Rapid Determination of Green Tea Origins by Near-Infrared Spectroscopy and Multi-Wavelength Statistical Discriminant Analysis
| Autor: | Lili Wang, H. F. Wang, J. X. Fang, X. S. Shi, X. G. Zhuang |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
business.industry
Calibration (statistics) 010401 analytical chemistry Near-infrared spectroscopy Pattern recognition 02 engineering and technology 021001 nanoscience & nanotechnology Condensed Matter Physics Linear discriminant analysis 01 natural sciences 0104 chemical sciences Wavelength Preprocessor Artificial intelligence Sensitivity (control systems) 0210 nano-technology business Spectroscopy Smoothing Mathematics |
| Zdroj: | Journal of Applied Spectroscopy. 86:76-82 |
| ISSN: | 1573-8647 0021-9037 |
| Popis: | A new simple classification modeling procedure, multi-wavelength statistical discriminant analysis (MW-SDA), is proposed for the identification of Shandong green tea origins coupled with near-infrared (NIR) spectroscopy. After smoothing and first derivative preprocessing, seven characteristic wavelengths (CW) were selected by enlarging the detailed information of preprocessed spectra. Then, for each characteristic wavelength, a classification threshold is calculated according to the differences in absorbance value, which can best separate the spectra for different origins. Based on the seven CWs and corresponding thresholds, seven classifiers were obtained, which form the classification model. The performance of the calibration model was evaluated according to sensitivity, specificity, and classification accuracy. Analysis results indicated that MW-SDA can be used well to build classification models. The predicted precision of the last model in prediction set was: sensitivity = 1, specificity = 0.967, and accuracy = 98.3%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|