Almost everytime I use aggregate_doc_topics I have to perform the following transformation to get an array of the topic distribution:
# v.aggregate_doc_topics returns an ndarray instead of just values.
theta = v.aggregate_doc_topics(all_ids)
theta = theta[np.argsort(theta['i'])]
theta = np.array(theta['value'])
I should add an aggregate_doc_topic_matrix to return this properly formatted list, or clean up our DataTable implementation. Since the second seems super hard, I'm going to go with the first.
Almost everytime I use
aggregate_doc_topicsI have to perform the following transformation to get an array of the topic distribution:I should add an
aggregate_doc_topic_matrixto return this properly formatted list, or clean up ourDataTableimplementation. Since the second seems super hard, I'm going to go with the first.