The successive projections algorithm

The successive projections algorithm (SPA) is a variable-selection technique that has attracted increasing interest in the analytical-chemistry community in the past 10 years. The present review presents the basic features of SPA for Multiple Linear Regression (MLR) and Linear Discriminant Analysis (LDA) and reports some variants that have been proposed for sample selection, calibration transfer and Quantitative Structure-Activity Relationship (QSAR) and Quantitative Structure-Property Relationship (QSPR) studies. We also discuss computational and pre-processing issues. By way of illustration we present two case studies involving near-infrared determination of protein in wheat and voltammetric classification of vegetable oils. The code employed in this article is freely available from us upon request. (c) 2012 Elsevier Ltd. All rights reserved.