PRINCIPAL COMPONENT ANALYSIS AND AN APPLICATION IN WINE DATA ANALYSIS

Abstract

Analyzing the relationship between chemical properties and wine quality will contribute to better wine’s price valuation. In this study, we use an availabe online dataset on white wines that produced from the "Vinho Verde" region in Portugal, which contains 11 variables measuring the chemical properties of white wines and 1 variable represents the quality of wine (ranked from 0 to 10 according to the sensory assessment of the wine tasters). The goal is to find chemical property variables that influence the wine quality classification. To achieve the goal, we propose to use principal component analysis in correlation and classification assessment. The results showed that PCA performs relatively well in classifying white wine based on its chemical properties. Chemical compositions such as residual sugar, total sulfur dioxide, density, alcohol, pH, and fixed acidity play an important role in the flavor of white wine. On the other hand, it was found that Alcohol and pH contributed highly to quality of the wine.