Sparse N-way partial least squares with R package sNPLS

Abstract We introduce the R package sNPLS that performs N -way partial least squares ( N -PLS) regression and Sparse (L1-penalized) N -PLS regression in three-way arrays. N -PLS regression is superior to other methods for three-way data based in unfolding, thanks to a better stabilization of the decomposition. This provides better interpretability and improves predictions. The sparse version also adds variable selection through L1 penalization. The sparse version of N -PLS is able to provide lower prediction errors and to further improve interpretability and usability of the N -PLS results. After a short introduction to both methods, the different functions of the package are presented by displaying their use in simulated and a real dataset.