Towards robustness and sensitivity of rapid Baijiu (Chinese liquor) discrimination using Raman spectroscopy and chemometrics: Dimension reduction, machine learning, and auxiliary sample

Wang, Chenhui; Shi, Zhuangwei; Shen, Haoqi; Fang, Yifei; He, Songgui; Bi, Hai

doi:10.1016/j.jfca.2023.105217

Cited by 11 publications

(1 citation statement)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…DA analysis is to establish a discriminant model by using the samples of known categories,to discriminate the categories of unknown samples,so that the cases of wrong judgment are minimized, Chenhui Wang et al (Wang et al 2023) By utilizing LDA-ML for data dimension reduction and enhancing discriminant ability, this approach achieves an overall test discriminant accuracy of 97%. Consequently, it enables robust and sensitive differentiation of highly similar liquor products, thereby highlighting the potential application of this technology in manufacturing quality control and product certi cation.…”

Section: For the Dummy Variable Ymentioning

confidence: 99%

Discrimination of Chinese Huzhu Qingke Liquor by using Ultra-Violet Spectroscopy Combined with Chemometrics

Zhao,

Zhang,

et al. 2024

Preprint

View full text Add to dashboard Cite

Qingke liquor, a renowned Tibetan alcoholic beverage derived from hull-less highland barley exclusively cultivated in the Qinghai-Tibetan Plateau, has witnessed a surge in sales. However, the issue of adulteration has emerged as a pressing concern demanding immediate attention. The research focuses on the rapid identification methods of 'Huzhu' brand Qingke liquor, a geographical indications protection product, using UV spectroscopy. Two approaches are proposed: principal component analysis-support vector machine (PCA-SVM) and multi-model partial least squares-discriminant analysis (MPLS-DA). Three categories of liquors are considered: Chinese 'Huzhu' Qingke Liquors (CHQL), Other Brand Qingke Liquors (OBQL), and Non-Qingke-Based Liquors (NQBL). SVM is performed using two principal components to solve the binary classification problem, while PLS1 algorithm is used for each column of the dummy variable Y in MPLS-DA to integrate prediction results from submodels. Both PCA-SVM and MPLS-DA successfully build discrimination models for CHQL. PCA-SVM distinguishes CHQL from OBQL and NQBL but cannot differentiate between OBQL and NQBL. In contrast, MPLS-DA correctly identifies all three classes of samples. These results demonstrate that the proposed method can serve as a simple and rapid identification approach for CHQL, with MPLS-DA exhibiting superior sample recognition capabilities.

show abstract

Section: For the Dummy Variable Ymentioning

confidence: 99%