Chemometrics, present and future success

How successful has chemometrics been? The answer depends, of course, on how success is defined and measured. A brief discussion of this subject is given, with reference to academic and industrial research and other application areas, notably industrial development and production. The areas where chemometrics has been most successful according to all measures are the following: (1) multivariate calibration, (2) structure—(re)activity modelling, (3) pattern recognition, classification, and discriminant analysis, and (4) multivariate process modelling and monitoring. Possible reasons are ventilated for these seemingly disparate success stories, together with some reflections of what remains to be done in these areas, and why success is slower in other areas. To continue the successful development of chemometrics, the most important is, in our opinion, that we continue to see ourselves primarily as chemical problem solvers—and only when needed, as developers of new methodology. To illustrate the chemistry driven development of chemometrics, we shall describe some recent work in multivariate modelling and analysis, applied in the areas of structure–activity relationships (peptides, proteins, RNA, DNA, hemes), modelling of batch processes and complicated kinetics, wavelet data compression, and orthogonal preprocessing of spectral data (NIR) for multivariate calibration.