Grouping of Bulgarian wines according to grape variety by using statistical methods

68 different types of Bulgarian wines were studied in accordance with 9 optical parameters as follows: color parameters in XYZ and SIE Lab color systems, lightness, Hue angle, chroma, fluorescence intensity and emission wavelength. The main objective of this research is using hierarchical cluster an...

Full description

Saved in:
Bibliographic Details
Main Authors: Milev, M., Nikolova, Kr, Ivanova, Ir, Minkova, St, Evtimov, T., Krustev, St
Format: Conference Proceeding
Language:English
Subjects:
Cluster analysis
Color
Correlation
Factor analysis
Fluorescence
Parameters
Physical properties
Principal components analysis
Statistical analysis
Statistical methods
Wines
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:68 different types of Bulgarian wines were studied in accordance with 9 optical parameters as follows: color parameters in XYZ and SIE Lab color systems, lightness, Hue angle, chroma, fluorescence intensity and emission wavelength. The main objective of this research is using hierarchical cluster analysis to evaluate the similarity and the distance between examined different types of Bulgarian wines and their grouping based on physical parameters. We have found that wines are grouped in clusters on the base of the degree of identity between them. There are two main clusters each one with two subclusters. The first one contains white wines and Sira, the second contains red wines and rose. The results from cluster analysis are presented graphically by a dendrogram. The other statistical technique used is factor analysis performed by the Method of Principal Components (PCA). The aim is to reduce the large number of variables to a few factors by grouping the correlated variables into one factor and subdividing the noncorrelated variables into different factors. Moreover the factor analysis provided the possibility to determine the parameters with the greatest influence over the distribution of samples in different clusters. In our study after the rotation of the factors with Varimax method the parameters were combined into two factors, which explain about 80 % of the total variation. The first one explains the 61.49% and correlates with color characteristics, the second one explains 18.34% from the variation and correlates with the parameters connected with fluorescence spectroscopy.
ISSN:0094-243X
1551-7616
DOI:10.1063/1.5013940