Predicting live weight of reindeer using a regression model

Relevance. Accounting for live weight in reindeer husbandry is an important indicator for breeding, but due to the difficulty of determining it, there is a need to develop a predictive matrix of animal weight through biometric exterior measurements. The purpose of the study is to develop a model for predicting the live weight of reindeer using regression analysis. Methods. The research was carried out using exterior measurements and weighing results of Nenets reindeer (males n = 48, females n = 50) from the Tazovsky district on the Yamal Peninsula at the age of 2 to 9 years. Calculation of statistical parameters, visualization of correlation data and regression analysis using the least squares method were carried out in MS Excel and R-studio. Results. The greatest variability among exterior measurements was noted in the indicators “chest width CW” (9.6%) and “live weight LW” (9.4%). When conducting a correlation analysis, multicollinearity was revealed between the height at the withers HW and the height at the elbow HE r = 0.824 (p ≤ 0.001). High and significant correlations of live weight LW with chest depth CD and chest girth CG r = 0.651 and r = 0.687 (p ≤ 0.001), head length HL r = 0.678 (p ≤ 0.001), height at withers HW r = 0.663 (p ≤ 0.001) and body length BL r = 0.639 (p ≤ 0.001). The most effective model m2 was determined, including chest girth and body length in its structure, the coefficient of determination of which was R2 = 0.70, with multiple R = 0.83, which reflects 70% of the explained variable in the model, with an approximation of 4.2%. As a result, a table was created for predicting the live weight of reindeer using biometric data, which will help simplify selection and breeding work in future populations of hard-to-reach areas.