Complete UV/VIS/NIR Systems 
Liquids and Gases Worldwide Sales and Support
Liquids and Gases Worldwide Sales and SupportComplete UV/VIS/NIR Systems Liquids and Gases Worldwide Sales and Support
Liquids and Gases Worldwide Sales and Support
OutlierDetect for Photometer Calibrations
OutlierDetect is a powerful plug-in program for Microsoft® Excel® that we use it to calibrate our Optical Solutions' brand SMART photometers.
Enter the absorbance data measured from the photometer, drag-and-click over the absorbance data, highlight the predicted property and then just click on one button.
OutlierDetect then goes to work using Principal Component Analysis (PCA) and Multiple Linear Regression (MLR) in the background to quickly provide you with your calibration coefficients.
Furthermore, it will highlight any sample that appears to be an chemical outlier based upon two outlier statistics. You can then decide to remove it from the model and recompute, again, just with a click of a button.
OutlierDetect was designed to be easy-to-use at the manufacturing plant by instrument specialists. No special Chemometrics training is required to use it.
Why It Works for Multiple Discrete Wavelengths
Once a calibration requires more than a single wavelength (i.e. a linear regression), it falls into the same category as a full spectrum analysis. The single wavelength example is referred to as a zero-order problem, as is a temperature, pressure or pH measurement. Once multiple wavelengths are used, it becomes a first-order problem. Thus, a photometer using four analytical wavelengths and one reference wavelength is fundamentally no different from a spectrophotometer analyzing 1000 contiguous wavelengths. As we will demonstrate herein, the same type of outlier statistics used in the full-spectrum analyzer can now be easily used in the ChemView® photometer.
How it Works
You put the wavelengths used in the first row of the spreadsheet. In this case, we are using data from 1500, 1594, 1134 and 1622 nm. Then, you enter your component concentration, in this case, %X. Next, you enter the absorbance data from the photometer and concentration data from you lab or sample preparation. You then drag-and-click over the absorbance data and click ALL in the floating menu.
OutlierDetect takes over, makes the computations and presents you with all of the results. It computes the 95% confidence limits for the M (Mahalanobis Distance) and Q (Sum of Squares) outlier statistics. It computes the SCORES for principal components 1 and 2. It highlights any sample the exceeds these outlier limits in Blue (see arrow below) so that you may remove that sample from the model and then recompute. Easy!
It produces a new worksheet page that shows the results of the MRL with ANOVA statistics and the new coefficients that you enter into your ClearView, ChemView, Simulplex or ChemViewMx photometer.
Now, you simply enter the coefficients shown in yellow into the photometer, zero it and you are ready to predict your chemistry on-line in real-time.