lda_fgs_lppv: Computes Log Posterior Predictive Value for an LDA model:

In clintpgeorge/ldamcmc: Markov chain Monte Carlo Algorithms for the Latent Dirichlet Allocation Model

Description

This is based on Zhe Chen (2015) [dissertation].

Usage

lda_fgs_lppv(K, V, wid, doc.N, alpha.v, eta, max.iter = 100, burn.in = 0,
  spacing = 1)

Arguments

K	Number of topics in the corpus
V	Vocabulary size
wid	a vector of vocabulary ids of every word instance in the corpus. Note: we assume vocabulary id starts with 1.
doc.N	a vector of word counts for each document in the corpus
alpha.v	hyperparameter vector for document Dirichlets θ
eta	hyperparameter value for topic Dirichlets β
max.iter	maximum number of Gibbs iterations to be performed
burn.in	burn-in period for the Gibbs sampler
spacing	spacing between the stored samples (to reduce correlation)

Value

A list that consists of (a) a (D x S) matrix of log posterior predictive values for each held-out using each sample, where S is the number of saved samples based on burn.in and spacing, and (b) log posterior predictive value of the corpus.

See Also

Other posterior predictive check (PPC) options: lda_acgs_lppv_R, lda_fgs_lppv_R

clintpgeorge/ldamcmc documentation built on Feb. 22, 2020, 12:39 p.m.