Discriminant analysis for unveiling the origin of roasted coffee samples: A tool for quality control of coffee related products
authors de Toledo, PRAB; de Melo, MMR; Pezza, HR; Toci, AT; Pezza, L; Silva, CM
nationality International
journal FOOD CONTROL
author keywords Chemical markers; Coffee quality; Discriminant analysis; Geographic origin; Volatiles composition
keywords CHROMATOGRAPHY-MASS-SPECTROMETRY; ARABICA COFFEE; HEADSPACE; AROMA; DIFFERENTIATION; FINGERPRINT; PROFILES; ODORANTS; BEVERAGE; IMPACT
abstract Coffee quality is highly dependent on geographical factors. Based on the chemical characterization of 25 coffee samples from worldwide provenances and same roasting degree, Discriminant Analysis (DA) was employed to develop models that are able to identify the continental or country (Brazil) provenance of blind coffee samples. These models are based on coffee composition, particularly on several key compounds either with or without significant impact on aroma, such as 2,3-butanedione, 2,3-pentanedione, 2-methylbutanal and 2-ethyl-6-methylpyrazine. All models were validated with new and independent data from literature, and also through cross validation and permutation tests. Furthermore, the robustness of the proposed models in case of incomplete characterization data was also tested, being concluded that missing data is supportable by the models. In the whole, this article provides compelling arguments for the development of DA-based tools with the purpose of controlling the quality of coffee in terms of their continental and/or national origins. (C) 2016 Elsevier Ltd. All rights reserved.
publisher ELSEVIER SCI LTD
issn 0956-7135
isbn 1873-7129
year published 2017
volume 73
beginning page 164
ending page 174
digital object identifier (doi) 10.1016/j.foodcont.2016.08.001
web of science category Food Science & Technology
subject category Food Science & Technology
unique article identifier WOS:000390965800006
  ciceco authors
  impact metrics
journal impact factor (jcr 2016): 3.496
5 year journal impact factor (jcr 2016): 3.584
category normalized journal impact factor percentile (jcr 2016): 91.085
dimensions (citation analysis):
altmetrics (social interaction):



 


Sponsors

1suponsers_list_ciceco.jpg