Choosing the Best Machine Learning Model for Weight Estimation at Different Growth Stages in Bali Cattle

Abstract

Background: Accurate weight estimation is essential for effective livestock management. Bali cattle, a vital component of Indonesia’s agricultural economy. Traditional weighing methods pose practical challenges, thus requiring adoption of machine learning (ML) techniques. Methods: This study compares five ML models-linear regression (LR), random forest regression (RFR), support vector regression (SVR), neural network regression (NNR) and polynomial regression (PR)-to determine the most accurate approach for predicting Bali cattle weight across different age groups. The research was conducted at a livestock breeding center in Bali, where morphological features such as body height, body length and heart girth were collected and used as predictors. Results indicate that polynomial regression consistently outperformed other models for middle-aged cattle (366-728 days), while random forest regression performed best for younger (240-365 days) and older (729+days) cattle. SVR and NNR struggled to generalize due to the characteristics of the dataset. Result: The study highlights the importance of age-specific modeling for precise weight prediction, offering valuable insights into precision livestock management. Future research should explore deep learning or hybrid approaches to improve predictive accuracy for mature cattle.