Chemometrics: the issues of measurement and modelling

In this paper, two spectral data sets have been used to illustrate the impo rtance of maintaining chemical information whilst generating predictive mul tivariate calibration models. The first data set is based on 26 duplicate U V/VIS spectra for four meal ions (Fe, Ni, Co, Cu) present at varying concen trations in aqueous solution. Spectra were collected across the range 180-8 00 nm at a resolution of 3.5 nm generating 211 data points for each sample. Calibration was carried out using multiple linear regression (MLR) and a K -matrix approach to demonstrate the advantages the latter method has in des cribing real spectral features. In addition, the limitation of MLR in accom modating noise and spectral overlap in the data is also illustrated. The se cond data set based on NIR spectroscopy, was generated using a four-level 2 factor Factorial design strategy and consisted of two additives present at a range of concentrations in an aqueous caustic system, with the spectra b eing collected over the range 10,000-3000 cm(-1). Whilst a conventional par tial least squares (PLS) model was applied to the data, it was through the use of variable selection (VS) prior to PLS and the application of weighted ridge regression (WRR) techniques that the need to develop chemometric met hodology which intuitively reflected chemical information has been demonstr ated. The results will also illustrate how a poorly designed experimental d esign protocol and missing data can limit the performance of the calibratio n models generated. The aims of this paper are not to prescribe ideal calib ration methodology but rather to demonstrate the relevance of selecting mul tivariate calibration methodology that relates more to the chem rather than just the metrics in chemometrics. (C) 1999 Elsevier Science B.V. All right s reserved.