Premium
Comparison of novel sensory panel performance evaluation techniques with e‐nose analysis integration
Author(s) -
Sipos László,
Kovács Zoltán,
Szöllősi Dániel,
Kókai Zoltán,
Dalmadi István,
Fekete András
Publication year - 2011
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.1391
Subject(s) - electronic nose , principal component analysis , pattern recognition (psychology) , artificial intelligence , sensory system , linear discriminant analysis , mathematics , sensory analysis , computer science , statistics , chemometrics , data mining , machine learning , psychology , cognitive psychology
Reliability and validity of sensory data is an important issue in scientific researches. If sensory analysis is performed in an analytical approach, the resulting data will show a similar structure to the chemical analyses. In the present paper the authors have used a complex approach to evaluate the performance of a sensory panel. The tested samples were black tea batches from different plantations of Sri Lanka. Profile analysis was applied to identify the odor profiles of the samples. Sensory profile data was submitted to two novel techniques of panel performance evaluation. GCAP (Gravity Center Area/Perimeter) is based on the profile polygons of the individual assessors. If the area/perimeter ratio of two panelists' profiles is similar and the gravity center is located near to each other, the panelists performed the tests consistently. CRRN (Compare Ranks with Random Numbers) is applicable not only to sensory data but also to other field of chemometrics. The essence of CRRN method is based on the evaluation of an ‘average’ vector, corresponding to the coordinate‐averages of the measured points, and on a produced random vector series of the same dimension as the measured points. Sensory and e‐nose data were evaluated with principal component analysis, cluster analysis and linear discriminant analysis. Partial least square regression and support vector machine regression were used to predict sensory data with electronic nose results. Prediction by support vector machine gave close correlation between the results of electronic nose measurement and odor attributes. Copyright © 2011 John Wiley & Sons, Ltd.