SCREAM (Shifted Covariates REgression Analysis for Multi-way data) combines PARAFAC2 and principal covariates regression (PCovR) for multi-way regression. Overcomes inaccuracy from shifts or peak shape changes in one mode common in hyphenated chromatography. Combines PARAFAC2 decomposition with PCovR-like regression coefficients, performing as well or better than N-PLS on real and simulated datasets.
Federico Marini [1]Rasmus Bro [2]
[1]Department of Chemistry, Sapienza University of Rome, Italy;[2]Department of Food Science, University of Copenhagen, Denmark