Skip to main content

Table 2 Comparison of PLS calibration models using vibrational spectroscopy and pyrolysis molecular beam mass spectrometry

From: High-throughput prediction of eucalypt lignin syringyl/guaiacyl content using multivariate analysis: a comparison between mid-infrared, near-infrared, and Raman spectroscopies for model development

Method SEL calibration1,a RMSEC2,a RM- SECV3,a R2Cal4 R2CV5 F6 Outliers7
Raman 2 nd deriv (19 pt) + SNV 32 scans 0.05 0.13 0.14 0.83 ± 0.02 0.81 ± 0.02 4-5 2
Raman 1 st deriv (7 pt) + EMSC 32 scans 0.05 0.13 0.14 0.845 ± 0.003 0.82 ± 0.01 4-5 3
Raman EMSC + 2 nd deriv (15 pt) 96 scans 0.05 0.13 0.14 0.83 ± 0.01 0.81 ± 0.01 5-6 2-3
Raman 2 nd deriv (15 pt) + SNV 96 scans 0.05 0.13 0.13 0.84 ± 0.01 0.82 ± 0.01 4-5 4-5
MIR EMSC + 2 nd deriv (15 pt) 0.05 0.13 0.14 0.84 ± 0.03 0.81 ± 0.03 3-4 1-2
MIR 2 nd deriv (17 pt) + MSC 0.05 0.13 0.14 0.82 ± 0.01 0.78 ± 0.01 3-4 1-2
MIR 2 nd deriv (17 pt) + SNV 0.05 0.13 0.14 0.85 ± 0.02 0.82 ± 0.03 3-4 2-3
NIR EMSC + 2 nd deriv (25 pt) 0.05 0.17 0.18 0.73 ± 0.01 0.681 ± 0.004 4-5 4-7
NIR 2 nd deriv (25 pt) + MSC 0.05 0.17 0.18 0.72 ± 0.02 0.68 ± 0.02 4-6 1-5
NIR 2 nd deriv (25 pt) + SNV 0.05 0.16 0.17 0.74 ± 0.01 0.70 ± 0.02 4-5 2-3
  1. 1Standard Error of the Laboratory for the calibration data.
  2. 2Root Mean Standard Error of Calibration.
  3. 3Root Mean Standard Error of Cross-Validation.
  4. 4Coefficient of determination for calibration set.
  5. 5Coefficient of determination for full cross-validation.
  6. 6Average number of factors used in model construction.
  7. 7Number of outliers removed from calibration models.
  8. aAverage errors of 3 randomly generated models using data provided. Models were not statistically different.
  9. The numbers listed parenthetically reflect the degree of Savitzky-Golay spectral smoothing.
  10. Statistical values are the average of 3 independent models.
  11. MIR = mid-infrared spectroscopy, NIR = near-infrared spectroscopy, EMSC = extended multiplicative scatter correction, MSC = multiplicative scatter correction, SNV = standard normal variate.