Flexible Clustering for High-Dimensional Data via Mixtures of Joint Generalized Hyperbolic Models

TL;DR

This paper introduces a mixture of joint generalized hyperbolic distributions (MJGHD) for flexible, high-dimensional asymmetric clustering, emphasizing cluster-specific subspaces for better visualization and parameter efficiency.

Contribution

The paper proposes a novel MJGHD model that accounts for cluster-specific subspaces, improving high-dimensional clustering and visualization capabilities.

Findings

Effective clustering on real datasets demonstrated

Model selection via Bayesian information criterion

Parameter estimation through multi-cycle ECM algorithm

Abstract

A mixture of joint generalized hyperbolic distributions (MJGHD) is introduced for asymmetric clustering for high-dimensional data. The MJGHD approach takes into account the cluster-specific subspace, thereby limiting the number of parameters to estimate while also facilitating visualization of results. Identifiability is discussed, and a multi-cycle ECM algorithm is outlined for parameter estimation. The MJGHD approach is illustrated on two real data sets, where the Bayesian information criterion is used for model selection.

Figures