Linear and Nonlinear Multivariate Classification of Iranian Bottled Mineral Waters According to Their Elemental Content Determined by ICP-OES

The combinations of inductively coupled plasma-optical emission spectrometry (ICP-OES) and three classification algorithms, i.e., partial least squares discriminant analysis (PLS-DA), least squares support vector machine (LS-SVM) and soft independent modeling of class analogies (SIMCA), for discriminating different brands of Iranian bottled mineral waters, were explored. ICP-OES was used for the determination of Li, Na, K, Ca, Mg, Sr, Ba, B, Si and Zn in bottled mineral waters (150 samples) from 30 brands. Hierarchical cluster analysis (HCA) and principal component analysis (PCA) showed differences in water samples according to the mineral composition. 120 samples (4 for each brand) were selected randomly for the calibration set, and 30 samples (1 for each brand) for the prediction set. PLS-DA, LS-SVM and SIMCA were implemented for calibration models. The results suggest that ICP-OES combined with PLS-DA, LS-SVM and SIMCA models had the capability to discriminate the different brands of mineral waters with high accuracy. The model can resolve the tap water samples from classified mineral waters accordingly.