lda_smoothed: Run LDA adapted to use a count matrix

In g-l-mansell/MSLDA:

Description

Run LDA adapted to use a count matrix

Usage

lda_smoothed(
  N,
  K,
  max_iter = 50,
  thresh = 1e-04,
  seed = NULL,
  cores = NULL,
  alpha = NULL,
  eta = NULL,
  NMF = FALSE
)

Arguments

N	matrix of word counts
K	the number of topics to look for
max_iter	the maximum number of EM iterations to run
thresh	threshold for L convergence, (L_i - L_i-1)/L_i < thresh
seed	set a seed for the random documents to initialise beta
cores	number of cores to run the E-step in parallel, if NULL all detected cores are used
alpha	if you want to set the exchangeable Dirichlet parameter for theta, if NULL a default value of 1/K is used
eta	the exchangeable Dirichlet parameter for beta, if NULL a default value of 1/K is used
NMF	logical indicating if lambda should be initialised using non-negative matrix factorisation, if FALSE it is generated using K random documents

g-l-mansell/MSLDA documentation built on Dec. 20, 2021, 9:43 a.m.