Elastic Net Grouping Variable Selection Combined with Partial Least Squares Regression (EN-PLSR) for the Analysis of Strongly Multi-collinear Spectroscopic Data

• DOI: 10.1366/10-06069
• Publication Date: 2011-03-14T17:22:58Z
• Authors: Guang-Hui Fu; Qing-Song Xu; Hong-Dong Li; Dong-Sheng Cao; Yi-Zeng Liang
• Keywords: Elastic net regularization; Least-squares function approximation

In this paper a novel wavelength region selection algorithm, called elastic net grouping variable combined with partial least squares regression (EN-PLSR), is proposed for multi-component spectral data analysis. The EN-PLSR algorithm can automatically select successive strongly correlated prediction groups related to the response using two steps. First, portion of predictors are selected and divided into subgroups by means effect estimation. Then, recursive leave-one-group-out strategy employed further shrink in terms root mean square error cross-validation (RMSECV) criterion. performance real near-infrared (NIR) spectroscopic sets shows that competitive full-spectrum PLS moving window (MWPLS) methods it suitable use data.