主成分分析和聚类分析在配方模块中的应用

Application of Principal Components Analysis and Cluster Analysis in Evaluation of Blending Module

为探索一种科学合理的小叶组配方烟叶质量评价与分类方法,对云南省楚雄州不同产地的初烤中部烟叶样品的外观品质因素指标、常规化学成分检测指标和感官质量评吸指标进行了主成分分析,并用快速聚类法(K-均值聚类法)对得到的主成分进行聚类分析;利用统计分析软件SPSS11.0提供的方法进行快速样本聚类,按聚类结果确定其在小叶组配方中的用途。结果表明,采用这种多元分析方法可以克服传统理化和感官指标检测方法对多个样品难以综合评价分类的缺陷,且运用计算机处理数据简便易行,是进行小叶组配方烟叶评价分类的一种便捷方法。

In order to develop a scientific and reasonable method for quality evaluation and classification of sub_blends, the factor analysis of appearance quality, routine chemical components and indexes of sensory evaluations of cured cutters of flue-cured tobacco grown in different areas in Chuxiong was carried out with principal components analysis, and the obtained principal components were fast clustered by K-mean clustering algorithm; the samples were further fast clustered with statistical analysis software SPSS11.0, thus the function of individual sample in a sub_blend was determined. The results showed that the multivariate analysis overcome the shortcomings of traditional physical, chemical and sensory index tests which failed to comprehensively evaluate and classify between numerous tobacco samples. The method is simple and fast with computer aided data processing.