Mixture ratios of the standard and verification samples are indicated in Tables 1 and 2. The absorption spectra for each brand of mineral water, measured using a quartz cell with a 10 mm optical path length and ion-exchange water as a reference, are shown in Fig. The UV absorption spectra of nine standard samples and three verification samples were measured using a UV-3600 UV-VIS-NIR spectrophotometer. Then samples with known mixture ratios were prepared for verification of the calibration model. First a calibration model was created using standard samples prepared by varying the mixture ratio.
In this example, three commercial brands of bottled mineral water – A, B, and C, were mixed in various proportions, then multiple regression was used to determine the mixture ratio of each sample. Multiple regression allows determining the quantity of each component (or target components) in a sample containing a mixture of components. In this article, multiple regression was selected as the quantitative method and principal component analysis (PCA) and cluster analysis were selected as the classification methods to illustrate multivariate analysis.Ģ Simultaneous Quantitation of Mineral Water Mixture Samples Using Multiple Regression 1 shows an overview of multivariate analysis. Using such techniques allows classifying multiple samples based on similarities to understand their characteristics or help identify where they were produced. Classification methods include principal component analysis (PCA), cluster analysis, discriminant analysis, and factor analysis. However, PCR and PLS are more susceptible to the effects of noise (data variability) than multiple regression. Multiple regression requires that the analyst select specific data (descriptive variables), but that is not required using PCR and PLS techniques, which allow using all data. Using such techniques allows determining the quantity of each component (or target components) in a sample containing a mixture of components. Quantitative methods include multiple regression (multiple linear regression (MLR) or inverse least squares (ILS)), principal component regression (PCR), and partial least squares (PLS).
0 kommentar(er)
