针对目前评酒员对葡萄酒评价存在较大的主观性和不稳定性的问题,该文先通过主成分分析法对酿酒葡萄和葡萄酒的理化指标进行降维,得到几组能分别代表一类理化指标的主成分.然后,分别讨论是否引入芳香物质指标两种情况,将各主成分与葡萄酒质量进行了两次多元线性拟合.结果发现,引入芳香物质能显著优化对白葡萄酒的拟合效果,但对红葡萄酒的优化不明显.故红葡萄酒的质量可用理化指标来评价;而白葡萄酒的质量则需用理化指标和芳香物质来评价.
There is great subjectivity and instability among the evaluations of wine critics. This paper first reduces the dimensions of physicochemical indexes of grapes and wine through principal component analysis, getting several principal components, each of which can represent a class of physicochemical index. Then, the paper discusses whether the aromatic substances indexes should be added, and sets up multiple linear regressions between the components and the wine quality twice. The paper finds that the introduction of aromatic substances can significantly optimize the fitting effect of white wine, but the optimization is not obvious for red wine. Therefore, the quality of red wine can be evaluated only by the physicochemical indexes, while the quality of white wine needs to be evaluated by both physicochemical indexes and aromatic substances.