Construction of Flavor Wheel for Indian Coffee Using Multivariate Exploratory Data Analysis

ABSTRACT Flavor wheels are chronicles of sensory descriptors arranged in a circle with unique colors that are used for sensory quality assessment. This study aimed to construct flavor wheels for grounded coffee from four Indian regions, such as Coorg, Chikmagalur, Shevaroy, and Bababudangiri. The volatile compounds of grounded coffee from four Indian regions were assessed using an E‐nose, and corresponding sensory descriptors were collected from the AroChem library. All sensory descriptors were statistically condensed to 13 sensory descriptors (fruity, floral, roasted, sweet, spicy, nutty, earthy, fermented, cocoa, sour, and chemical). The Sensory Descriptor Index was calculated from the 13 sensory descriptors and then subjected to Principal Component Analysis to understand the relationship between each sensory descriptor using a covariance matrix. The sensory descriptors were clustered and placed on the outer layer of the Flavor wheel by Agglomerative Hierarchical Clustering and pairwise Euclidean distance, respectively. The middle layer was constructed using key volatile compounds. The key volatile compounds contributing to each note, e.g., ethyl butyrate, 2‐methylpropanal (fruity), citronellol, nerol (floral), furfural (sweet/roasted), and methyl eugenol (spicy), were identified using VIP scores above one. The inner flavor wheel was constructed based on the classes of each volatile compound. The developed integrated the sensory descriptors, volatile compounds, and their chemical classes. The developed flavor wheels can be used as a sensory lexicon for coffee cupping training and analysis, quality control, and geographical identification.