The present paper aims to predict egg weight from some egg quality characteristics (egg weight, yolk weight, albumen weight, shell thickness, shell weight, shape index, albumen index, yolk index, yolk scale, and Haugh unit) in Japanese quails and to classify 94 eggs collected in regards to their shell colors. In the present study, CART and MARS data mining algorithms were assessed in the prediction of egg weight of the quails with the content of detecting egg quality standards of the studied quail genotypes for breeding and marketing strategies. In the classification of the collected quail eggs on their shell colors, classification performances of CART, CHAID, exhaustive CHAID, and QUEST algorithms were measured. Among those, CART was selected as the best classification algorithm according to eggshell color. The highest significant correlations were obtained for egg weight-yolk weight (0.740) and egg weight-albumen weight (0.735), respectively, in considering egg internal and external quality traits in quails. CART algorithm more accurately classified all eggshell colors compared with other algorithms. MARS showed much better predictive performance than CART that produced 0.850 Rsq and 0.728 cross-validation Rsq for prediction of egg weight in quails. In conclusion, the obtained results revealed that data mining algorithms may be useful references in practice for quail breeders in the development of new selection strategies and characterization of the studied animal materials.
Keywords: Data mining; Egg quality; Eggshell color; Quail.