T
he combination of
1H NMR fingerprinting wit
h multivariate analysis provides an original approac
h to study t
he profile of olive oil in relation to its geograp
hical origin and processing. T
he present work aims at illustrating t
he relevance of
1H NMR fingerprints for assessing t
he geograp
hical origin and t
he year of production for olive oils from various Mediterranean areas. Multivariate (c
hemometric) tec
hniques are able to filter out t
he most relevant information from a spectrum, e.g. for a classification. Principal component analysis (PCA) was carried out on t
he
![](<font color=)
http://www.sciencedirect.com/scidirimg/entities/223c.gif" alt="not, vert, similar" border=0>12,000 variables (c
hemical s
hifts) and four data sets were defined prior to PCA. Linear discriminant analysis (LDA) of t
he first 50 PC's was applied for classification of olive oil samples (97 or 91) according to t
he geograp
hic origin and year of production. T
he data analysis
has been carried out wit
h and wit
hout outliers, as well. Variable selection for LDA was ac
hieved using: (i) t
he best five variables and (ii) an interactive forward stepwise manner. Using LDA on t
he external validation sets t
he correct classification varied between 47 and 75%(random selection), and between 35 and 92%(Kennard–Stone selection (KS)) depending on geograp
hic origin (country) and production years. A similar success rate could be ac
hieved using partial least squares discriminant analysis (PLS DA). T
he success rate can be considerably improved by using probabilistic neural networks (PNN). Correct classification by PNN varied between 58 and 100%on t
he external validation sets. Ot
her c
hemometric tec
hniques, suc
h as multiple linear regression, or generalized pair-wise correlation, did not give better results.