Comparison of Multivariate Adaptive Regression Splines and Classification Regression Tree for Prediction of Body Weight of Bapedi Sheep

The study aimed to compare the performance of multiple adaptive regression splines and classification regression trees for the prediction of the body weight of Bapedi sheep. A total of 100 Bapedi sheep aged between one and five years old of different sexes were employed. The study measured the following: Body Length (BL), Withers Height (WH), Heart Girth (HG), Rump Height (RH), Body Weight (BW) and Sternum Height (SH). The model's performances were evaluated using goodness of fit criteria while the association between body measures and BW was discovered using a correlation matrix. Multivariate Adaptive Regression Splines (MARS) and Classification and Regression Tree (CART) established the model for the prediction of BW. The findings indicated that CART performed well. Correlation matrix results indicated that BW had positive statistical significance (p<0.05) with SH (r = 0.53) and BL (0.47) and a statistically significant relationship (p<0.01) with HG (r = 0.89), WH (r = 0.74) and RH (r = 0.64). CART model indicated that HG, BL, and BL could be used to predict BW while MARS indicated that HG, the interaction of AGE and HG, and the interaction of AGE, BL, and HG play a role in the prediction of BW. The CART model appears to be the most effective model for predicting BW based on goodness of fit results. The correlation results imply that HG can be applied to enhance the BW of Bapedi sheep. CART and MARS model results suggest that HG and the interaction of AGE and HG are the best explanatory variables of BW