Coverage for dariah_topics/modeling.py : 25%

import lda 

from dariah_topics import utils 

import gensim 

def lda(document_term_matrix, topics, iterations=1000, implementation='lda', gensim_corpus=None, 

type2id=None, path_to_mallet=None, clean_tokenized_corpus=None, document_labels=None, 

output_topic_keys=None, output_doc_topics=None, **kwargs): 

if implementation == 'lda': 

model = lda.LDA(n_topics=topics, n_iter=iterations, **kwargs) 

model.fit(document_term_matrix) 

return model 

elif implementation == 'gensim': 

model = LdaMulticore(corpus=gensim_corpus, id2word=type2id, num_topics=topics, iterations=iterations, **kwargs) 

return model 

elif implementation == 'mallet': 

Mallet = utils.Mallet(path_to_mallet) 

mallet_corpus = Mallet.import_tokenized_corpus(clean_tokenized_corpus, document_labels) 

Mallet.train_topics(mallet_corpus, 

output_topic_keys=output_topic_keys, 

output_doc_topics=output_topic_keys, 

num_topics=topics, 

num_iterations=iterations, 

**kwargs) 

else: 

raise ValueError("{} is no supported LDA implementation".fromat(implementation))