Discriminant Partial Least Square on Compositional Data: a Comparison with the Log-Contrast Principal Component Analysis


Abstract


En
Discriminant Partial Least Squares for Compositional data (DPLS-CO) was recently proposed by Gallo (2008). The aim of this paper is to show that DPLS-CO is a better dimensionality reduction technique than the LogContrats Principal Component Analysis (LCPCA) for dimensional reduction aimed at discrimination when a compositional training dataset is available.

DOI Code: §

Keywords: Compositional observation; Dimension reduction; Linear discrimination

Full Text: PDF
کاغذ a4

Creative Commons License
This work is licensed under a Creative Commons Attribuzione - Non commerciale - Non opere derivate 3.0 Italia License.