Del Campo Gloria, Santos J Ignacio, Iturriza Nuria, Berregi Iñaki, Munduate Arantxa
Applied Chemistry Department, Faculty of Chemistry, University of the Basque Country, P.O. Box 1072, E-20018 San Sebastian, Spain.
J Agric Food Chem. 2006 Apr 19;54(8):3095-100. doi: 10.1021/jf051818c.
The low field region (5.8-9.0 ppm) corresponding to aromatic protons and the region 1.8-3.0 ppm of the (1)H NMR spectra were used for characterization and chemometric differentiation of 52 apple juices obtained from six cider apple varieties. The data set consisted of 14 integrated areas corresponding to resonances from acids and phenolic compounds. Multivariate procedures based on hierarchical cluster and discriminant analysis were performed on selected signals of the spectra to determine whether it was possible to distinguish the different juices. Cluster analysis was able to satisfactorily classify the six apple varieties. Discriminant analysis, by means of stepwise procedure for variables selection and leave-one-out for cross-validation, was applied to 40 samples from the year 2001, obtaining recognition and prediction abilities of 100%. The most discriminant variables corresponded to poliphenols, (-)-epicatechin, phloridzin-phloretin, and p-coumaric, chlorogenic, and malic acids. The classification model was applied to 12 samples from apples harvested in the years 2002 and 2003, and the prediction ability was 91.7%.
对应于芳香族质子的低场区域（5.8 - 9.0 ppm）以及（1）H NMR光谱中1.8 - 3.0 ppm的区域，被用于对从六个苹果酒用苹果品种获得的52种苹果汁进行表征和化学计量学区分。数据集由14个对应于酸和酚类化合物共振的积分区域组成。基于层次聚类和判别分析的多变量程序对光谱的选定信号进行了处理，以确定是否能够区分不同的果汁。聚类分析能够令人满意地对六个苹果品种进行分类。判别分析通过变量选择的逐步程序和留一法交叉验证，应用于2001年的40个样本，获得了100%的识别和预测能力。最具判别力的变量对应于多酚、（-）-表儿茶素、根皮苷 - 根皮素以及对香豆酸、绿原酸和苹果酸。该分类模型应用于2002年和2003年收获的苹果的12个样本，预测能力为91.7%。
