Variable Selection Method in the NIR Quantitative Analysis Model of Total Saponins in Red Ginseng Extract

Si-yu AN; Lei ZHANG; Xian-zhao SHANG; Hong-shui YUE; Wen-yuan LIU; Ai-chun JU

doi:10.3964/j.issn.1000-0593(2021)01-0206-04

Journals >Spectroscopy and Spectral Analysis >Volume 41 >Issue 1 >Page 206 > Article

Spectroscopy and Spectral Analysis
Vol. 41, Issue 1, 206 (2021)

Variable Selection Method in the NIR Quantitative Analysis Model of Total Saponins in Red Ginseng Extract

Si-yu AN^1、1、*, Lei ZHANG^1、1, Xian-zhao SHANG^1、1, Hong-shui YUE^1、1, Wen-yuan LIU^1、1, and Ai-chun JU^1、1

Author Affiliations

¹[in Chinese]

¹1. Tianjin Tasly Pride Pharmaceutical Co., Ltd., Tianjin Key Laboratory of Safety Evaluation Enterprise of TCM Injections, Tianjin 300402, China

show less

DOI: 10.3964/j.issn.1000-0593(2021)01-0206-04 Cite this Article

Si-yu AN, Lei ZHANG, Xian-zhao SHANG, Hong-shui YUE, Wen-yuan LIU, Ai-chun JU. Variable Selection Method in the NIR Quantitative Analysis Model of Total Saponins in Red Ginseng Extract[J]. Spectroscopy and Spectral Analysis, 2021, 41(1): 206 Copy Citation Text

show less

Fig. 1. Original near infrared (NIR) spectra of red ginseng

Download full size | View in the Article

Fig. 2. The frequency of variables selected by VCPA-IRIV

Download full size | View in the Article

样本类型	样本数	最小值/ (mg·mL^-1)	最大值/ (mg·mL^-1)	均值/ (mg·mL^-1)	标准差
校正集	40	8.08	14.72	10.587	1.34
检验集	15	9.52	12.93	10.685	0.938

Table 1. Statistical characteristics of total saponins value

View in the Article

波数变量类型	分类规则
强信息变量	DEMEAN<0, p<0.05
弱信息变量	DEMEAN<0, p>0.05
无信息变量	DEMEAN>0, p>0.05
干扰信息变量	DEMEAN>0, p<0.05

Table 2. Variable classification rules

View in the Article

建模方法	筛选变量数	主成分数	校正集		验证集
建模方法	筛选变量数	主成分数	R_c	RMSEC	RMSEP	RSEP/%
FULL-PLS	2 074	1	0.560 1	1.110 5	1.231	11.48
OPUS-PLS	93	3	0.601 3	1.086 4	1.018	9.49
CARS-PLS	51	2	0.565 3	1.105 7	1.130 3	10.54
RF-PLS	10	3	0.644	1.025 5	1.095 0	10.21
VCPA-PLS	12	10	0.951 2	0.412 3	1.067 2	9.95
VCPA-IRIV-PLS	18	10	0.928	0.494 1	0.856 7	7.99

Table 3. Comparison on modeling results by different variables screening methods

Download Citation

Save the article for my favorites

Paper Information