Use of CART and CHAID Algorithms in Karayaka Sheep Breeding

Abstract

The aim of this study was to determine the effect of some factors (sex, birth type, farm type, birth weight and weighting time) on weaning weight through CART and CHAID data mining algorithms. The classification and regression trees are modern analytic techniques that construct tree-based data-mining algorithms. Regression trees are used for the purpose of preliminary selection of the traits affecting the continuous dependent variable. The studied data were consisted of 366 records from Karayaka sheep breed. The CHAID algorithms results revealed that; predictors such as weighting time, sex and farm type statistically influenced weaning weight Regression tree diagram constructed by CART algorithm depicted that birth type was effect the weaning weight, and in this tree weighting time of single born lambs was affected the birth type. The predicted values and original values were correlated (P<0.05). As a result, it could be suggested that CHAID algorithm was found more useful biologically than CART.