Comparison of several growth models for cattle
2001
Nesetrilova, H. (Ceska Zemedelska Univ., Prague-Suchdol (Czech Republic))
Five growth models (logistic, Gompertz, Richards, Morgan-Mercer-Flodin and Weibull) and their suitability for modelling of body growth in cattle are studied in this paper. As these models are nonlinear regression models and their properties can be studied only in the combination model/data, growth data on breeding bulls of Czech Pied cattle were used for the purposes of the present analysis. Residual variance was used as a criterion of the quality of fit. This criterion was supplemented by statistical properties of the model parameter estimators, degree of nonlinearity of individual models for the considered data and visual inspection of the growth curve. Considering these criterions, the Gompertz model was the better of the three-parameter models in parametrization 2, while among the four-parameter models, Richards model proved to be unsuitable for the data due to the high values of residual variance and high degree of nonlinearity. The remaining two four-parameter models (Morgan-Mercer-Flodin and Weibull) had low residual variances but the curves lacked elasticity and both models overestimated the position of the asymptote which is important for breeders (final animal weight).
Показать больше [+] Меньше [-]Ключевые слова АГРОВОК
Библиографическая информация
Эту запись предоставил Library of Antonin Svehla