Discrimination of Bayberry Juice Varieties by Vis/NIR Spectroscopy Based on PLS Analysis and Bayesian-SDA

Author

Cen, Hai-yan ; Bao, Yi-Dan ; He, Yong

Author_Institution

Coll. of Biosyst. Eng. & Food Sci., Zhejiang Univ., Hangzhou

fYear

2006

fDate

13-16 Aug. 2006

Firstpage

3181

Lastpage

3185

Abstract

Bayberry juice is a green beverage contained many different kinds of nutrition components. Visible and near infrared reflectance spectroscopy (Vis/NIRS) offers a promising means for the detection of bayberry juice varieties. A total of 60 bayberry juice samples (20 for each species) were investigated for Vis/NIRS on 325-1075 nm using a field spectroradiometer. Absorbance of 20 samples of each variety after pretreatment was exported and the lacked Y variables were assigned independent values for partial least squares (PLS) analysis. 8 principal components from PLS were used as 8 variables in Bayesian stepwise discriminant analysis (Bayesian-SDA). The full-cross validation results of PLS, i.e., standard error of prediction (SEV) 0.415, correlation coefficient (r) 0.865 and root mean square error of validation (RMSEV) 0.412, indicated an optimum model for bayberry juice identification. By handling these 8 principal components from PLS analysis with Bayesian-SDA, the results with deviation presented a 100% bayberry juice detection rate. Thus, it could be concluded that PLS analysis combined with Bayesian-SDA was an available alternative for species recognition based on Vis/NIR spectroscopy

Keywords

Bayes methods; beverages; infrared spectroscopy; least mean squares methods; principal component analysis; visible spectroscopy; 325 to 1075 nm; Bayesian stepwise discriminant analysis; bayberry juice detection rate; bayberry juice identification; correlation coefficient; field spectroradiometer; green beverage; near infrared reflectance spectroscopy; partial least squares analysis; principal component analysis; root mean square error of validation; standard error of prediction; visible spectroscopy; Bayesian methods; Independent component analysis; Infrared detectors; Infrared spectra; Least squares methods; Predictive models; Reflectivity; Root mean square; Spectroradiometers; Spectroscopy; Bayberry juice; Bayesian SDA; PLS analysis; Vis/NIRS;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics, 2006 International Conference on

Conference_Location

Dalian, China

Print_ISBN

1-4244-0061-9

Type

conf

DOI

10.1109/ICMLC.2006.258415

Filename

4028614