The classification of Phyllanthus niruri Linn. according to location by infrared spectroscopy

This preliminary work investigates whether Fourier transform infrared spectroscopy (FT-IR), in combination with multivariate analysis can be used to distinguish Phyllanthus niruri Linn. samples from different locations. A FT-IR spectrometer used in this work is equipped with a transmittance accessory. All samples were scanned in the mid-infrared range of 400–4000 cm−1 but only the region of 400–2000 cm−1 that contained the most spectral information was analyzed with multivariate techniques of principal component analysis (PCA), genetic algorithm (GA), linear discriminant analysis (LDA) and SIMCA. PCA was used to reduce the spectra to three principal components, which explained 96.2% of the whole variance and a plot of principal components two and three showed groupings of the samples according to origin. Secondly, the spectra compressed with PCA were used to develop a LDA. The GA applied for 50 or 100 generations used concept of maximizing the ratio of between-groups variance to within-groups variance and identified the most discriminating variable for subsequent LDA. The SIMCA technique consisted of constructing an enclosure for each location using separate principal component models and using these models to classify the whole data. The best discriminatory approach was the technique using GA with LDA, where using just six wavenumber variables obtained from application of GA for 100 generations, gave 100% correct prediction for both originally grouped and cross-validated data. However, both PCA–LDA and SIMCA also gave very good classification. Basically, FT-IR analysis is rapid and has the power to discriminate samples in terms of geographical origin with the aid of multivariate analyses.