Número mínimo de pesagens para estimação dos parâmetros de curvas de crescimento para ovinos de corte cruzados

• Autor(a) principal: Cordeiro, Liliane Lopes
• Data de Publicação: 2011
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: LOCUS Repositório Institucional da UFV
• Texto Completo: http://locus.ufv.br/handle/123456789/4036
• topic: Análise de diagnóstico; Modelo não linear; Ovinos cruzados; Qualidade de ajuste; Diagnostic analysis; Non-linear model; Sheep crossed; Quality of fit; CNPQ::CIENCIAS EXATAS E DA TERRA::PROBABILIDADE E ESTATISTICA::ESTATISTICA

This study aims to verify the effect of restrictions in the data set as the minimum number of weighings per animal on the nonlinear models parameters estimation of growth curves for sheep crossed. It was used data on average weight for age and individual weight for age of 74 crossbred, from the Texel and Santa Inês cross grown in the northeastern region in order to estimate the growth curves parameters. It was used seven data structures with restriction of at least 5, 6, 7, 8, 9 or 10 weights per animal and without restriction. Brody, von Bertalanffy, Logistic and Gompertz models were fitted to weight-age data. The quality of the models adjustment according to the restrictions on data was evaluated by the coefficient of determination (R2 aj), Mean prediction error (MPE) and Mean Square Residual (MSR). Usual diagnostic techniques, such as leverage points, aberrant points and influential measures were used to identify atypical points. All the structures presented estimates for mature weight, maturity index, instantaneous growth rate, relative instantaneous growth rate and inflection point very close and consistent to the ones presented by the literature. Once the values of R2 aj, EPM and QMR were close, it can be concluded that with or without restriction, the quality of fit was very similar, then the use of data with large variation in the number of weighings per animal (4-13) did not compromise the quality of the models adjustments. It can be concluded also that it is possible to obtain good fit, even for data set with few animals, since they present with a greater frequency.