Multivariate Adaptive Regression Splines Data Mining Algorithm for Prediction of Body Weight of Hy-Line Silver Brown Commercial Layer Chicken Breed

Abstract

Multivariate Adaptive Regression Splines (MARS) data mining algorithm is a non-parametric regression method employed to obtain the prediction of live weight by using body measurements. The study was conducted to investigate the relationship between body weight, linear body measurement traits and the effect of linear body measurement traits on body weight of Hy-Line silver brown commercial layer. A total of one hundred (n= 100) Hy-Line silver brown commercial layers aged 22 weeks were used for body measurements viz; body weight (BW) in kilograms, Beak Length (BK), Body Length (BL), Body Girth (BG), Shank Length (SL) and Wing Length (WL) in centimetres. Furthermore, Pearson correlation and MARS methods were used for data analysis. Correlation results revealed that BW had a negative statistically high significant correlation with WL (r= -0.48**) and BL (r= -0.61**). MARS results developed a non-parametric regression model with coefficient of determination (R2) = 1, adjusted coefficient of determination (R2 adj.)= 1, standard deviation ration (SD ratio) = 0.006, root mean square error (RMSE) = 0.001 and Pearson correlation (r) = 1 between predicted and actual values (P < 0.01) of body weight. MARS model revealed that WL and BL had an effect on BW of Hy-Line silver brown commercial layer. The findings suggest that WL and BL had an effect on BW, therefore chicken layer farmers might use WL and BL for selection during breeding to improve BW. In conclusion, MARS models developed in this study might be used by chicken layer farmers for selection during breeding.