Application of genetic algorithm-PLS for feature selection in spectral data sets

After suitable modifications, genetic algorithms can be a useful tool in th e problem of wavelength selection in the case of a multivariate calibration performed by PLS. Unlike what happens with the majority of feature selecti on methods applied to spectral data, the variables selected by the algorith m often correspond to well-defined and characteristic spectral regions inst ead of being single variables scattered throughout the spectrum. This leads to a model having a better predictive ability than the full-spectrum model ; furthermore, the analysis of the selected regions can be a valuable help in understanding which are the relevant parts of the spectra. After the pre sentation of the algorithm, several real cases are shown. Copyright (C) 200 0 John Wiley & Sons, Ltd.