V. F. Akhmedli, S. Yu. Bakoev, D. R. Matyukov, O. N. Lukonina, L. V. Getmantseva
All-Russian Research Institute of Breeding, Lesnye Polyany settlement, Moscow region, Russia
E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
Volume 25 No. 10
Date of paper submission: 13.08.2025, date of review: 04.09.2025, date of acceptance: 15.09.2025.
Published: 10/31/2025
Abstract. Data quality and reliability are the cornerstone for making effective management, breeding, and scientific decisions in animal husbandry. Inaccurate data can lead to significant economic losses and erroneous scientific conclusions. The purpose of the study was to systematize and analyze the multi-level arsenal of methods developed for data verification and cleaning, assessing their effectiveness from simple deterministic rules to complex intelligent systems. Research methods. Traditional approaches are considered, including range checks, logical and contextual validations, which serve as a basic filter for gross errors. Further, statistical methods such as Z-scores and the interquartile range (IQR) are analyzed, identifying statistical outliers but primarily limited to univariate analysis. To address this limitation, the review extensively covers machine learning methods. The application of principal component analysis (PCA) for detecting multidimensional anomalies in phenotypic and productivity profiles is examined. Clustering methods, such as k-means for identifying atypical prototypes and DBSCAN for filtering data noise, are discussed. As the most advanced approach, decision trees and their ensembles (Isolation Forest), as well as neural network architectures particularly autoencoders capable of unsupervised learning to detect complex nonlinear patterns and hidden anomalies are presented. Research results demonstrate that a combination of these multi-level approaches enables a shift from error correction to a proactive strategy for ensuring data integrity. The scientific novelty lies in the fact that, for the first time, data verification methods from classical to modern (machine learning) have been systematized and compared, with an emphasis on their applicability in animal husbandry. This is crucial for the advancement of modern livestock farming, minimizing economic losses, and improving breeding decisions.
Keywords: data quality control, livestock farming, machine learning, PCA, clustering, autoencoders, anomalies, statistical methods, biological norms, intelligent analysis
Acknowledgments. The study was carried out in accordance with the State Assignment No. 082-00240-25-00 for 2025. Agreement No. 082-03-20254-011.
For citation: Akhmedli V. F., Bakoev S. Yu., Matyukov D. R., Lukonina O. N., Getmantseva L. V. Data quality control in livestock farming based on biological norms and intelligent analysis. Agrarian Bulletin of the Urals. 2025; 25 (10): 1589‒1598. https://doi.org/10.32417/1997-4868-2025-25-10-1589-1598 (In Russ.)
Download the full text of the article
