Robust regression models for predicting the lean meat proportion of lambs carcasses

• Autor(a) principal: Xavier, Cristina
• Data de Publicação: 2012
• Outros Autores: Cadavez, Vasco
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10198/17609
• topic: Carcass; Quality; Ordinary least squares; Robust regression

The aim of this study was to develop and evaluate robust regression models for predicting the carcass composition of lambs. One hundred and twenty lambs (34 females and 86 males) were slaughtered and their carcasses were cooled for 24 hours. The subcutaneous fat thickness (C12) was measured between the 12th and 13th rib, and the left side of carcasses was dissected and the proportions of lean meat (LMP) was calculated. A multiple regression model was fitted using robust regression (RR) methods, and the results were compared to ordinary least squares (OLS) estimates. For RR methods, the Bisquare and Welsch weighting functions were used, and model fitting quality was evaluated by the following statistics: the root mean square error (RMSE), the median absolute deviation (MAD), the mean absolute error (MAE), and the coefficient of determination (R2). The parameters obtained by RR presented lower standard error for C12 measurement (decreases by 12% when compared with OLS estimates). The RR methods or weighted least squares methods represents a good alternative to OLS approach for modelling the LMP of lambs carcass. In this study, the Bisquare weighting method presented the best results, however other weighting functions are available and should be tested and compared in the near future.