Prediction of Live Weight in Romanov Lambs Using Body Measurements and Machine Learning Algorithms

This study aimed to compare the predictive performance of different machine learning algorithms for estimating live weight in Romanov lambs using body measurements. The dataset consisted of repeated live weight and morphometric measurements obtained from 50 Romanov lambs reared in the Niksar district of Tokat, Türkiye. Withers height, rump height, body length, chest depth, chest girth, and chest width between the shoulders were used as predictor variables. Random Forest, Support Vector Machine, XGBoost, LightGBM, and Gradient Boosting Machine algorithms were trained and evaluated using a 75% training and 25% testing data split, while 5-fold cross-validation was applied for hyperparameter optimization. Model performances were assessed using mean squared error (MSE), root mean squared error (RMSE), and the coefficient of determination (R²). Among the evaluated models, Gradient Boosting Machine achieved the best predictive performance, with an MSE of 0.991, an RMSE of 0.995, and an R² of 0.980. The findings indicate that machine learning algorithms, particularly tree-based ensemble methods, can be effectively used as practical decision-support tools for live weight estimation in Romanov lambs.