hdp_extract_components: Extract major components from the raw clusters

In nicolaroberts/hdp: Hierarchical Dirichlet Process for categorical count data

Description

If prior components included via hdp_prior_init are preserved by hdp_extract_components, they are prefixed with "P". Any new components in this case are prefixed with "N".

Usage

hdp_extract_components(x, cos.merge = 0.9, min.sample = 1)

Arguments

x	hdpSampleChain or hdpSampleMulti object
cos.merge	Merge components with cosine similarity above this threshold (default 0.90)
min.sample	Components must have significant exposure in at least this many samples (i.e. those DP nodes with data assigned) (default 1)

Value

A hdpSampleChain or hdpSampleMulti object updated with component information

See Also

hdp_posterior, hdp_multi_chain, plot_comp_size, plot_comp_distn, plot_dp_comp_exposure

Examples

hdp_extract_components(mut_example_multi)

nicolaroberts/hdp documentation built on May 23, 2019, 5:09 p.m.