Improvement of PLS model transferability by robust wavelength selection

Abstract Frequently, a calibration model is adapted after being transferred to another instrument by, e.g., direct standardization (DS) or piecewise direct standardization (PDS). For this, a subset from the calibration set should be measured on both instruments. Usually, however, the calibration samples cannot be measured on both instruments. Another approach is to make the model robust with respect to transfer to another instrument during the development of the model, by data preprocessing. In this paper, the robustness of the calibration model is enhanced by using variable selection as data preprocessing. In the case under consideration, variable selection consists in the calculation of a calibration model with a subset of the original wavelengths (of spectroscopic data) that retains its predictive ability when it is transferred to another instrument. Both approaches (variable selection and (P)DS) are applied to the transferability of a PLS model which determines the water content in tablets. To this end, 140 tablets were measured on 2 near-infrared (NIR) reflectance instruments. It has been found that variable selection by simulated annealing (SA) enhances the model's robustness with respect to model transfer and also improves its predictive ability.

[1]  Bruce R. Kowalski,et al.  Propagation of measurement errors for the validation of predictions obtained by principal component regression and partial least squares , 1997 .

[2]  Paul Geladi,et al.  Interactive variable selection (IVS) for PLS. Part 1: Theory and algorithms , 1994 .

[3]  B. Kowalski,et al.  Multivariate instrument standardization , 1991 .

[4]  John H. Kalivas,et al.  Further investigation on a comparative study of simulated annealing and genetic algorithm for wavelength selection , 1995 .

[5]  Luis A. Sarabia,et al.  Genetic-algorithm-based wavelength selection in multicomponent spectrometric determinations by PLS: application on indomethacin and acemethacin mixture , 1997 .

[6]  Desire L. Massart,et al.  Improvement of the piecewise direct standardisation procedure for the transfer of NIR spectra for multivariate calibration , 1996 .

[7]  Desire L. Massart,et al.  Investigation of sources of variance which contribute to NIR-spectroscopic measurement of pharmaceutical formulations , 1997 .

[8]  Hideyuki Abe,et al.  Analysis of Hydrogen Bonding Related to Water in Foods , 1995 .

[9]  Desire L. Massart,et al.  Random correlation in variable selection for multivariate calibration with a genetic algorithm , 1996 .

[10]  John H. Kalivas,et al.  Adaption of simulated annealing to chemical optimization problems , 1995 .

[11]  B. Skagerberg,et al.  Predictive ability of regression models. Part I: Standard deviation of prediction errors (SDEP) , 1992 .

[12]  D. Massart,et al.  Elimination of uninformative variables for multivariate calibration. , 1996, Analytical chemistry.

[13]  H. Swierenga,et al.  Comparison of Two Different Approaches toward Model Transferability in NIR Spectroscopy , 1998 .

[14]  John H. Kalivas,et al.  Global optimization by simulated annealing with wavelength selection for ultraviolet-visible spectrophotometry , 1989 .

[15]  Jr Jerry Workman.  The Properties of Fibre Optics , 1995 .

[16]  Onno E. de Noord,et al.  Multivariate calibration standardization , 1994 .

[17]  Bruce R. Kowalski,et al.  Additive Background Correction in Multivariate Instrument Standardization , 1995 .

[18]  Riccardo Leardi,et al.  Genetic Algorithms as a Tool for Wavelength Selection in Multivariate Calibration , 1995 .

[19]  R. Wehrens,et al.  Bootstrapping principal component regression models , 1997 .

[20]  M A Arnold,et al.  Genetic algorithm-based method for selecting wavelengths and model size for use with partial least-squares regression: application to near-infrared spectroscopy. , 1996, Analytical chemistry.

[21]  C. B. Lucasius,et al.  Genetic algorithms in wavelength selection: a comparative study , 1994 .

[22]  F. Navarro-Villoslada,et al.  Selection of calibration mixtures and wavelengths for different multivariate calibration methods , 1995 .

[23]  Charles E. Miller,et al.  Sources of Non-Linearity in near Infrared Methods , 1993 .

[24]  John H. Kalivas,et al.  Wavelength Selection Characterization for NIR Spectra , 1997 .

[25]  S. D. Jong.  SIMPLS: an alternative approach to partial least squares regression , 1993 .

[26]  G. Cruciani,et al.  Predictive ability of regression models. Part II: Selection of the best predictive PLS model , 1992 .

[27]  Desire L. Massart,et al.  Comparison of multivariate methods based on latent vectors and methods based on wavelength selection for the analysis of near-infrared spectroscopic data , 1995 .

[28]  John H. Kalivas,et al.  Simulated‐annealing‐based optimization algorithms: Fundamentals and wavelength selection applications , 1995 .