Application of Principal Component Regression in Estimating the Body Weight of SimPO and LimPO Cattle Using Body Measurements

Abstract This study aims to apply Principal Component Regression (PCR) to estimate the body weight of SimPO and LimPO cattle based on body measurements and compare its accuracy with conventional regression methods which often encounter multicollinearity issues when using multiple correlated variables. This study involves 115 SimPO and 114 LimPO cattle, aged 2.5-4.5 years, non-pregnant, and easy to handle during measurement and weighing. Data on the body measurements including body length (BL), chest girth (CG), shoulder height (SH), and chest width (CW), and body weight (BW) of SimPO and LimPO cattle were collected from the Maju Sejahtera Livestock Production Cooperative farms. Principal Component Regression (PCR) Analysis of the body measurements, and multiple linear regression analysis were conducted using the R software. The selection criteria for the best model was based on a high coefficient of determination (R 2 ) and Adjusted R 2 , as well as low values of RSE (Residual Standard Error), AIC (Akaike Information Criterion), and BIC (Bayesian Information Criterion). Results indicate that PCR models offer comparable accuracy to conventional models while reducing redundancy and simplifying the prediction process. The body measurements of SimPO cattle were found to be more reliable predictors of body weight, as evidenced by higher R 2 and adjusted R 2 values and by lower RSE, AIC, and BIC, compared to LimPO cattle. The study concludes that PCR is an effective method for improving the accuracy and efficiency of body weight estimation in livestock, providing a valuable tool for farmers and researchers in the cattle industry.