BUCHI NIRCal - Manual (Page 196)

NIRCal 5.5 Software Manual
196 NIRCal 5.5 Manual, Version A
Y = (X-X mean) * P + Y mean
Where:
Y
predicted property value
X pretreated spectrum
X mean Mean of C-Set spectra ("C-Set Mean Spectrum")
Y mean Mean Property value of the C-Set spectra ("C-Set Mean Property")
P Regression Coefficients
The Regression Coefficients can be opened as Excel table in the Regression Coefficients plot
pressing "G".
The value of "Y mean" can be found in the Matrices: "C-Set Mean Property".
3.18.74 Sample
Description Contains the sample index number.
Use For 1D-scatter plots. Can be used for Outlier selection.
Method PCR / PLS / Cluster (CLU) / SIMCA / MLR
Matrices ID 113
Tip Spectra from the same sample get the same sample index.
Details Sample is a column vector.
Related Topic
Description
Contains the sample index number.
Use For 1D-scatter plots. Can be used for Outlier selection.
Method PCR / PLS / Cluster (CLU) / SIMCA / MLR
Matrices ID 114
Tip Spectra from the same sample get the same sample index.
Details Sample is a row vector.
Related Topic
3.18.75 Scores
Description Each spectra is placed in the n-dimensional score space. The position of the
spectra is given by the n-dimensional coordinate of the scores.
Use Similar spectra are placed near each other. Look for clustering effects and for
Outliers.
Method PCR / PLS / Cluster (CLU) / SIMCA
Matrices ID 6
Tip Looks very nice in a 3D scatter plot or 2D scatter plots.
Details Vin: score of the i PC and n spectrum
Formula
Reconstruction of a spectrum
Related Topic Loadings
Scores are the weightings of each PC after the pretreated spectrum has been transformed by PCA. A
score is the portion of a PC used for the spectra reconstruction. Each spectrum has different scores
for each primary factors.
The scores are visible in a 2 or 3 dimensional scatter plot.
In qualitative calibration, the separate scores determine the number of secondary PCs.