Application of machine learning to improve the interpretability of the relationships between silage quality and dairy production

While the impact of forage quality on dairy-milk production is clear, much work to date on such relationships as digestibility, fermentation characteristics and nutritional constituents with milk yield and composition have not easily been transferred to producers, given the complexity of analytical tests as well as their interpretation. With the growing interest in the field of precision dairy farming, the objective of this research was to provide better interpretability of the relationship between silage quality and milk production by performing an in-depth machine learning data analysis of forage variables such as digestibility, fermentation characteristics, and nutritional constituents. Dairy production and forage data – characterized as either grass and legume silage or corn silage – were provided by Lactanet (Québec Dairy Herd Improvement). Since production data do not typically contain complete forage information, missing impact variables (like digestibility and fermentation characteristics) were predicted using supervised machine-learning algorithms; multi-input and multi-output (MIMO) regression was performed using a Meta-estimator with the Extra Tree algorithm as a base regressor for grass and legume silage, and a regressor chain based AdaBoost algorithm with an Extra Tree algorithm as the base regressor was used for corn silage. Factorial analysis was used to extract key silage quality characteristics (e.g., neutral detergent fiber digestibility, heat damage, rumen protein degradability, legume proportion, homolactic fermentation, length of initial fermentation, soil contamination, bad fermentation pattern), and a linear mixed effects model was used to estimate the effects of silage quality on average herd milk production and composition (fat, protein, milk-urea nitrogen and somatic cell score).Increases in fermentation length, as well as the proportion of grass and legume mixture and corn silage or concentrate from feed, were associated with higher milk production, while the proportion of corn silage from feed showed an increasing trend in somatic cell count and milk urea nitrogen. Results from the research support the importance of forage quality on milk production and provide a method for completing missing forage values relative to time of feeding. Through this demonstration of the impact of forage quality, it is hoped to encourage better attention to this important input resource. The possibility of developing a decision-support tool for silage quality evaluation seems reasonable, thereby helping to provide consultants and producers with a practical guide to improved dairy profitability from forage

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.