Abstract
A new MATLAB graphical interface toolbox for implementing third-order multivariate calibration methodologies is discussed. Multivariate calibration 3 (MVC3) is a sequel of the already described first-order (MVC1) and second-order (MVC2) toolboxes. MVC3 accepts a variety of ASCII data for input, depending on whether the third-order data are vectorized or matricized. If required, data for sample sets are arranged into four-way arrays for processing with several quadrilinear and non-quadrilinear algorithms. Quadrilinear decomposition techniques and latent structured models based on partial least-squares regression and residual trilinearization are included in the software. Appropriate working sensor regions in the three data dimensions 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.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
