https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html https://www.euro-online.org/enog/inoc2007/Papers/m https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html

Calibration for Protein in Soybean meal

jhonattan.baez's picture
Forums: 

Hi Everybody
I´m making a calibration for protein, but I have obtained a high RMSECV and low R^2, I have 120 samples, I don´t know why it´s happens. I´m working with QUANT software (Bruker), and I have already optimized the wavelength and the proccesing data, nevertheless the minimun RMSECV is aproximately 0.6 and the maximun R^2 is 0.7. I have a previous calibration with 90 spectrums with some of this samples that were taken before, for this calibration  the minimun RMSECV is aproximately 0.1 and the maximun R^2 is 0.98. I don´t understand why when I take the spectra in other time, and I include new samples, the calibration is not same.

thanks for your help, and i apologise if my english is not good.

DagmarB's picture

Hi Jonathan,
there could be different reasons: different sample matrix, not sufficient sample presentation or even a change in instrument settings etc.
I would recomment to contact your local Bruker office or you contact me in Germany. I'm food and feed manager and a NIR specialist. If you send some older and some new spectra we can check.
Where are you located?
Send me an email to [email protected]
Thanks and regards
Andreas