Unsupervised Machine Learning Techniques for Detecting Outliers in Dairy Cow Milk Yield Data

Strategic Insights -

A new study published in the Journal of Dairy Science has explored the application of unsupervised machine learning techniques in detecting outliers in daily milk yield data of dairy cows. The research aims to evaluate the suitability of these techniques in estimating the unperturbed lactation curve, which is essential for developing effective feeding plans and optimizing breeding for dairy farms.

Key Takeaways:

  • The study evaluated three unsupervised machine learning models, including one-class support vector machines, isolation forest, and local outlier factor, for detecting outliers in daily milk yield data.
  • The models were compared to two previously proposed models, the perturbed lactation model (PLM) and the iterative Wood model (IWM), in a simulation study using 1,000 simulated lactations.
  • The results showed that the unsupervised machine learning models outperformed the baseline Wood model in sensitivity and F-score, while maintaining comparable precision.
  • The study also applied the models to observed daily milk yield data from 2,831 lactation records of 1,636 Holstein cows collected over a 10-year period.
  • The results showed that the unsupervised machine learning models demonstrated relatively high computational efficiency and established ull curves that showed better goodness-of-fit and shapes more consistent with the baseline Wood curve.

Statistics:

  • 1,000 simulated lactations were used in the simulation study.
  • 4.00 (± 1.46) perturbations were included in each simulated lactation.
  • Sensitivities of 61% were achieved across all unsupervised machine learning models.
  • Precisions of 82% were achieved across all unsupervised machine learning models.
  • F-scores of 70% were achieved across all unsupervised machine learning models.
  • The unsupervised machine learning models outperformed the baseline Wood model in F-score (64.2% vs. 53.5%).
  • The unsupervised machine learning models outperformed the PLM in F-score (70% vs. 53.2%).
  • The unsupervised machine learning models outperformed the IWM in F-score (70% vs. 66.8%).

Sources:

  • Leveraging unsupervised machine learning techniques for detecting outliers in the daily milk yield data of dairy cows. Journal of Dairy Science, 2025;108(9):9696-9711.
  • Journal of Dairy Science: www.journals.elsevier.com/journal-of-dairy-science/
  • Elsevier Science Inc: www.elsevier.com