An empirical comparison and characterisation of nine popular clustering methods

Nine popular clustering methods are applied to 42 real data sets. The aim is give a detailed characterisation of the by means several cluster validation indexes that measure various individual aspects resulting clusters such as small within-cluster distances, separation clusters, closeness Gaussian distribution etc. introduced in Hennig (in: Data analysis and applications 1: regression, modeling—estimating, forecasting mining, ISTE Ltd., London, 2019). 30 sets come with “true” clustering. On these similarity clusterings from nine explored. Furthermore, mixed effects regression relates observable clusterings, which problems unobservable. study gives new insight not only into ability discover but also properties can be expected methods, crucial for choice method situation without given