Use of multivariate adaptive regression splines (MARS) for predicting parameters of breast meat in quails

Abstract

The aim of this study was to determine the effects of variety and sex on the color of the breast meat (brightness: L*, red color: a*, yellow color: b*) in quails. In this study, a total of 144 quails from three different varieties (Wild-type, Dark Brown and Golden) were employed. The color and pH parameters of the breast meat were measured in quails slaughtered in week 10. In order to predict the brightness (L*), red color (a*), and yellow color (b*) values of the breast meat, Multivariate Adaptive Regression Splines (MARS) models were implemented. When determining the best model, attention was paid to minimize the Generalized Cross Validation (GCV), Root Mean Square Error (RMSE), and Mean Absolute Deviation (MAD) statistics and to maximize coefficient of determination (R2) and adjusted R2 values. In the MARS models constructed to predict L*, a* and b*, it was found that R2 values were 0.999, 0.999, and 0.999; adjusted R2 values were 0.997, 0.992, and 0.996; and RMSE values were 0.068, 0.082, and 0.038, respectively. As a result, it could be suggested that MARS modeling may be a useful tool for the prediction of the color parameters of the breast meat.