基于主成分和聚类分析的曲拉品质的综合评价

为提高牧区牦牛曲拉品质的一致性,对我国曲拉主产区的曲拉样品进行综合评价。采集8个地区95份牦牛曲拉样品,对其营养成分、抗氧化指标、色度值及5-羟甲基糠醛(5-hydroxymethylfurfural,5-HMF)进行测定与分析,应用主成分分析筛选曲拉品质评价指标,同时通过聚类分析对曲拉样品进行分类,运用方差分析对曲拉进行综合评价。结果表明,根据相关性分析得到大多数指标间均存在极显著(P0.01)或显著(P0.05)相关性;主成分分析筛选出2个主成分因子,PC1(72.846%)为外观色泽因子,PC2(13.763%)为营养品质因子,L*值、a*值、b*值以及5-HMF含量在PC1上的载荷因子均在0.9以上,说明外观色泽是评价曲拉品质的主要指标;聚类分析可将95份曲拉样品分为4类,且此分类结果与主成分分析结果基本一致。该4类曲拉样品品质存在极显著差异(P0.01),其中总体14.74%左右的曲拉样品品质不佳,69.47%左右的曲拉样品品质良好,15.79%左右的曲拉样品品质优良。

Comprehensive Evaluation of the Quality Qula,Dried Residue of Naturally Fermented Skim Yak Milk,Based on Principal Component Analysis and Cluster Analysis

In order to improve the quality consistency of Qula, the dried residue of naturally fermented skim yak milk, in pastoral areas of China, a comprehensive quality evaluation was conducted for Qula samples collected from some major producing areas in the country. A total of 95 Qula samples from 8 producing areas were analyzed for nutritional composition, antioxidant properties, color values and 5-hydroxymethylfurfural (5-HMF). The appropriate indicators to evaluate the quality of Qula were screened by principal component analysis (PCA) and the samples were classified by cluster analysis. At the same time, analysis of variance was used to comprehensively evaluate the quality of Qula samples. The results showed that majority indicators were extremely significant (P < 0.01) and significant (P < 0.05) correlation. The first and second principal components (PC1 and PC2) identified by PCA, accounting for 72.846% and 13.763% of the total variance, were interpreted as a ‘color component’ and a ‘nutritional quality component’, respectively. The PC1 loading factors for color L*, a* and b* values and 5-HMF content were all higher than 0.9, suggesting that the color could be considered the major indicator of Qula quality. Cluster analysis suggested that these Qula samples were classified into four categories, showing good consistency with the result from principal component analysis. There were extremely significant differences between four categories of Qula samples in terms of quality indicators (P < 0.01). Approximately 14.74% of Qula samples were bad in quality, 69.47% good and 15.79% excellent.