I am using the lda package in R to perform Latent Dirichlet Allocation modelling. However, each time I run the program I get a different output.
Using set.seed() doesn't seem to help like with the topicmodels package.
Assuming an identical input, is there a way to ensure that identical topics are found on subsequent executions of the code?
I execute the function as follows:
set.seed(11)
fit1 <- lda.collapsed.gibbs.sampler(documents = documents, K = topics, vocab = vocab,
num.iterations = iterations, alpha = alpha,
eta = eta, initial = NULL, burnin = 500,
compute.log.likelihood = TRUE)
