Comparison of tree-based regression tree methods for predicting live body weight from morphological traits in Hy-line silver brown commercial layer and indigenous Potchefstroom Koekoek breeds raised in South Africa.

In animal breeding, more considerable attention is drawn to reveal the relationship between live body weight and morphological traits in identifying breed and species standards. The aim of this study was to predict live body weight from morphological characteristics in the Hy-line silver brown commercial layer and indigenous Potchefstroom Koekoek breed, native to South African. In the prediction of live body weight, eleven morphological measurements, i.e., wing length, back length, beak length, shank length, shank circumference, chest circumference, wingspan, keel length, body girth, toe length, and body length, were taken. As tree-based regression tree methods, predictive performances of CART, CHAID, and exhaustive CHAID algorithms were measured for body weight prediction. Among those, CART was found to be the best decision tree algorithm that gave the highest predictive accuracy. CART visual results reflected that the heaviest body weight mean (2.000 kg) was obtained from the chickens with 10.250 cm < WL ≤ 10.500 cm. As a result, it could be suggested that the CART decision tree might help to determine breed standards of the Hy-line silver brown commercial layer and, especially, indigenous Potchefstroom Koekoek breeds for breeding programs.