Application des réseaux de neurones artificiels pour la prédiction du poids de l’oeuf de la cane locale « Anas platyrhynchos var domesticus »

Abstract

The objective of this study found to predict the whole weight of the egg from other quantitative characteristics of the eggs from a local breed of duck, for this a total of 102 fresh eggs were collected at the level of the Tiarat region. The eggs were cleaned, numbered and weighed, and then their internal and external characteristics were determined. The average egg weight was estimated to be 58.72g, while that of the shell; albumen and yolk were 5.74g, 31.4g and 21.58g respectively. Correlations between internal and external quality characteristics of duck egg were overall significant (P <0.01) with the exception of PH and Haugh unit. The principal component analysis, which was applied to the morph-weight characteristics of the egg, made it possible to retain five principal components which explained 85% of the total variance. Using these components, a multiple regression model was established to predict whole egg weight; (Whole egg weight = 21.578 + 2.363 * CP1-1.333 * CP2 + 0.22 * CP3 + 0.169 * CP4) with an accuracy of (adjusted R2 = 0.981). These results play an essential role in the design and implementation of breeding strategies for the improvement of genetic potential.