Smarter ways to predict rabbit body weight across multiple breeds

Area of Science:

• Animal Science
• Quantitative Genetics
• Machine Learning in Agriculture

Background:

• Accurate live weight estimation is crucial for effective animal breeding programs.
• Morphometric measurements offer a non-invasive method for assessing animal growth and development.
• Traditional linear models and modern machine learning techniques are increasingly applied to improve prediction accuracy.

Purpose of the Study:

• To compare the predictive precision of linear regression and Random Forest (RF) models for live weight estimation in rabbits.
• To evaluate the effectiveness of morphometric measurements in improving the accuracy of body weight prediction.
• To identify key morphometric traits that significantly influence rabbit body weight.

Main Methods:

• A study involving 228 rabbits across six breeds (Satin, Rex, New Zealand White, Hyla, Hycole, Reza).
• Collection of morphometric data including body weight, head dimensions, chest circumference, body length, and hip width.
• Application of linear regression and Random Forest (RF) algorithms using R software and relevant packages (caret, randomForest).

Main Results:

• Linear regression achieved a higher R-squared value (0.82) and lower Root Mean Squared Error (RMSE: 300.16) compared to RF (R-squared: 0.8, RMSE: 326.37).
• Chest circumference and body length were identified as the most significant predictors of body weight in the linear regression model.
• Head length was also highlighted as an important predictor by the Random Forest model's IncNodePurity metric.

Conclusions:

• Linear regression models demonstrate superior performance over Random Forest for predicting rabbit body weight using morphometric data.
• Morphometric measurements, particularly chest circumference and body length, are valuable predictors for live weight estimation in rabbit breeding.
• The findings support the use of optimized linear models for enhancing precision in animal breeding programs.