Prediction of growth and feed efficiency in mink using machine learning algorithms.

Area of Science:

• Animal Science
• Machine Learning Applications
• Agricultural Economics

Background:

• Feed efficiency is critical for mink industry profitability, but traditional measurement methods are costly and labor-intensive.
• Accurate feed efficiency metrics are essential for improving mink farming competitiveness.
• Current methods for measuring feed intake and body weight in mink are impractical for widespread adoption.

Purpose of the Study:

• To evaluate machine learning algorithms for predicting mink average daily gain (ADG), feed conversion ratio (FCR), and residual feed intake (RFI).
• To identify key features (sex, color, age, body weight, length) for accurate prediction of feed efficiency metrics.
• To determine the most significant features influencing ADG, FCR, and RFI in mink.

Main Methods:

• Utilized machine learning algorithms to predict ADG, FCR, and RFI over a 15-week period.
• Collected data on sex, color type, age, body weight (BW), and body length at 3-week intervals.
• Compared prediction accuracy using various combinations of features, with initial measurements (August 1st) proving most effective.

Main Results:

• The Extreme Gradient Boosting (XGB) algorithm demonstrated high accuracy in predicting ADG (R²=0.71), FCR (R²=0.74), and RFI (R²=0.76).
• Including all initial features (sex, color, age, BW, length on August 1st) yielded the most accurate predictions.
• Sex was identified as the most significant predictor for ADG, FCR, and RFI.

Conclusions:

• Machine learning, particularly XGB, can accurately predict mink feed efficiency metrics without direct measurement of daily feed intake.
• Initial measurements of sex, color, age, BW, and length are sufficient for reliable feed efficiency prediction.
• The findings offer a cost-effective solution for evaluating feed efficiency, enhancing mink industry competitiveness.