Figures of Merit in Multiway Calibration

Abstract

In this chapter, the estimation of relevant analytical figures of merit is reviewed, from univariate to multivariate and multi-way calibration: sensitivity, limit of detection and other analytical parameters. After a brief discussion on data properties, models and algorithms, different sensitivity expressions are presented, based on the concept of net analyte signal as a function of concentration, and those derived from uncertainty propagation, as applied to univariate, first-order and multiway (higher-order) calibrations. Relevant algorithms for which figures of merit are detailed are those based on multilinear decomposition, multivariate curve resolution-alternating least-squares and partial least-squares with residual multilinearization. Comparison is made of the values furnished by the discussed expressions in going from lower to higher-order data.