Curvas de crescimento difÃsicas de fÃmeas Hereford com erros auto-regressivos e heterogeneidade de variÃncias / Difasic growth curves of Hereford females with auto-regressive errors and heterogenety of variances.

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

In the present study, the models of non-linear Difasic Logistic and Difasic Gompertz were adjusted, both weighed by the inverse of the variance with three different errors structures: independent errors (IE), first-class auto-regressive (AR (1)) and second-class auto-regressive (AR (2)) to weight-age data of 55 females of the Hereford race, raised in the Bagà region, RS, Brazil, evaluated since birth till 675 days old. The study objective is to compare these functions and shape the errors structure inherent to each adjustment. The Gauss-Newton interactive process made the esteem of the parameters to non-linear regression models. For this, it was used the model procedure of the Software Statistical Analysis System (SAS). Average and individual adjustments had been gotten for the animals in study. The comparison between the models was carried out through on the biological interpretation basis of the parameters and in the adjustment of quality appraisers (adjusted determination coefficient, Durbin-Watson test, residual standard desviation, iterations number), beyond the Akaike information criteria (AIC) and the F test to model comparison. Both studied functions, although showed good data adjustment, tend to sub estimate the adult animal weight, except when the errors structure AR (2) is considered. The individual adjustments were more accurate then the average adjustments, although the last ones needed less number of iterations till they reach convergence.

ASSUNTO(S)

heterogeneidade de variÃncias agronomia heterogenety of variances hereford race erros auto-regressivos modelos nÃo-lineares raÃa hereford models non-linear growth curves curvas de crescimento errors auto-regressive

ACESSO AO ARTIGO

Documentos Relacionados