ABSTRACT
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
Chemometrics is often defined as the application of statistics and mathematics to the analysis of chemical data. Without arguing the sufficiency of this definition, it is safe to say that the application of multivariate statistical and mathematical spectral analysis methods to near-infrared (NIR) data provides an intriguing set of advantages absent in univariate analysis of NIR data. Foremost of these advantages are the abilities to preprocess NIR spectra for removal of complex background signals, perform multianalyte calibration and calibration in the presence of multiple changing chemical
DBURNS: “7393_c010” — 2007/7/23 — 21:53 — page 208 — #2
that deviate from the bulk of the calibration set.
