Characterizing Shared Sensory Priorities and Age-Group Differences for Feng-Flavor Baijiu: A Consumer Survey Using Hierarchical Weighting and Interpretable Machine Learning.

As Feng-flavor Baijiu increasingly shifts toward consumer-oriented product development, clarifying consumers' shared sensory priorities and age-related differences is essential for aligning its traditional style with evolving market demands. This study collected 1523 questionnaires across China, of which 1357 valid responses were analyzed. A hierarchical sensory system comprising three primary dimensions-aroma, taste, and drinking perception-and 24 secondary attributes was evaluated using hierarchical weighting, a proportional sensory flavor wheel, machine learning, and SHAP interpretation. Drinking perception received the highest weight (35.23%), although it was only slightly higher than aroma (33.54%) and taste (31.22%). Jiuhai aroma, sweetness, comfort, and purity were identified as the leading attributes within their respective dimensions. Machine learning further identified purity, persistence, saltiness, comfort, and harmony as attributes with relatively high contributions to age-group classification. Age-group differences were not concentrated in any single sensory attribute but were reflected in the classification information jointly provided by multiple sensory attributes. Descriptive group comparisons showed decreasing trends in the relative weights of saltiness and sourness across the 18-30-, 31-45-, and ≥46-year groups, although neither difference remained statistically significant after Benjamini-Hochberg correction. Comfort, bitterness, and honey aroma were relatively more prominent in the ≥46-year group, whereas grain aroma was relatively more prominent in the 31-45-year group. Sweetness and Jiuhai aroma had relatively high overall weights but contributed less to age-group classification, suggesting that they may represent more broadly shared sensory priorities across age groups. These findings identify candidate sensory priorities and segmentation indicators for subsequent validation using actual products, consumer liking tests, and ideal-intensity assessments.