Comparison of models for some testicular characteristics in Karakaş male lambs

Abstract

This study was aimed to search testis growth of Karakaş male lambs raised in intensive conditions and to build the forecasting model of live weight in lambs using 4 data mining CHAID, Exhaustive CHAID, CART, and multivariate adaptive regression splines (MARS). Effects of genotype and some environment factors such as birth type, dam age, lamb' age in control and live weight on testis characteristics were also researched. The furthest importance sequence was obtained for testis length (TESLENG) (100%), followed by age, testis diameter (TDIA), age in control (DAGE), birth weight (BW), around scrotum (SCC), scrotum length (SCL) and birth type (BT). TDIA > 8.5 and DAGE will provide 21.06 unit to live weight (LW). BW > 4.2 will provide 18.43 kg to LW. TESLENG > 8.5, SCC < 2.4 and SCL will provide 8.227 to LW. TDIA < 4.5 and TESLENG > 9 will provide 4.095 to LW. Analytic results shown that MARS outperformed CHAID, Exhaustive CHAID, and CART approaches in terms of R2, Adj. R2, RMSE, SD ratio, MAPE and AIC provided effective alternatives for forecasting live weight of lambs. Purpose of MARS algorithm having very high predictive precision in the prediction of LW from body features may put onward a new view point for 39 lambs breeders.