Research Article
Robust and Automated Internal Quality Grading of a Chinese Green Tea (Longjing) by Near-Infrared Spectroscopy and Chemometrics
Table 2
Classification results of three grades of Longjing with different preprocessing methods.
| | Objects | Preprocessing | Sensitivity | Specificity |
| | | Raw (20a) | 1.000 (161/161b) | 0.967 (726/751c) | | Grade I | Smoothing (20) | 0.994 (160/161) | 0.955 (717/751) | | D2 (15) | 0.975 (157/161) | 0.963 (723/751) | | | SNV (20) | 1.000 (161/161) | 0.979 (735/751) | | | Raw (20) | 0.938 (273/291) | 0.942 (585/621) | | Grade II | Smoothing (20) | 0.914 (266/291) | 0.931 (578/621) | | D2 (15) | 0.924 (269/291) | 0.957 (594/621) | | | SNV (20) | 0.955 (278/291) | 0.952 (591/621) | | | Raw (20) | 0.907 (417/460) | 1.000 (452/452) | | Grade III | Smoothing (19) | 0.889 (409/460) | 1.000 (452/452) | | D2 (15) | 0.933 (429/460) | 0.996 (450/452) | | | SNV (20) | 0.924 (425/460) | 0.996 (450/452) |
|
|
number of PLSDA components used to distinguish one grade from the other two grades.
bTP/(TP + FN).
cTN/(TN + FP).
|
