lda_example.py

import numpy as np
import lda
import lda.datasets


# ----------------------------
# Load LDA built-in datasets
# ----------------------------
X=lda.datasets.load_reuters()
vocab=lda.datasets.load_reuters_vocab()
titles=lda.datasets.load_reuters_titles()

# ---------------------
# LDA model fitting
# ---------------------
model=lda.LDA(n_topics=30,n_iter=1500, random_state=1)
model.fit(X)
topic_word=model.topic_word_   # it's the topic distribution in the model


# ---------------------------------------
# Print the first 10 words of each topic 
# ---------------------------------------
n_top_words=10
for i, topic_dist in enumerate(topic_word):
	topic_words=np.array(vocab)[np.argsort(topic_dist)][:-(n_top_words+1):-1]
	print('Topic {0} : {1}'.format(i, ' '.join(topic_words)))