Skip to main content
ARS Home » Research » Publications at this Location » Publication #151279
Title: IMPROVEMENT IN UPDATING CHEMOMETRIC MODELS WITH NEAR-INFRARED AND RAMAN DATA: PREDICTION OF PROTEIN AND AMYLOSE

Author
item Himmelsbach, David
item SOHN, MI RYEONG - FAS/OICD
item Barton Ii, Franklin

Submitted to: International Conference on Vibrational Spectroscopy
Publication Type: Abstract Only
Publication Acceptance Date: 8/20/2003
Publication Date: 8/24/2003
Citation: Himmelsbach, D.S., Sohn, M., Barton II, F.E. 2003. Improvement in Updating Chemometric Models with Near-infrared and Raman Data: Prediction of Protein and Amylose. Book of Abstracts, the 2nd International Conference on Vibrational Spectroscopy ICAVS-2. p. 242, PO183.

Interpretive Summary:

Technical Abstract: Habitual problems with chemometric models are their periodic requirement for update and essentially manual determination of the best pre-processing parameters. This is complicated by continual changes in instrument performance. A regime of data standardization and automated parameter determination can overcome most of these problems. As an example, improved chemometric models were produced for near-infrared and Raman spectral data collected over multiple years for the prediction of protein and amylose content in rice flours. First, direct standardization was employed to generate an instrument transfer function to make the first year's data appear as it was collected under the same instrumental conditions as the second year's data. This overcame residual instrumental changes that could not be corrected by typical instrument corrections. The combined data sets were then processed using the automated routines [1], developed in MATLAB (tm) with PLS_Toolbox, to produce root mean squared error (RMSE) of cross-validation calibration vs. smoothing parameters and smoothing parameters vs. latent variables or principal components. This data handling reduced the errors of prediction and permitted efficient updating of calibrations. [1] D.D. Archibald, S.E. Kays, D.S. Himmelsbach and F.E. Barton, II, Applied Spectrosc., 1998 52 (1), 22.