4

I have the LDA model and the document-topic probabilities.

# build the model on the corpus
ldam = LdaModel(corpus=corpus, num_topics=20, id2word=dictionary) 
# get the document-topic probabilities
theta, _ = ldam.inference(corpus)

I also need the distribution of words for all the topics i.e. a topic-word probability matrix. Is there a way to extract this information?

Thanks!

Clock Slave
  • 7,627
  • 15
  • 68
  • 109

1 Answers1

6

The topics-term matrix (lambda) is accessible via : 

topics_terms = ldam.state.get_lambda()

If you want a probability distribution just normalize it : 

topics_terms_proba = np.apply_along_axis(lambda x: x/x.sum(),1,topics_terms)
arthur
  • 2,319
  • 1
  • 17
  • 24
  • when I use `ldam.state.get_lambda()` I get a numpy matrix but the there are no column names. How do I identify the words? – Clock Slave Feb 19 '17 at 09:09
  • 1
    To know which word corresponds to a given index, use `ldam.id2word`. For example `ldam.id2word[0]` is the word corresponding to the first column of the matrix. – arthur Feb 20 '17 at 10:04