Assuring specificity for a multivariate near-infrared (NIR) calibration: The example of the Chambersburg Shoot-out 2002 data set

A method for demonstrating specificity has been developed for NIR calibrations involving the use of partial least squares (PLS) regressions. The method is demonstrated with near-infrared transmittance data on pharmaceutical tablets. A regression was performed with PLS and the first principal component spectrum is shown to match the spectrum of the active ingredient as well as a spectrum extracted from the original file. A good match demonstrates that the calibration is specific for the active component. The effects on the match for different pretreatments were evaluated.