LOCAL CENTERING IN MULTIVARIATE CALIBRATION

Multivariate calibration models are constructed using measured respons es of variables (e.g. spectra) on a set of calibration samples and val ues of a quantity of interest (e.g. concentration) measured by a refer ence method. The goal is to replace the reference method. Traditionall y the calibration data are mean centered, which insures minimum averag e prediction error (for a prediction set having the same distribution as the calibration set). An alternative to this preliminary data treat ment is presented. Instead of using the entire calibration set for cen tering, a subset of samples from the calibration set that are closest to the unknown is selected for centering. This preliminary data treatm ent reduces reliance on regression. Thus it is expected to perform wel l in cases where model errors are dominating or extrapolation occurs. The method is tested on data from near-infrared reflectance and infrar ed emission spectroscopy, showing that an average improvement of 20% i n prediction accuracy is achievable. This method is fundamentally diff erent from locally weighted regression because it uses the entire cali bration set for the regression step.