Prediction of Second Parity Milk Performance of Dairy Cows from First Parity Information Using Artificial Neural Network and Multiple Linear Regression Methods

摘要

A mathematical model for prediction of second parity milk yield and fat percentage, with the use of first parity information seems to be helpful in order to predict the performance of prospective productive cows. As a tool for this prediction, back propagation neural network and multiple linear regression methods were compared based on their prediction differences with observed values. While, multiple linear regressions are based on linear relationships between variables, artificial neural network system also considers non-linear relationships between parameters. Data was collected from 4 medium sized dairy herds in Isfahan, Iran, which was divided into three parts in order to train, verify and test the artificial neutral network system and estimation of regression coefficients, verify and test the multiple linear regression method. The results of the simulation showed that evaluations from both multiple linear regression and artificial neural network methods are good predictors for second parity production estimated from first parity information. However, artificial neural network predictions showed lower differences with the observed values and better quality parameters than multiple linear regression predictions, which made this assumption that artificial neural network system is more accurate in prediction.