BUCHI NIRCal - Manual (Page 198)

NIRCal 5.5 Software Manual
198 NIRCal 5.5 Manual, Version A
3.18.76 Score Adjacency
Description Distance in score space between each spectra against all others.
Use The neighborhood relationships between the spectra for all PCs at once.
For special interest with the Cluster method.
Method PCR / PLS / Cluster (CLU) / SIMCA
Matrices ID 40
Tip Look at it in a 2D-Plot in top view to see hidden effects.
Details Secondary PCs take an effect.
Related Topic Scores , Leverage
3.18.77 SEP Generalized Cross Validation
Description The determination of the SEP (Standard Error of Prediction) with a Cross
Validation (CV) is very time consuming.
The SEP value determined with the GCV (Generalized Cross Validation)
process is from theoretically side at least equally good as the SEP of a
conventional CV process.
Use Select the number of PC, where the SEP Generalized Cross Validation has
a minimum.
Method MLR / PCR / PLS
Matrices ID 39
Tip While SEP Generalized Cross Validation does not always have a minimum, it
is not always optimal to use this for the secondary PCs selection.
Details See also: Gene H. Golub, Michael Heath, and grace Wahba. Generalized
crossvalidation as a method for choosing a good ridge parameter.
Technometrics, 21(2):215-223,1979.
Related Topic V-Set SEE (SEP)
Formula:
where:

n : number of calibration spectra

a : number of secondary PCs
3.18.78 Spectra
Description Contains the spectra index number.
Use For 1D-scatter plots.
Method PCR / PLS / Cluster (CLU) / SIMCA / MLR
Matrices ID 80
Tip
Details Spectra is a column vector.
Related Topic
Description
Contains the spectra index number.
Use For 1D-scatter plots.
Method PCR / PLS / Cluster (CLU) / SIMCA / MLR
Matrices ID 81
Tip
Details Spectra is a row vector.
Related Topic