Chemometrics: Views and Propositions

Science is hungry for new tools, and, when one is discovered, history often repeats itself. The interest and support rise rapidly at first. Then, if the use of the tool is not somewhat conservative, the tool is often misused. The reasons behind the misuse are complex and numerous but include pressures to publish in quantity and the desire to execute application “firsts”. Misuse of a tool often leads to disenchantment with it, andl, coupled with other factors (finite funds, etc.), progress will decline and then oscillate for a time. Eventually, if the tool is really useful, real progress can proceed. Pattern recognition is a collection of problem solving methods that, in my opinion, collectively comprise a tool that can be applied to a wide range of information extraction application which includes several in the field of chemistry. I t is, perhaps, no better or worse than several other collections of data analysis methods (standard statistics, optimization, etc.) and, like these many other collections, can be extremely useful in doing something that is sorely needed in chemistry: extracting useful chemical information from large amounts of measurements or raw data. I, and others, prefer to aLssemble any and all methods that can be used to extract useful chemical information from raw data under the general title of “chemometrics”. With this assemblance in mind, I now present some of my views on the present and future of chemometrics (emphasis on pattern recognition).