ON THE USE OF CONDITIONAL MAXIMIZATION IN CHEMOMETRICS

The purpose of this short communication is to illustrate the use of co nditional maximization (CM) in chemometric applications. The CM algori thm is useful in reducing the computational complexity when a high-dim ensional and complicated maximization problem arises from fitting chem ometric models. It can also be efficiently combined with the expectati on-maximization (EM) algorithm for handling incomplete data, a problem that sometimes arises when only a part of the intended data can be co llected. Three models from fluorescence spectroscopy are used for illu stration.