Predicting dry matter intake in Pelibuey sheep using machine learning methods

Area of Science:

• Animal Science
• Machine Learning
• Agricultural Engineering

Background:

• Accurate prediction of Dry Matter Intake (DMI) is crucial for optimizing nutrition in growing livestock.
• Pelibuey sheep production is significant in tropical regions, necessitating efficient management strategies.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting DMI in growing male Pelibuey sheep.
• To compare the predictive performance of Multivariate Adaptive Regression Splines (MARS), Classification and Regression Tree (CART), and Support Vector Regression (SVR) algorithms.

Main Methods:

• Data from 130 Pelibuey sheep (average body weight 23 ± 6 kg) were collected under tropical conditions.
• Key variables included diet concentrate (CON), initial and final body weight (IBW, FBW), mean metabolic body weight (MBW), average daily gain (ADG), crude protein (CP), and neutral detergent fiber (NDF).
• MARS, CART, and SVR algorithms were employed to build predictive models for DMI.

Main Results:

• The MARS algorithm achieved a determination coefficient exceeding 0.90.
• All three machine learning methods were utilized to develop predictive algorithms for DMI.
• MARS proved to be a highly reliable model for predicting DMI in this sheep breed.

Conclusions:

• The MARS algorithm is a robust and reliable tool for predicting Dry Matter Intake in growing Pelibuey sheep.
• Machine learning approaches, particularly MARS, offer effective solutions for nutritional management in sheep farming.
• This study provides a validated method for enhancing precision feeding strategies in Pelibuey sheep production.