Data Mining Strategies for Understanding Birth Patterns in Dairy Cattle: Single and Multiple Birth Analysis
Authors:
- Mostafa Ghaderi‐ZefreheiMGMostafa Ghaderi‐Zefrehei
- Maryam Montazeri‐NajafabadiMMMaryam Montazeri‐Najafabadi
- Farjad RafeieFRFarjad Rafeie
- Effat Nasre EsfahaniEEEffat Nasre Esfahani
- Mohammad Reza Abbaszadeh Bavi SoflaeiMSMohammad Reza Abbaszadeh Bavi Soflaei
- Mojtaba KafiMKMojtaba Kafi
- Mohammad Hossein BanabaziMBMohammad Hossein Banabazi
Open Access
- 25 March 2026
- journal article
- Published by Wiley in Veterinary Medicine and Science
- Vol. 12 (2) , e70890
Abstract
Data mining is a robust methodology designed to predict feature performance, identify trends within datasets, classify data based on similarities and optimize resource allocation. In the context of dairy farming, multiple birth is generally regarded as an undesirable outcome due to its adverse effects on cow fertility rates and the health of multiple calves, ultimately impacting farm profitability. This study conducts a comprehensive evaluation of various machine learning algorithms for the classification of multiple birth data using the Waikato Environment for Knowledge Analysis (WEKA) platform. The performance of 21 algorithms was evaluated using multiple metrics, including mean absolute error (MAE), root mean square error (RMSE), Kappa statistic, classification accuracy and computational time. The results reveal significant variability in the performance of the algorithms. Notably, the radial basis function (RBF) and support vector machine (SVM) algorithms outperformed others, achieving the lowest MAE and RMSE values, along with the highest Kappa statistics and classification accuracy on both training and test datasets. Specifically, the RBF model achieved a perfect Kappa statistic of 1 on the training set, indicating flawless agreement between predicted and actual classifications. In contrast, algorithms such as Voted Perceptron and Simple Cart exhibited higher error rates and lower classification accuracy, rendering them less suitable for this task. The analysis also highlighted trade‐offs between accuracy and computational efficiency, with algorithms like K ‐nearest neighbours ( K NN) and Random Forest providing a balanced performance. These findings emphasize the critical importance of selecting appropriate algorithms and evaluation metrics to achieve robust and reliable classification outcomes in multiple birth data analysis.Keywords
This publication has 64 references indexed in Scilit:
- Detecting heat events in dairy cows using accelerometers and unsupervised learningComputers and Electronics in Agriculture, 2016
- Recognition of culling reasons in Polish dairy cows using data mining methodsComputers and Electronics in Agriculture, 2016
- Statistical modeling of calving ease and stillbirths in dairy cattle using the classification tree techniqueLivestock Science, 2013
- Detection of cows with insemination problems using selected classification modelsComputers and Electronics in Agriculture, 2010
- Effect of calving ease on the subsequent performance of cow and calf in UK Holstein-Friesian cattleAnimal - science proceedings, 2010
- Associations between age at first calving and subsequent performance in Irish spring calving Holstein–Friesian dairy cowsLivestock Science, 2009
- Mastitis detection in dairy cows by application of neural networksLivestock Science, 2008
- Ovulation failure and double ovulation in dairy cattle: risk factors and effectsTheriogenology, 2005
- Support-vector networksMachine Learning, 1995
- Genetic Aspects of Twinning in Dairy CattleActa Agriculturae Scandinavica, 1974