Akademik Veri Yönetim Sistemi
4. INTERNATIONAL CAPPADOCIA SCIENTIFIC RESEARCH CONGRESS, Nevşehir, Türkiye, 16 - 17 Nisan 2023, ss.271-272, (Özet Bildiri)
COMPARISON OF THE EGG PRODUCTION CURVE OF CHUKAR PARTRIDGES (ALECTORIS CHUKAR) WITH NON-LINEAR MODELS AND ARTIFICIAL NEURAL NETWORKS (ANN)
ABSTRACT
Chukar partridges are known as a game bird and although generally live in wild, raised in intensive system for restock the hunting stocks throughout the world. Therefore, the level of egg production of partridges is crucial. The study was aimed to determine the best fit egg production curve of the Chukar partridges during the laying period (age at 34-51 weeks) using with 7 non-linear models (Wood, Compartmental, Modified Compartmental, McNally, Adams Bell, Lokhorst, Narushin Takma) and Artificial Neural Networks (ANN). As a research material, 1050 Chukar partridges were used which reared in semi-open cages dimension of which 6.0 × 1.2 × 1.5 meters in outdoor. During the laying period, the partridges were allowed ad-libitum access to water and diet. Parameters of the non-linear models were estimated using with curve fitting toolbox application of MATLAB R2020b software. In ANN model, ANN toolbox application of MATLAB R2020b software was used to estimate hen-day egg production. ANN model was designed 70% training, 15% validation, 15% testing and as iteration criteria Levenberg–Marquardt algorithm was used. To determine best fit egg production curve model coefficient of determination (R2), mean square error (MSE), root mean square error (RMSE), mean absolute deviation (MAD) and mean absolute percentage error (MAPE) were used as goodness of fit criterias. According to the goodness of fit criterias ANN was determined the best fit egg production curve (R2: 0.981, MSE: 0.0003, RMSE: 0.0170, MAD: 0.0112, MAPE: 6.7025). Followed by ANN, it can be said that Narushin Takma, Adams Bell and Lokhorst model could be used as second-best fit model because of the fact that gave closer results each other. In addition, Compartmental model gave the worst results (R2: 0.746, MSE: 0.0038, RMSE: 0.0620, MAD: 0.0516, MAPE: 110.5419). In conclusion, Artificial Neural Networks was determined the best fit egg production curve for Chukar partridges which would be advisable for Chukar breeders to estimate the hen-day egg production.
Keywords: Artificial Neural Networks, Chukar Partridge, Egg Production Curve, Hen-day Egg Production, Non-linear Regression