A synthetic review of some recent extensions of ComDim

Abstract

The simultaneous analysis of several data matrices related to the same set of samples can be done with the use of multiblock methods. Common components and specific weights analysis (CCSWA), also called ComDim, is one method enabling the analysis of such data. CCSWA, which is a multiblock version of principal components analysis (PCA), is particularly interesting as it can easily be modified and transformed, either by replacing its PCA‐based data decomposition by another multivariate method, such as independent components analysis or partial least‐squares regression, for example, or by segmenting the data matrices before multiblock analysis, or both. These modifications aim at improving the models and their interpretation. This article will give a concise description of ComDim and then a presentation of several recent extensions of the method, showing its usefulness as both a supervised and a non‐supervised method.

Improvement of Qualitative Analyses of Aliphatic Alcohols Using Direct Catalytic Fuel Cell and Chemometric Analysis Format

“…Briefly [10,11,[16][17][18], we consider a set of M data matrixes Xm m = 1…M sharing the same number of samples n, but possibly with a different number of variables. Each matrix is assumed to be mean-centered and further scaled to unit Frobenius' norm, so to make the variance of the different blocks comparable.…”

“…ComDim can be considered a generalization of principal component analysis (PCA) [12][13][14][15] for cases where multiple data matrices are collected to describe the same set of samples. It was first applied in sensory analysis and has gained some popularity in both sensorimetry and chemometrics [10,11,[16][17][18].…”

“…Briefly [10,11,[16][17][18], we consider a set of M data matrixes Xm m = 1. . .M sharing the same number of samples n, but possibly with a different number of variables.…”