Chemometric analyses with self organising feature maps: A worked example of the analysis of cosmetics using Raman spectroscopy

Abstract

This chapter explains the dispersive Raman spectroscopy that was used to gain very high dimensional chemical fingerprints from perfumes and lipsticks. Spectral acquisition was reproducible, accurate, non-invasive, and very rapid; the typical analysis time per sample was only 1 minute. To observe the relationship among these cosmetics, based on their spectral fingerprints, it was necessary to reduce the dimensionality of these hyperspectral data by unsupervised feature extraction methods. The neural computational pattern recognition technique of self organizing feature maps (SOMs) was therefore employed and the clusters observed compared with the groups obtained from the more conventional statistical approaches of principal components analysis (PCA) and hierarchical cluster analysis (HCA). All chemometric cluster analyses gave identical results. Very successful exploratory analyses were performed for the SOM analysis of the lipsticks. SOMs were also able unequivocally to classify and to identify all the perfumes analyzed. This chapter demonstrates the potential of dispersive Raman spectroscopy for the non-invasive, non-destructive discrimination of perfumes and lipsticks, and may find application in authenticity testing of perfumes and as a forensic investigative tool for the identification of lipsticks from crime scenes.