Comparison of different data analysis tools to study the effect of storage conditions on wine sensory attributes and trace metal composition

Storing Cabernet Sauvignon wine at three different temperatures and in four different packaging configurations led to significant changes of the elemental and sensory profiles. Two different exploratory data analysis techniques-principal component analysis (PCA) and canonical variate analysis CVA)-were used to analyze both data sets separately to study the different data analysis outcomes. Additionally, partial least squares regression (PLSR) was used to correlate the two data sets to each other. Both unsupervised PCA and supervised CVA methods separated the samples due to the storage temperatures and packaging configurations, but showed some differences in the relative similarities between the treatments. While the sensory attributes changed to a larger degree as a function of the storage temperature, the elemental profile was most affected by the packaging configuration. Using the elements as predictor matrix for the sensory variables in the PLSR, no good model was found, indicating that the sensory changes cannot be solely explained by the metal changes. Overall, analyzing the same data set with different methods leads to similar but not identical conclusions, and depending on the research question, one method may be advantageous over the other. However, by comparing these methods we can gain a deeper understanding of the advantages, limitations, and potential applications of these statistical approaches, which can be useful in later analyses.