Multivariate Analysis and Classification of 146 Odor Character Descriptors

Smells can be described by assigning the words that come to mind when sniffing an odorous material. A great number of terms applied, but not all them are independent, and it is possible establish groups often applied together describing a smell. Such classification olfactory descriptors scientific interest in order better understand dimensionality structure human perception space. For this purpose, compilations profiles contain valuable information may lead certain consensus odor classification. One most comprehensive databases Dravnieks’ Atlas, which contains quantitative for 160 samples. each one, large panel rated applicability 146 character on numeric scale. By applying principal component analysis 105 were reorganized 24 classes, 33 attributes considered as odors intermediate two or three categories. The similarities between classes studied means further multivariate based latent variables, provides about salient dimensions Consistent with other reported statistical analyses databases, perceptual space multidimensional 20–30 dimensions, continuum spectrum rather than segmented Attempts classify restricted appear inappropriate. Instead, categories related proposed here, regarding rest smells, assuming independent follow hierarchy according particular underlying dimensions.