Maschinelle Lernverfahren zur Prognose von Tierwohlrisiken in der Schweinehaltung

Abstract

Animal welfare is a quality indicator of modern pig farming and increasingly important to society. Animal welfare risks have multiple factors and should be recognized and mitigated early on to prevent economic risks. In this work, we use machine learning models to predict animal welfare risks. Our dataset comprises data for over 57,000 pigs with indications of 10 animal welfare risks and 14 suckling phase features. We contribute a prediction model for suckling phase deaths with an accuracy of 80.4% – providing a sizeable improvement over a majority vote‘s accuracy of only 53.1%. The proposed model may help pig farmers to prevent deaths in the suckling phase of pigs at an early stage by taking countermeasure