Skip to main content

Table 3 Different spectral transformations. Effect of the different preprocessing of the spectra on the prediction of total sugar, xylose and glucose release during bioethanol production (R2 coefficient of determination, RMSE root-mean-square error, CV cross-validation data set, EV external validation data set, F number of factors used in calibration)

From: Rapid estimation of sugar release from winter wheat straw during bioethanol production using FTIR-photoacoustic spectroscopy

Preprocessing Total sugar Glucose Xylose
R 2 F RMSE (g g−1 dm) R 2 F RMSE (g g−1 dm) R 2 F RMSE (g g−1 dm)
CV EV CV EV CV EV CV EV CV EV CV EV
Savitzky-Golay smoothing (seven points) 0.70 0.69 5 0.029 0.030 0.65 0.63 5 0.019 0.019 0.67 0.65 5 0.015 0.015
Normalisation by mean
Savitzky-Golay first derivative 0.71 0.70 4 0.029 0.030 0.65 0.64 4 0.019 0.019 0.68 0.66 4 0.014 0.015
(second order polynomial, seven smoothing points)
Savitzky-Golay second derivative 0.71 0.63 4 0.029 0.030 0.66 0.55 4 0.018 0.021 0.70 0.60 5 0.014 0.016
(second order polynomial, seven smoothing points)
Savitzky-Golay smoothing (seven points) 0.71 0.69 5 0.029 0.030 0.66 0.63 5 0.018 0.019 0.67 0.65 5 0.015 0.015
Standard normal variate (SNV)