Files and links (1)
CC BY V4.0, Open Access
Journal article
Analyzing Cinnamon Spice Adulteration with Spectroscopy: The Influence of Data Preprocessing on Multivariate Prediction Models
Innovative Food Technologies, Vol.13(1), pp.1-12
2025
Abstract
Detecting fraud in the cinnamon supply chain is critical for ensuring consumer safety and maintaining product integrity. Recent advances in spectral data preprocessing techniques offer enhanced accuracy in identifying adulterants in spices like cinnamon. This study investigates the impact of different spectral preprocessing techniques on predicting adulterants—specifically soybean powder, hazelnut shell powder, and dry bread powder—mixed with cinnamon powder using spectroscopy combined with multivariate analysis. The transmittance spectra were collected across the mid-infrared range of 400–4000 cm⁻¹, and Partial Least Squares Regression (PLSR) was employed to model the adulteration levels based on these spectra. Various preprocessing methods were applied to optimize the spectral data. Among them, orthogonal signal correction (OSC) combined with detrending yielded the highest predictive accuracy, with a coefficient of prediction (R²p) ranging from 0.900 to 0.981. Conversely, Extended Multiplicative Scatter Correction (EMSC) and Savitzky-Golay second derivative (D2) were less effective, with R²p values between 0.115 and 0.931. Soybean powder was the easiest adulterant to detect, with a prediction error range of 5–10%. These findings underscore the importance of selecting appropriate preprocessing techniques to improve the accuracy of fraud detection in cinnamon powder using spectroscopic methods.
Details
- Title
- Analyzing Cinnamon Spice Adulteration with Spectroscopy: The Influence of Data Preprocessing on Multivariate Prediction Models
- Publication Details
- Innovative Food Technologies, Vol.13(1), pp.1-12
- Publisher
- Iranian Research Organization for Science and Technology
- Identifiers
- 991005821344607891
- Murdoch Affiliation
- School of Information Technology
- Resource Type
- Journal article
Metrics
2 File views/ downloads
13 Record Views