Comparison for Fitting of Lactation Curve and Artificial Neural Network Models to Lactation Milk Yields in Saanen Goats
Project no. 21580 · Status: Kapanmış · 2020
Summary
In animal production, determining the milk yield abilities of the animals reared in the direction of milk yield is important for the continuity of the herd and the profitability of the enterprise. For this reason, it is necessary to know the milk yield in the lactation period, which starts with the birth of the animal and ends with the drying up, and the negative environmental factors affecting this period should be kept under control. Various statistical models, called lactation curve models, have been developed for the estimation of lactation milk yield. The success and effectiveness of these models vary according to the species under investigation. On the other hand, artificial neural networks (ANNs) are also nonlinear statistical models and are used in solving complex problems in many branches of science today. In this study, lactation milk yields of 40 pure Saanen goats were estimated by examining the environmental factors that are thought to be effective on the 15-day control day milk measurements taken individually, and by fitting the Wood and Cobby Le Du lactation curve and artificial neural network (ANN) models. The effects of birth type, parity, and body condition score on milk yields on the control day were examined and all factors were found to be not statistically significant (p>0,05). When the fitting abilities of the estimation models are compared, the corrected determination of coefficient (R2) is Root Mean Square Error (RMSE), Mean Absolute Deviation (MAD), and Mean Absolute Percent Error (MAPE) criteria were used in order to determine the best model. When the high R2 and low MAPE values, which are used as a basis for determining the model that will provide the best fit according to the mentioned performance criteria, are examined, it can be said that the ANN model is better than the lactation curve models. Keywords: Saanen, Lactation Milk Yield, Nonlinear Models, Artificial Neural Network