Analytical Figures of Merit including Net Analyte Signal and Selectivity
The function figmerit.m calculates analytical figures of merit for PLS and PCR models as given in the recent article by Lorber, Faber and Kowalski (Anal. Chem., Vol. 69, No. 8, April 15, 1997). The inputs are the preprocessed (usually centered and scaled) spectral data, the preprocessed analyte data, and the PCR or PLS approximation to x, Rhat. Generally, Rhat is found by multiplying the scores by the loadings from PLS or PCR, using the number of LVs or PCs in the corresponding calibration model. The outputs are the matrix of net analyte signals for each of the spectra, the norm of the net analyte signal for each sample (nnas), the matrix of sensitivities for each sample (sens), the vector of selectivities for each sample (sel), and the “noise filtered” estimate of the net analyte signal (nfnas), which is just the multiple of the regression vector that best fits the nas.
For example, given the 7 LV PLS model formed using the PLS function in the PLS_Toolbox:
>> [reg,ssq,p,q,w,t,u,b] = pls(x,y,7); >> Rhat = t*p'; >> [nas,nnas,sens,sel,nfnas] = figmerit(x,y,Rhat);
Note that in later releases of PLS_Toolbox, the figmerit function is included and the I/O is different. Please see the figmerit documentation page for more information.
Requirements for running figmerit
- MATLAB 4.2 or 5.0
- No other toolboxes required
- Must be a registered PLS_Toolbox user.
Barry M. Wise
Eigenvector Research, Inc.
To get it, simply click on figmerit.m for Mac, or figmerit.m for PC, then enter your username and password. Move the file to a folder on your MATLAB path and you’re done.