Premium
Partial linear fit: A new NMR spectroscopy preprocessing tool for pattern recognition applications
Author(s) -
Vogels J. T. W. E.,
Tas A. C.,
Venekamp J.,
van der Greef J.
Publication year - 1996
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/(sici)1099-128x(199609)10:5/6<425::aid-cem442>3.0.co;2-s
Subject(s) - preprocessor , multivariate statistics , multivariate analysis , stability (learning theory) , data pre processing , computer science , partial least squares regression , pattern recognition (psychology) , resolution (logic) , artificial intelligence , biological system , data mining , mathematics , machine learning , biology
NMR spectroscopy is increasingly used in combination with multivariate analysis applications. Especially in the analysis of food products and the study of natural processes it has proved its usefulness. The samples used in these evaluations are however, often difficult to control. ‘Positional’ shifts of peaks due to differences in pH and other physico‐chemical interactions are quite common. A reduction of the resolution of the spectra is generally sufficient to correct for these effects. This approach is, however, not possible if the fine structure in the data is important in the analysis. A solution to this problem is to use the partial linear fit (PLF) algorithm described here. Using PLF preprocessing the fine structure in the data is utilized to correct for any ‘positional’ variances, which results in a significant improvement in the classification ability and a greater stability of the multivariate data analysis. © 1996 by John Wiley & Sons, Ltd.