Skip to main content

Table 2 Summary statistics for external validation of PLSR calibration models for chemical components

From: An NIRS-based assay of chemical composition and biomass digestibility for rapid selection of Jerusalem artichoke clones

Parameter

N

Full spectra

CARS-optimized spectra

RF-optimized spectra

RMSEP

R 2 V

RPD

RER

RMSEP

R 2 V

RPD

RER

RMSEP

R 2 V

RPD

RER

Soluble sugars

13

2.17

0.91

3.07

14.18

2.55

0.85

2.61

12.07

1.57

0.94

4.24

19.60

Cellulose

14

0.75

0.92

3.21

17.49

0.61

0.93

3.95

21.51

0.54

0.94

4.46

24.29

Hemicellulose

14

0.39

0.95

4.18

19.12

0.23

0.98

7.08

32.42

0.35

0.95

4.65

21.31

Lignin

14

0.58

0.89

2.98

12.97

0.66

0.93

2.61

11.40

0.61

0.84

2.83

12.33

Ash

14

0.21

0.58

3.13

16.05

0.31

0.08

2.12

10.87

0.23

0.86

2.86

14.65

  1. N number of samples, CARS competitive adaptive reweighted sampling, RF random frog, RMSEP root mean square error of prediction, R2V coefficient determination of validation, RPD the ratio of performance to deviation, RER the range error ratio