蔬菜与水果的成分因子聚类分析

FACTORIAL CLUSTER ANALYSIS OF VEGETABLES AND FRUITS BASED ON THE COMPOSITION

本研究主要探索蔬菜与水果依据营养成分指标的分类问题。从《中国食物成分表2002》给出的256种蔬菜和162种水果中选择了具有完整数据的116种蔬菜和55种水果。利用SAS统计分析软件的FASTCLUS过程直接进行依据成分指标的快速聚类分析,未得到满意的分类结果。通过相关分析可以得知,有109对成分指标中等相关,有12对成分指标高度相关。对指标变量进行因子分析,试验发现,指定6个主因子比较适宜。当PR IORS选项为MAX时进行的方差最大因子旋转模型,能看出主因子较为明确意义。而且借助相应的主因子得分进行的快速聚类分析可得较为满意的分类结果。

The classifications of vegetables and fruits on the nutrition i ents are dealt with in this paper. The composition tables of 256 vegetables and 162 fruits are given in the book, China Food Composition 2002. 116 vegetables and 55 fruits with perfect data are slected. None of results of the FASTCLUS procedure on the composition variables are satisfied with the aid of SAS software. There are 12 correlative coefficients greater than 0.8 and 121 correlative coefficients greater than 0.5 in the correlation matrix of correlation analysis. The FAC- TOR procedure on the composition variables performs a variety of common factor and component analyses and rotations. It is suitable to assign 6 principal factors by experiments and comparisons. The significance of Rota- ted Factor Pattern can be found out when the option METHOD chooses PRINCIPAL and PRIORS selects / ASMC or MAX. Two satisfied results of the FASTCLUS procedure on 6 principal factor variables can be obtained.