Classical vs inverse model for calibration
Spectrophotometers allow the determination of analyte concentration via an absorbance measurement (or Optical Density). This requires
a preliminary calibration phase (calibration) via modeling of the Absorbance vs. Concentration response. This modeling is done classically
by application of the Beer-Lambert physical model with the absorbance in Y and the concentration in X. Even if this "classic" model is very relevant,
you can also do a so-called "inverse" calibration, with the concentration in Y and the absorbance in X. This model is a priori counter-intuitive because it is the
concentration variation which causes the absorbance variation and the error on the preparation of the concentration standards can be finely controlled.
The results are generally almost identical for the 2 approaches, but the inverse model can present certain advantages which will be presented.
It is moreover an inverse modeling that is used systematically for multivariate models, where all of the spectral information is
used for prediction (several wavelengths).
