Multivariate Data Processing

In some cases, many different spectra (or chromatograms) of the same object are available. For inhomogenous objects, for example, several samples of different constitution can be taken. This allows to apply multivariate data processing techniques. When the signals of the compounds in the sample are specific and linearly additive, the number of compounds which contribute to the signal, can be determined by a Principal Components Analysis (PCA) (see Sect. 3.2.1). Without knowing the identity of all compounds, which are present and without knowing their spectra, a calibration by partical least squares (PLC) allows to quantify the compounds of interest. 

0 S PflR riON 0 HPLCAJVVIS-pC MS 0 EXCIT. -EMISSION 9 SPECTROMETRY 

0 NUMBER OF COMPOUNDS 0 CRfWIDRTE COMPOUNDS 0 SPECTRR OF CRND. COMP. 

0 MORE (DIFFERENT)0 RRFR 0 GSflW 9 CRFR 9 TTFR 0 MCR (f R) 

0 NUMBER OF COMPOUNDS 9 CRNDIDRTE COMPOUNDS 0 SPECTRR or CRND. COMP. 

---

Above mentioned examples clearly show that if multivariate data processing methods are applicable, analytical information can be derived with a minimal amount of pre-information and a foreseeing of a maximum of problems. When the sampled object is homogenous, multivariate methods are only applicable when the analytical method itself produces multivariate signals. This is the case when several signals (e.g. spectra) are obtained for the sample as a function of another variable (e.g. time, excitation wavelength). For e mple in GC-MS, a mass spectrum is m sured of the eluents every. 1 a 1 second. In excitation-emission spectroscopy, spectra are measured at several excitation-wavelengths. The potentials of the application of multivariate... 

Multivariate data analysis has been developed as an independent field of statistics and has numerous practical and theoretical applications. This explains the existence of a variety of printed materials and of different computer programs available for multivariate data processing [71a], Only some aspects of multivariate data analysis with application to the processing of Py-MS data will be discussed here. 

Most of the procedures using multivariate data processing were used for qualitative purposes such as classification and identification of microorganisms, differentiation of capsular polysaccharides of different bacteria, or fossil biomaterials, etc. Quantitation was also possible in identifying adulteration of several foods [86]. 

Figure 13. Multivariate data processing techniques employed by electronic noses. See reference [2] for definition of acronyms.

Multivariate Data Processing in Spectrophotometric Analysis of Complex... 

Multivariate Data Processing in Specirophotometric Analysis of Complex Chemical Systems... 

R. (2009) Characterization of interfacial chemistry of adhesive/dentin bond using FTIR chemical imaging with univariate and multivariate data processing. /. Biomed. Mater. Res. A, 91 (1), 251-262. 

---