Definition of Outliers Using Unsupervised Principal Component Similarity Analysis for Sensory Evaluation of Foods

Abstract

As unsupervised classifications, principal component similarity (PCS) and cluster analysis (CA) were compared for outlier detectability in panel evaluation. By rotating the reference, PCS can define outlying panelists based on the similarity of their evaluation patterns with that of the reference panelist. As a result, the outliers detected on PCS scattergrams are dependent on the reference selected, whereas, outliers detected by CA are based on dissimilarity, thus being rather unilateral. The definition of outliers in PCS is new as it is different from the currently most popular definitions based on dissimilarity. For verifying the outliers thus obtained, random-centroid optimization (RCO) was applied for selecting the best samples by each cluster of panelists. This combination of PCS=RCO may be useful in finding the likeness distribution among consumers and then in creating food products to correctly respond to the demands of different consumer groups. *Corresponding

Unsupervised classification methods in food sciences: discussion and outlook

“…Nakai et al 37 claimed that a human was able to classify objects, on the basis of a bi‐plot, better than a computer. By that, they confirmed the opinion of Buydens et al ,50 who stated that computer methods were still not able to recognise data patterns better than a human.…”

“…Although Vodovotz et al 35 reported lower classification accuracy of PCS compared with SDA, they concluded that the former is useful because it can be used with a small number of samples and has the capacity of detecting outliers. An application of PCS in detecting outliers is presented in the paper by Nakai et al 37…”

“…Nakai et al 37 compared PCS, combined with random centroid optimisation, and heuristic cluster analysis (HCA) for outlier detection in panel evaluation. They used the methods to find outliers in an investigation on glass shape optimisation, which was prepared subject to just one attribute, namely shape.…”

Pattern similarity study of functional sites in protein sequences: lysozymes and cystatins

“…For classification of uncharacterized sequences, the PCS scatterplots are also useful as shown in Figures 1, 5 and 7. The PCA demonstrated the classifying capacity superior to that of distance-based cluster analysis [39]. The PCS is more flexible than cluster analysis as different pattern similarity patterns can be drawn by rotating the reference segment for searching.…”

“…The method described in the previous papers [9,39] was followed. Principal components analysis (PCA) was modified to principal components similarity (PCS) by incorporating linear regression of PC scores to be able to account for more than three PC scores on a 2D scatter plot.…”