HRClusterpath

HRClusterpath is a R-package which provides tools for the variable clustering of Husler-Reiss models, specially using the graphical models structure.

Installation

To install the package, one can write this in the R-terminal :

remotes::install_github("alxkpl/HRClusterpath")

Usage

This section gives an overview of the package’s tools.

Clusterpath algorithm for hierarchical clustering

The first method you can use with the package is a hierarchical clustering (for variables) using a Clusterpath algorithm applied to the likelihood of the precision matrix of the Husler Reiss graphical model with a fused-Lasso penalty.

You can use :

• get_cluster to get optimal cluster with fixed parameter \(\lambda\) and estimated variogram \(\hat \Gamma\) with customizable weights.
• HR_Clusterpath that build a list of optimal results with a grid of \(\lambda\) and the standards exponnential weights.
• gg_cluster which provides the dendrogram induced by the results of the HR_Clusterpath function.
• and others functions to analyze the results.