Spurious principal components in the set of spectra subjected to disturbances: I. Presentation of the problem

作者: R Siuda , G Balcerowska , D Aberdam

DOI: 10.1016/S0169-7439(97)00086-5

关键词:

摘要: Abstract It has been shown that a kind of imperfections which can be met in spectrometers, or small shifts the recorded lines, influence properties spectra sets such way number principal components (PCs) necessary to describe non-noisy part total variance set increases compared when disturbances are absent. The additional (spurious) PCs depends on magnitude and signal-to-noise ratio (SNR). mechanism generates spurious is discussed, criterion enabling an evaluation effect becomes important, given. Numerical simulations enabled inference empirical relationship between PCs, imperfections, SNR. Finally, procedure proposed check if spectrometer free discussed.

参考文章(4)
M. J. G. Wenham, I. R. Barkshire, M. Prutton, R. H. Roberts, D. K. Wilkinson, An examination of the use of principal component analysis and target factor analysis for the determination of Auger depth profiles in two magnetic multilayer metal systems—Cu/Co and Co/Pt Surface and Interface Analysis. ,vol. 23, pp. 858- 872 ,(1995) , 10.1002/SIA.740231303
Bart J. Kooi, Marcel A. J. Somers, Factor analysis of Fe M2,3 VV spectra of oxidizing iron: The role of the background in the emergence of ghost components Surface and Interface Analysis. ,vol. 21, pp. 501- 513 ,(1994) , 10.1002/SIA.740210802
Larry G. Blackwood, Factor Analysis in Chemistry (2nd Ed.) Technometrics. ,vol. 36, pp. 115- 115 ,(1994) , 10.1080/00401706.1994.10485411
Ian T. Jolliffe, Principal Component Analysis ,(1986)