Meta-attributes in sensory descriptive analysis
Descriptive analysis was conducted by a trained panel on potato varieties, forty in 2010 and forty-four in 2014. The panel evaluated 52 well-defined sensory attributes on line- scales anchored at 0 and 100. Our objective was to determine if there were groups of homogeneous attributes, or “meta-attributes.” We utilized a well-known exploratory procedure and fitted a parsimonious Gaussian mixture model (PGMM) to each data set. The Bayesian information criterion (BIC) was used to select the number of components and the number of latent factors. We initialized the PGMMs in two different ways and compared the meta-attributes found within and between each data set. Our results indicated the presence of at least two meta-attributes. We propose that these meta-attributes be used to determine product dissimilarity in the construction of a sensory informed design for consumer research (SID; Franczak et al., 2015) or to reduce the number of attributes needed when performing quality control.
Franczak, B. C., Findlay, C. J., & McNicholas, P. D. (2015). Meta-attributes in sensory descriptive analysis. In: 11th Pangborn Sensory Science Symposium. 23-27 August. Gothenburg, Sweden.