hdpSampleChain-class: hdpSampleChain class for posterior samples off one MCMC chain

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

Description

hdpSampleChain class for posterior samples off one MCMC chain

Usage

## S4 method for signature 'hdpSampleChain'
as.list(x, ...)

## S4 method for signature 'hdpSampleChain'
sampling_seed(x, ...)

## S4 method for signature 'hdpSampleChain'
hdp_settings(x, ...)

## S4 method for signature 'hdpSampleChain'
final_hdpState(x, ...)

## S4 method for signature 'hdpSampleChain'
lik(x, ...)

## S4 method for signature 'hdpSampleChain'
numcluster(x, ...)

## S4 method for signature 'hdpSampleChain'
cp_values(x, ...)

## S4 method for signature 'hdpSampleChain'
clust_categ_counts(x, ...)

## S4 method for signature 'hdpSampleChain'
clust_dp_counts(x, ...)

## S4 method for signature 'hdpSampleChain'
numcomp(x)

## S4 method for signature 'hdpSampleChain'
prop.ex(x)

## S4 method for signature 'hdpSampleChain'
comp_cos_merge(x)

## S4 method for signature 'hdpSampleChain'
comp_categ_counts(x)

## S4 method for signature 'hdpSampleChain'
comp_dp_counts(x)

## S4 method for signature 'hdpSampleChain'
comp_categ_distn(x)

## S4 method for signature 'hdpSampleChain'
comp_dp_distn(x)

Arguments

x	Object of class hdpSampleChain
...	unused

Methods (by generic)

• as.list: Convert to list class

• sampling_seed: Get random seed used by hdp_posterior

• hdp_settings: Get settings of posterior sampling chain

• final_hdpState: Get hdpState object from the end of the posterior sampling chain

• lik: Get likelihood of data given model over all iterations

• numcluster: Get the number of clusters for each posterior sample

• cp_values: Get matrix of concentration parameter values for each posterior sample

• clust_categ_counts: Get category vs cluster counts for each posterior sample

• clust_dp_counts: Get dp node vs cluster counts for each posterior sample

• numcomp: Get number of extracted components for hdpSampleChain

• prop.ex: Get proportion of dataset explained (on average) for hdpSampleChain

• comp_cos_merge: Get cos.merge setting for hdpSampleChain

• comp_categ_counts: Get sample vs category counts for each component

• comp_dp_counts: Get sample vs component counts for each DP

• comp_categ_distn: Get mean distribution over data categories for each component

• comp_dp_distn: Get mean distribution over components for each DP

Slots

seed

Random seed used by hdp_posterior

settings

Settings of the posterior sampling chain: burnin, n (number of samples collected), space (iters between samples), cpiter (con param moves between iters)

hdp

hdpState object after the final iteration

lik

Likelihood of data given model at each iteration

numcluster

Number of raw data clusters in each posterior sample

cp_values

Matrix of concentration parameter values (one column for each parameter) in each posterior sample (rows).

clust_categ_counts

List of matrices (one from each posterior sample) counting the category-cluster data assignment across all DP nodes. Number of rows is the number of categories (constant), and number of columns is the number of clusters in that posterior sample (variable).

clust_dp_counts

List of matrices (one from each posterior sample) counting within-DP cluster assignment (aggregating across data categories). Number of rows is the number of DPs (constant), and number of columns is the number of clusters in that posterior sample (variable).

numcomp

Number of global components extracted by hdp_extract_components (not including component 0)

prop.ex

(Average) proportion of dataset explained by the extracted components

comp_cos_merge

cos.merge setting used by hdp_extract_components

comp_categ_counts

List of matrices (one for each component) counting the sample-category data assignment across all DP nodes. Number of rows is the number of posterior samples, and number of columns is the number of data categories.

comp_dp_counts

List of matrices (one for each DP) counting sample-component assignment (aggregating across data categories). Number of rows is the number of posterior samples, and number of columns is the number of components.

comp_categ_distn

List with elements mean and cred.int, containing matrices with the mean (and lower/upper 95% credibility interval) distribution over data categories for each component. Number of rows is the number of components, and number of columns is the number of data categories. Rows sum to 1.

comp_dp_distn

List with elements "mean" and "cred.int", containing matrices with the mean (and lower/upper 95% credibility interval) distribution over components for each DP. Number of rows is the number of DPs, and number of columns is the number of components. Rows sum to 1.

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