Performance of Egg Sexing Classification Models in Philippine Native Duck

Egg sexing is one of the major concerns in the field of smart agriculture where local farmers and growers might be benefitted by increasing the number of laying ducks and eggs produced for additional income and supplies in the market. This study aims to present classification models for egg sexing of the Philippine Native Duck (Anas Platyrhynchos Domestica). In this work, 503 duck eggs were hatched where 403 are used for training and 100 for validating using KNN, SVM, Decision Tree, Linear Regression, and Naïve Bayes Models. Results show that the SVM classification model outperforms other classification models by attaining an accuracy rate of 87%. The morphological features such as eccentricity, sphericity, and shape Index affect the performance of the algorithms. Hence, the SVM algorithm can perform accurate classification given a good number of morphological features than other classifiers. This study presents some future work and practical recommendations.