LAUSR

Abstract An upgrade is presented of a MATLAB graphical user interface toolbox for implementing third-order/four-way multivariate calibration models. The new Multivariate Calibration 3 (MVC3_GUI) incorporates new models and features that make it a very versatile tool for four-way data processing. In addition to the quadrilinear decomposition (QLD) and latent structure models based on partial least-squares regression and residual trilinearization, included in the earlier software version, non-QLD models are now available. The latter include extended multivariate curve resolution-alternating least-squares (MCR-ALS), augmented parallel factor analysis (Augmented PARAFAC) and PARAFAC2. The software is presented as both a set of MATLAB codes and as a standalone program. MVC3_GUI accepts a variety of ASCII data for input. Appropriate working sensor regions in the different data modes can be selected. Model development and its subsequent application to unknown samples are straightforward from the interface. Prediction results are provided along with analytical figures of merit and standard concentration errors, as calculated by modern concepts of uncertainty propagation. Different examples of use of this updated interface are given in this work.