Abstract
A variation projection method is proposed to 1) remove the redundant variables in spectrum, and improve the measurement accuracy in partial least squares (PLS) prediction. The algorithm was tested on urine Raman spectra taken on a liquid core optical fiber Raman spectroscopy system. The variables were effectively removed without affecting the final prediction accuracy.
© 2012 Optical Society of America
PDF ArticleMore Like This
Anthony J. Durkin, L. Stephanie Matchette, and Marwood N. Ediger
BTuC3 Biomedical Optical Spectroscopy and Diagnostics (BIOMED) 1998
Sandeep Kumar, Shovan K. Majumder, Pradeep K. Gupta, Harshul Patidar, and Chandrahas V. Kulkarni
T3A.65 International Conference on Fibre Optics and Photonics (Photonics) 2014
Sabbir Liakat, Kevin A. Bors, Tzu-Yung Huang, Anna P. M. Michel, Eric Zanghi, and Claire F. Gmachl
FTu2D.3 Fourier Transform Spectroscopy (FTS) 2013