Prediction of Milk Production in Dairy Cows Using Statistical Regression Model and Machine Learning Methods

Prediction of milk production from dairy cows can be used as an important indicator to evaluate the productivity of livestock farms. This study proposes methods to predict milk production from dairy cows raised in livestock farms by period. To this end, the study was conducted in three steps. First, a livestock farm in Boseong-gun, Jeollanam-do was selected, and feed intake, age, parity, number of milking days, and milk production were collected for each of 49 dairy cows. Second, based on the collected data, statistical analysis was performed to find out how much the four characteristics affect milk production through scatter plots and correlation analysis. Third, in order to accurately predict the milk production of each dairy cow by period, four prediction models were used: the linear regression model, random forest regression model (RFRM), XGBoost technique and 1D-CNN+BiLSTM. Through various experimental results, it was found that the machine learning methods, random forest and XGBoost methods, had the best predictive power, followed by 1D-CNN+BiLSTM model, while the linear regression model had the worst predictive power. In addition, through the XGBoost analysis technique, it was found that the variables that had the greatest influence on milk production were age, milking days, feed intake, and parity, in that order.