Using machine learning procedures to ascertain the influence of beef carcass profiles on carcass conformation scores.

In this study, a total of 163 young-bull carcasses belonging to seven Spanish native beef cattle breeds showing substantial carcass variation were photographed in order to obtain digital assessments of carcass dimensions and profiles. This dataset was then analysed using machine learning (ML) methodologies to ascertain the influence of carcass profiles on the grade obtained using the SEUROP system. To achieve this goal, carcasses were obtained using the same standard feeding regime and classified homogeneous conditions in order to avoid non-linear behaviour in grading performance. Carcass weight affects grading to a large extent and the classification error obtained when this attribute was included in the training sets was consistently lower than when it was not. However, carcass profile information was considered non-relevant by the ML algorithm in earlier stages of the analysis. Furthermore, when carcass weight was taken into account, the ML algorithm used only easy-to-measure attributes to clone the classifiers decisions. Here we confirm the possibility of designing a more objective and easy-to-interpret system to classify the most common types of carcass in the territory of the EU using only a few single attributes that are easily obtained in an industrial environment.