merge_classes: Merge CLIMB classes into parent groups.

In hillarykoch/CLIMB: Composite LIkelihood eMpirical Bayes

Merge CLIMB classes into parent groups.

Description

Use a theoretical distance between estimated classes to merge them together. This can be useful for more parsimonious visualizations.

Usage

merge_classes(n_groups, chain, burnin)

Arguments

n_groups	Integer. How many parent groups to merge the classes into?
chain	An object output from the extract_chains function.
burnin	Iterations to be used as burn-in.
multichain	Logical. Defaults to FALSE. Set to true if analyzing multiple different MCMC outputs together. 'chain' should be a list of objects output from the extract_chains function.
method	Agglomeration method to be used. See hclust.
...	Other arguments to be passed to hclust.

Value

merged_z	New group labels assigned to each observation
merged_mu	A matrix of merged class means with each column corresponding to a dimension and each row corresponding to a group.
merged_prop	Merged class mixing weights.
An object output from hclust, showing hierarchical relationship among the non-empty classes.

Author(s)

[e-mail]

hillarykoch/CLIMB documentation built on Oct. 24, 2022, 4:27 a.m.