Dual-sPLS: a family of Dual Sparse Partial Least Squares regressions for feature selection and prediction with tunable sparsity; evaluation on simulated and near-infrared (NIR) data

TL;DR

Dual-sPLS is a novel family of sparse PLS regressions that balances prediction accuracy and interpretability, utilizing dual norm penalizations for feature selection, demonstrated on simulated and NIR data.

Contribution

It introduces Dual-sPLS, a generalized sparse PLS method that incorporates dual norm penalizations and a shrinking ratio for tunable sparsity, enhancing feature selection and prediction.

Findings

Outperforms similar methods on simulated data

Effective feature selection in high-dimensional chemometric data

Provides an open-source R package for implementation

Abstract

Relating a set of variables X to a response y is crucial in chemometrics. A quantitative prediction objective can be enriched by qualitative data interpretation, for instance by locating the most influential features. When high-dimensional problems arise, dimension reduction techniques can be used. Most notable are projections (e.g. Partial Least Squares or PLS ) or variable selections (e.g. lasso). Sparse partial least squares combine both strategies, by blending variable selection into PLS. The variant presented in this paper, Dual-sPLS, generalizes the classical PLS1 algorithm. It provides balance between accurate prediction and efficient interpretation. It is based on penalizations inspired by classical regression methods (lasso, group lasso, least squares, ridge) and uses the dual norm notion. The resulting sparsity is enforced by an intuitive shrinking ratio parameter. Dual-sPLS favorably compares to similar regression methods, on simulated and real chemical data. Code is provided as an open-source package in R: \url{https://CRAN.R-project.org/package=dual.spls}.