Avaliação de curvas de crescimento em suínos

Abstract

This study aimed to simulate data to the weight of boar and pig sows (1 to 154 days old) through the uniform probability distribution on an interval [a, b]. Five regression models (logistic, hyperbolic tangent, power, gamma and cubic polynomial) were adjusted to account for the possible weight ratio to tage of pigs in days. The logistic and gamma cubic polynomial models showed the best estimates to account for the weight of the behavior of boar and sow pigs. The logistic model showed R2 = 0.975 and AIC (Akaike’s information criterium)= 13.34 for boars and R2= 0.9787 and AIC= 10.16 for sows. The gamma model presented a degree of explanation of R²= 0.991 and AIC= 99.33 for boars and R²= 0.9941 and AIC= 90.602 for sows, while the cubic polynomial model showed R²= 0.9995 and AIC= 64.2 for boars and R²= 0.9995 and AIC= 31.728 for sows. the cubic polynomial model was chosen to explain the behavior of the simulated weight of the pigs, whose model showed the best fit with fewer parameters than the other models.