Table 1 The PLS-DA model of different preprocessing methods for screening the specific wavebands by SiPLS.

From: Novelty application of multi-omics correlation in the discrimination of sulfur-fumigation and non-sulfur-fumigation Ophiopogonis Radix

 

LVs

Model evaluation

Discriminant results

 

RMSECV

RMSEC

RMSEP

Se

Sp

Accuracy

Raw*

5

0.6459

0.4452

1.1014

0.8

1

0.9

MSC

15

0.527

4.02E-13

1.5248

0.6

0.4

0.5

SNV

14

0.5487

8.06E-06

1.5012

1

0.6

0.8

Baseline

1

1.0358

0.9997

0.9948

0.4

0.6

0.5

Normalization

4

0.6899

0.5948

0.6705

0.6

1

0.8

S-T

4

0.6365

0.5019

0.6748

1

1

1

WDS

11

0.5436

0.0254

0.4097

1

1

1

S-G(9) + 1st

4

0.696

0.5597

0.8499

1

0.8

0.9

S-G(11) + 1st

11

0.6004

0.003

0.7203

0.6

1

0.8

S-G(9) + 2nd

1

1.0057

0.6242

0.8726

0.8

0.6

0.7

S-G(11) + 2nd

1

1.0218

0.7848

0.8603

1

0.8

0.9

  1. *A variety of preprocessing methods to extract the useful information from noise for the spectroscopic data were compared, such as multiplicative scatter correction (MSC), standard normal variate transformation (SNV), baseline, normalization, spectroscopic transformation (ST), wavelet denosing of spectra (WDS), Savitzky-Golay smoothing with 9 points (SG(9)) plus first-order derivatives, SG(9) plus second-order derivatives, SG(11) plus first-order derivatives, and SG(11) plus second-order derivatives.
Back to article page