pivmet: Pivotal Methods for Bayesian Relabelling and k-Means Clustering

TL;DR

The paper introduces the pivmet R package, which offers methods for extracting pivotal units to improve Bayesian relabeling and clustering tasks, including consensus clustering and visualization tools.

Contribution

It presents novel pivotal methods integrated into an R package for Bayesian relabeling, consensus clustering, and visualization, enhancing mixture model analysis.

Findings

Effective relabeling of Bayesian mixture models

Improved clustering through pivotal unit seeding

Versatile tools for Gaussian mixture analysis

Abstract

The identification of groups' prototypes, i.e. elements of a dataset that represent different groups of data points, may be relevant to the tasks of clustering, classification and mixture modeling. The R package pivmet presented in this paper includes different methods for extracting pivotal units from a dataset. One of the main applications of pivotal methods is a Markov Chain Monte Carlo (MCMC) relabelling procedure to solve the label switching in Bayesian estimation of mixture models. Each method returns posterior estimates, and a set of graphical tools for visualizing the output. The package offers JAGS and Stan sampling procedures for Gaussian mixtures, and allows for user-defined priors' parameters. The package also provides functions to perform consensus clustering based on pivotal units, which may allow to improve classical techniques (e.g. k-means) by means of a careful seeding. The paper provides examples of applications to both real and simulated datasets.