Paper ID: 55
REVISITING THE APPLICATION OF CLASSICAL FORMULAS TO ESTIMATE BALI CATTLE’S BODY WEIGHT BASED ON BODY MEASUREMENT VARIABLES
Nuzul Widyas1*, Luthfi Adya Pradista1, Rohmad Setiaji1, Ratih Dewi Hapsari2 and Sigit Prastowo1
1 Animal Science Department, Universitas Sebelas Maret Surakarta, Indonesia
2 Bali cattle Breeding Center Pulukan, Indonesia
Predicting cattle’s body weight is a common practice considering various reasons. Using the approximate of cattle’s body weight as observed variable in experiments, however, poses questions on how accurate are the estimates. This paper revisits four classical formulas commonly found in papers published by Indonesian researchers in predicting cattle’s body weight based on their body measurements namely Schoorl, Winter, Smith and Lambourne models. Data on body weight (BW) and body measurements (Chest Girth=CG and Body Length=BL) of 118 male and 106 female Bali cattle (2-3 yo) were collected from Bali cattle Breeding Center. The estimates from the prediction formulas were compared to the actual body weight. We run 10-folds cross validation procedure to obtain the predictive ability parameters. The mean BW, CG and BL for male cattle were 439.14±113.55lbs; 295.98±22.62 and 273.80±22.01 inches; whereas for females were 355.70±75.73lbs; 318.68±29.60 and 223.72±13.89 inches respectively. The accuracy of the estimates was defined as the correlation coefficient between actual and predicted BW. All four formulas have the accuracy between 84.90 to maximum of 89.68% in both male and female cattle groups. RMSE were considerably higher in both male group (38.97 – 73.44) and female group (25.26 – 45.81). Although the correlations between actual and predicted BW are high, further study need to be done to determine whether the utilization of these predicted values as a response variable will introduce enough bias to affect the results of a research. Or, alternative methods to predict BW with higher accuracy need to be explored.