Determination of the Number of Components in Spectroscopy from the Multilinear Model Fitting

  • 발행 : 1999.08.01

초록

Biological specimens contain several components and multilinear models are very useful in analyzing these data. After fitting the model the number of components are determined by the change of mean squared error however this method is quite rule of thumb. in this paper we suggest a measure to decide the number of components based on the relative change of to mean squared error. Simulations are done and applications to real data set are given as illustrations.

키워드

참고문헌

  1. Communications in Statistics - Simulations and Computations v.25 Reference values for Cook's distance Kim, C.;Storer, B.E.
  2. Journal of Science v.53 Trilinear analysis of fluorescene spectra of chromatographic eluents Lee, C.H.
  3. Korean Biochemistry journal v.24 Trilinear analysis for the resolution of overlapping fluorescene spectra Lee, C.H.;Kim, K.;Ross, R.T.
  4. Communications in Statistics - Simulations and Computations v.26 An algorithm for weightes bilinear regression Lee, C.H.;Ezzeddine, B.M.;Kim, C.;Ross, R.T.
  5. Korean Biochemistry journal v.24 Trilinear analysis for the resolution of overlapping flurescene spectra Lee, C.H.;Kim. K.;Ross, R.T.
  6. Statistical Science v.7 Multi Models : Applications in spectroscopy Leugans, S.;Ross, R.T.
  7. Density Estimation for Statistics and Data Analysis Silverman, B.W.