A method is described whereby the wealth of information present in the Fourier transform infrared spectrum of a material may be extracted and used to estimate useful properties of the material. Factor analysis is used to condense the data, after which the use of multiple linear regression allows correlations to be established for a calibration set, from which properties of unknown samples can be estimated.
Peter M. Fredericks, James B. Lee, Paul R. Osborn, and Dom A. J. Swinkels, "Materials Characterization Using Factor Analysis of FT-IR Spectra. Part 2: Mathematical and Statistical Considerations," Appl. Spectrosc. 39, 311-316 (1985)