Learning-based estimation of cattle weight gain and its influencing factors
🤖 AI Summary
Traditional manual weight measurement for beef cattle is labor-intensive, time-consuming, and highly stressful for animals; meanwhile, existing machine learning approaches suffer from poor generalizability and limited interpretability. To address these challenges, this paper proposes an interpretable machine learning framework explicitly incorporating livestock physiological constraints for accurate prediction of daily weight gain and multi-source factor attribution analysis. Methodologically, the framework integrates XGBoost, SHAP-based interpretability techniques, and multivariate time-series feature engineering; crucially, it is the first to embed domain-specific physiological knowledge directly into model architecture, unifying predictive accuracy with causal inference. Evaluated on real-world ranch data, the framework achieves a mean absolute error (MAE) of 0.18 kg/day and identifies key influencing factors with over 92% accuracy—substantially enhancing the scientific rigor and precision of feeding decision-making.
Application Category
Problem
Research questions and friction points this paper is trying to address.
Automate cattle weight gain estimation
Use machine learning for remote monitoring
Address data challenges in weight prediction
Innovation
Methods, ideas, or system contributions that make the work stand out.
Machine learning for cattle weight estimation
Remote monitoring system for livestock
Advanced ML techniques for predictive analysis
🔎 Similar Papers
No similar papers found.
