Quantitative methods of standardization in cluster analysis

Abstract

The aim of this paper is to evaluate the impact of three standardization methods (z-score, log10 and improved min–max) in determining the number of clusters for a dataset of 146 archaeological ceramic fragments in which mass fractions of chemical elements were determined by INAA. The results showed a tendency towards clustering, which did not occur to the non-standardized data. The standardization methods indicated the presence of three groups within the database. Quality evaluation of these clusters, by means of internal validation indexes, showed that the best performance was obtained with the log10 transformation. This transformation also performed well in the calculation of compactness, while the improved min–max showed better performance in terms of separability.