Sheep category can be classified using machine learning techniques applied to fatty acid profiles derivatised as trimethylsilyl esters

Abstract

Eruption of permanent incisors (dentition) is used as a proxy for age for defining meat quality in Australian sheep meat. However, this approach may not be reliable. While not presently available, an objective method could be used to determine sheep age, and thus sheep category, which would then potentially remove any inaccuracies that may occur in classifying sheep meat product. Statistical classification algorithms have been successfully used in bioinformatics. In this paper we review the performance of three algorithms (support vector machines, recursive partitioning and random forests) for determining sheep age. The algorithms were applied to the measured fatty acid profiles of fat samples from 533 carcasses; 254 lamb (<1 year old), 131 hogget (~1–2 years old) and 148 mutton (>2 years old) samples. Three data pretreatments (range transformation, column mean centering and range transformation with mean centering) were also examined to determine their impact on the performance of the algorithms. The random forests algorithm, when applied to mean-centred data, gave 100% predictive accuracy when classifying sheep category. This approach could be used for the development of an objective test for determining sheep age and category.