Fruit growing
and viticulture of South Russia
Ufimtseva Larisa Viktorovna
Articles in journal: (total 1)
The crops products taste is the main characteristic that determines its appeal to consumers. Therefore, at different stages of the breeding process, it is important to have a competent tasting procedure, results of which require statistical analysis and graphics. The article proposes to use cluster heat maps (HM) to visualize the experts agreement when tasting products. The material for the study was questionnaires obtained in 2020 during the assessments of 10 varieties of blue honeysuckle Lonicera caerulea L. 12 experts were involved in the tasting, who evaluated the sensory characteristics of berries, including taste, on a five-point scale. Statistical analysis of the obtained data included the calculation of the Kendalls coefficient of concordance W, the average scores, as well as the construction of a cluster HM. To group the columns and rows of the data matrix in the CH, we applied hierarchical cluster analysis using Spearman's correlation coefficient as a measure of similarity, and Ward's method as an agglomerative algorithm. Calculations were carried out in the DescTool and pheatmap packages of the statistical environment R. As a result, it was found that the agreement of experts in the ranking of varieties was statistically significant (P < 0.001), but weak: W=0.374. Therefore, the average scores for varieties do not contain all the information and a heatmap was built to expand it. This article describes a statistical standardization procedure for color-coding cells of HM. Based on the heatmaps analysis, the individual characteristics of the honeysuckle variety and the individual taste of experts, the grouping of varieties and the grouping of experts and/or consumers are discussed. It is concluded that the cluster heatmap is a useful tool combining statistical and graphical techniques that is useful even in the absence of expert agreement.