Formation of homogeneous groups of bovine milk production units via multivariate statistical techniques

ABSTRACT: Milk supply chain in Brazil exhibits significant production system heterogeneity in all federal units. Thus, the objective of this study was to form homogeneous groups of bovine milk production units based on the chemical and microbiological quality of the milk via multivariate statistical techniques. A total of 1,541 milk producing units (MPUs), corresponding to 44,089 samples, were analyzed. The first three principal components accounted for 81.38% of the total variation in the data. Principal component 1 (PC1) was associated with the chemical quality of milk (fat, protein [PROT] and total dry extract [TDE] content), while PC2 and PC3 were associated with microbiological quality (somatic cell count [SCC] and total bacterial count [TBC]). The concurrent analysis of the two two-dimensional projections characterized the different productive strata by their quality attributes and identified the positive/negative points of milk microbiological characteristics in each production group. Thus, the dimensionality of the set of 1,541 MPUs was reduced to 15 homogeneous production groups. This method optimizes the use of the dairy industry monthly database and characterizes all the heterogeneities present in dairy production systems.

摘要：巴西牛奶供应链在所有联邦辖区内均表现出显著的生产系统异质性。为此，本研究旨在借助多变量统计技术，基于牛奶的化学与微生物品质，对牛乳生产单元（milk producing units, MPUs）进行同质分组。本研究共分析了1541个牛乳生产单元，对应44089份样品。前三个主成分（principal component, PC）累计解释了数据集总变异的81.38%。主成分1（PC1）与牛奶化学品质相关，涵盖脂肪、蛋白质（protein, PROT）及总干提取物（total dry extract, TDE）的含量；而PC2与PC3则与牛奶微生物品质相关，涉及体细胞计数（somatic cell count, SCC）与总细菌计数（total bacterial count, TBC）。通过对两组二维投影的联合分析，本研究依据品质特征对不同生产层级进行了表征，并识别出各生产组中牛奶微生物学特性的优劣点。最终，1541个MPUs的数据集维度被降至15个同质生产组。该方法优化了乳业月度数据库的利用效率，并全面表征了乳业生产系统中存在的所有异质性。