Estimating Ross 308 Broiler Chicken Weight Through Integration of Random Forest Model and Metaheuristic Algorithms.

Area of Science:

• Agricultural Science
• Computational Intelligence
• Machine Learning

Background:

• Accurate broiler chicken weight (CW) estimation is crucial for optimizing poultry production.
• Traditional methods may lack the precision required for dynamic farm conditions.
• Hyperparameter tuning is essential for maximizing the performance of machine learning models.

Purpose of the Study:

• To develop and evaluate a novel hybrid method for accurate broiler chicken weight estimation.
• To compare the performance of optimized Random Forest (RF) models with benchmark algorithms.
• To assess the efficiency and accuracy of different optimization techniques for CW prediction.

Main Methods:

• Developed a hybrid approach integrating Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO) to tune Random Forest (RF) hyperparameters.
• Collected broiler chicken data including temperature, humidity, feed consumption, and CW from six farms in Türkiye (2014-2021).
• Compared the predictive performance of RF models optimized with PSO and ACO against standard RF and Linear Regression (LR).

Main Results:

• The RF-PSO and RF-ACO models demonstrated significant improvements in CW prediction accuracy compared to standard RF and LR.
• RF-PSO reduced Mean Absolute Error (MAE) by 5.081% to 60.707%, while RF-ACO achieved MAE reductions of 3.066% to 43.399%.
• Both RF-PSO and RF-ACO exhibited considerable computational efficiency, requiring low training times.

Conclusions:

• Hybrid methods combining RF with PSO or ACO are highly effective for accurate broiler chicken weight estimation.
• PSO and ACO are computationally efficient algorithms for optimizing machine learning models in precision agriculture.
• The developed hybrid approach offers a promising solution for enhancing poultry farm management and productivity.