- Application Developers get useful information for their own NIR model development process or insight model validation.
- Chip sensor manufacturers can reduce detector/sensor hardware cost
- The array can be restricted to the minimum necessary area/pixels for a specific application, allow lower sensor cost.
- Dedicated application / sensor device suppliers can reduce scanning time
- The measurement process can skip over the gaps (MEMS) in the wavelength range for dedicated applications, allow faster measurement time.
Why do Wavelength Selection for Calibration and Prediction?
- Because the selected wavelength / wavenumber range depends heavily on the specific application.
- To eliminate unwanted or noisy wave bands from the measured samples (application), optical path (e.g. fiber optics) or detector ranges (signal to noise).
- The Data-Preprocessing does ignore noisy wave bands (no artifacts or amplifying noise).
- The model robustness can be increased.
- The model prediction performance (accuracy, precision) can be increased.
How it works
- Send your NIR and Lab-data to create an fully optimized calibration for your NIRS-Application (how the Calibration Model service works, see NIR-Predicor, see also Anonymize your NIR spectroscopy data).
- You can test the calibration on new measured NIR data with the free NIR-Predictor software.
- Get payed access to the selected wavelengths of the optimize calibration.
- The selected wavelengths are delivered in the form of
“7828-6476, 5560-4208 [1/cm] (number of data points = 678).”
If you supply the spectral data in wavelengths [nm] instead of wavenumbers [1/cm], this is taken into account accordingly. The number of data points of the “number of data points” selection is specified for control purposes to check whether the interpreted ranges include the number of points.
See also Anonymize your NIR spectroscopy data
Learn more about NIR-Predicor
NIR Absorption Bands , Near Infrared Absorption Bands , NIR absorbance peaks , wavenumber and wavelength ranges , combination bands , functional groups CH, CH2, CH3, CHO, NH2, OH, ArOh, ArCH, CONH2, H2O , How to find the most relevant predictors in NIR , variable selection to improve stability and prediction , subset selection methods , NIR wavelength range , NIR wavenumber range , NIR wavelengths selection , NIR wavenumbers selection , NIR variable selection , select predictors , variable selection in multivariate data , reduce model complexity , NIR predictors selection , significant variables selection , select significant variables , NIR multivariate predictors selection , influence of spectral range and resolution on model performance