Comparison of the data mining and machine learning algorithms for predicting the final body weight for Romane sheep breed

• DOI: 10.1371/journal.pone.0289348
• Publication Date: 2023-08-03T17:40:15Z
• Authors: Cem Tırınk; Hasan Önder; Dominique Francois; Didier Marcon; Uğur Şen; Kymbat Shaikenova; Karlygash Omarova; Thobela Louis Tyasi
• Keywords: 2. Zero hunger; [SDV.BA.MVSA]Life Sciences [q-bio]/Animal biology/Veterinary medicine and animal Health; Sheep; [SDV]Life Sciences [q-bio]; Body Weight; 610; Weaning; [INFO.INFO-MO]Computer Science [cs]/Modeling and Simulation; Machine Learning; Artificial Intelligence; Animals; Birth Weight; Algorithms; Research Article

The current study aimed to predict final body weight (weight of fourth months age select the future reproducers) by using birth weight, type, sex, suckling at weaning and for Romane sheep breed. For this purpose, classification regression tree (CART), multivariate adaptive splines (MARS), support vector machine (SVR) algorithms were used training (80%) testing (20%) sets. Different data mining learning 393 (238 female 155 male animals) with different artificial intelligence algorithms. best prediction model was obtained CART model, both set. Constructed models indicated that could be as an indirect selection measure get a superior flock on sheep. If genetically established, whose sex is female, over 142 days, 28 kg chosen affording genetic improvement in weight. In conclusion, usage procedure may worthy reflection identifying breed standards choosing meat yield France.