows using an integrative approach and machine learning: model performance metrics.

Autores

Descripción

The dry matter intake (DMI) of dairy cows has been estimated through equations developed to accommodate general features of a given group of cows or conditions, mostly purebred cows in temperate climates. Thus, the ability of such equations to accurately predict individual DMI is limited, especially for crossbred cows like the Holstein × Gyr (Girolando) raised in Brazil. Machine-learning (ML) algorithms such as Random Forests (RF) have off ered the opportunity to integrate various features routinely collected within a herd to tailor predictions at the cow level. Random Forests have been applied in diff erent contexts of precision dairy science, including predictions of daily eating time (Foldager et al., 2020). This study aimed to develop a cow-level DMI prediction model for lactating Girolando cows using RF and an integrative approach, which incorporated data from behavior-monitoring collars, phenotypic traits, weather, and diet characteristics to customize predictions.
SIMLEITE.