Prediction of Inner Quality Characteristics of Eggs Using Partial Least Squares Regression

Abstract

This study was carried out to obtain a prediction model for egg albumen and yolk weight, which are the internal quality characteristics of egg predicted from external quality characteristics of egg. For this purpose partial least squares regression method was applied to the data set used in the study and the results were compared with the principal component regression method. In the partial least squares regression analysis for egg albumen and yolk weight, the number of latent factor was 1 and the determination coefficients were 68.44% and 63.40%, respectively. For the egg albumen weight, the coefficients of determination for the principal component regression with one latent factor were 63.40% and 53.80%. When there is no restriction for the number of factors in the principal component regression, for the egg albumen weight the number of latent factors was five and the coefficients of determination was 79.77%; for the egg yolk weight the values were two and 75.35%, respectively. The results shown that the partial least squares regression method was more effective than the principal component regression method in dimension reduction, and more reliable predictions can be obtained in small sample sets with multicollinearity using the partial least squares regression method. © 2018, Centenary University. All rights reserved.