A comparative study of machine learning methods: A case study of weight and growth of livestock

The observation of daily weights and growth patterns of animals is important to detect the health and nutritional status of animals. The research investigated the usefulness of machine learning (ML) models and compared the performance of Multiple Linear Regression (MLR), Decision Tree (DT), Random Forest (RF), Support Vector Machines (SVM) and XGBoost (XGB) algorithms to predict the daily weight and growth curves of beef cattle. The min, mean, max and SD of daily weights in kg predicted using the best performing model (XGB) were 36.71, 251.74, 493.85 and 99.51 respectively. The results suggest that ML algorithms could be used as a tool to predict weight and growth of cattle using predictors such as age, sex, and breed. The model predictions and validation statistics showed that XGB is the best ML algorithm (R2 = 0.869) for predicting cattle weight using age, sex, and breed as predictors.