BUCHI NIRCal - Manual (Page 195)

Chemometrics
NIRCal 5.5 Manual, Version A 195
3.18.70 Reconstructed Spectra
Description Reconstructed pretreated spectra with the number of primary PCs.
Use See how good the reconstruction works.
Method PCR / PLS / Cluster (CLU) / SIMCA
Matrices ID 8
Tip It should be as similar to the pretreated spectra as possible.
Details Reconstructed spectra are plotted as the intensity values after the used
pretreatments according the wavelengths after the reconstruction of the PCA
using the primary PCs.
Related Topic Pretreated Spectra Spectra Residuum
Available in the menubar: Graphics / Spectra / Reconstructed.
A reconstructed spectrum is the sum of [mean spectrum and the sum of PCs multiplied with the scores
values].
3.18.71 Residuals
Description Contains the residual value against the PCs for the different spectra.
Use For selection of the primary PCs.
Method PCR / PLS / Cluster (CLU) / SIMCA
Matrices ID 95
Tip Select the smallest number of PCs, where the residuals are similar small and
do not change anymore.
Details Residual is the sum of squares of the residuum spectrum. Residuum
spectrum = pretreated spectrum
– reconstructed spectrum.
Related Topic Spectra Residuum , X-PRESS
3.18.72 Regression Coefficients [1]
Description It shows the regression coefficients in function of the PCs. Each regression
coefficients name has the name of the PC.
Use Help for the optimal secondary PCs selection.
Method PCR / PLS
Matrices ID 79
Tip Select PCs, which have stable regression coefficients (not noisy).
Details [1] means the 1st C-Set property. In general for each quantitative application
only one C-Set property is allowed, so for each property a separate
calibration is necessary (NIRCal could handle more in file-mode, but
NIRWare is conceived with only one property per calibration for quantitative
measurements).
Related Topic Regression Coefficient , CV Regression Coefficients[1]
3.18.73 Regression Coefficients / Property Spectra (CLU)
Description Spectra of the property. This matrices is called Regression Coefficient for
quantitative calibration and Property Spectra for Cluster calibrations.
Use Useful for wavelength selection.
Method PCR / PLS / Cluster (CLU)
Matrices ID 31
Tip Select wavelengths with high absolute values.
Details Cluster method calculates the mean spectra of all C-Set spectra with the
same property.
Related Topic Loadings , Property Wavelength Regression
In quantitative calibration the Regression Coefficients is used for building up the prediction
equation: