Showing 1–1 of 1 results for author: Kontemeniotis, N

Simplicial clustering using the $α$--transformation

Authors: Michail Tsagris, Nikolaos Kontemeniotis

Abstract: We introduce two simplicial clustering approaches for compositional data, that are adaptations of the $K$--means and of the Gaussian mixture models algorithms, by employing the $α$--transformation. By utilizing clustering validation indices we can decide on the number of clusters and choose the value of $α$ for the $K$--means, while for the model-based clustering approach information criteria comp… ▽ More We introduce two simplicial clustering approaches for compositional data, that are adaptations of the $K$--means and of the Gaussian mixture models algorithms, by employing the $α$--transformation. By utilizing clustering validation indices we can decide on the number of clusters and choose the value of $α$ for the $K$--means, while for the model-based clustering approach information criteria complete this task. extensive simulation studies compare the performance of these two approaches and a real data set illustrates their performance in real world settings. △ Less

Submitted 28 September, 2025; v1 submitted 7 September, 2025; originally announced September 2025.