Classification of sheep genetic resources based on morphometric characteristics using classical and machine learning methods

One classical method-linear discriminant analysis and two machine learning methods-artificial neural network (ANN) and support vector machine (SVM) were used to classify animals of eight sheep breeds of India (Kolhapuri, Lonand, Marwari, Munjal, Muzaffarnagri, Madgyal, Sangamneri and Solapuri) on the basis of morphometric characters. Atotal of 1558 adult ewes were included in the study. The set ratio of training and test data consisted of 70:30. Thirty experiments were conducted using tuned characters on each set of data. The study showed that performance of three methods was similar while doing pair-wise classification of breeds from distant regions. When the breed-pair have neighbouring habitats and closer in morphometric characters, SVM was the best method. When a breed is compared with combined number of other breeds, SVM was the best method for classification of sheep genetic resources.