Cluster analysis applied to the exploratory analysis of commercial spanish olive oils by means of excitation-emission fluorescence spectroscopy.

Olive oil fluorescence is related to oil composition. Here it is shown that the natural clustering of different types of commercial Spanish olive oils depends on their fluorescence excitation-emission matrices (EEMs). Fifty-six commercial samples of olive oil (29 virgin olive oils, 20 pure olive oils, and 7 olive-pomace oils) were used. The clustering method was hierarchical agglomerative clustering using the Euclidean distance as a similarity measure and the average linkage. Two spectral ranges were considered (which either contained the fluorescence peak of the chlorophylls or did not), and various methods for preprocessing the fluorescence spectra were compared. The oils were clearly distinguished using the unfolded EEMs measured between lambda(ex) = 300-400 nm and lambda(em) = 400-600 nm. The optimal preprocessing was normalization of the unfolded spectra followed by column autoscaling. Also shown are the advantages of using second-order data (EEMs) instead of first-order data (a single fluorescence spectrum) for each sample.