2
$\begingroup$

I am doing text mining to extract topics from documents. I started with Latent Dirichlet Allocation (LDA), which worked great, but then I came across TF-IDF with K-Means clustering, which worked better for me. I'd like to evaluate both but I can'f find any useful validation or metric to compare these two. How is it possible to compare these two with a useful metric?

$\endgroup$
1

1 Answer 1

2
$\begingroup$

If you have the ground truth value of the documents (their topics) all you gotta do is pick a metric and compare results. For classification problems, as yours, a common metric would be f1_score; reference: http://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html

$\endgroup$
2
  • $\begingroup$ K-means returns cluster centroids as "topics" and LDA assigns words to the different topics. I have a bunch of other topics to describe these documents but how can I use them as the ground truth value? Could you please give me an example? $\endgroup$
    – P.Tail
    Commented Jun 15, 2017 at 6:58
  • $\begingroup$ Do you have any good examples of a set of documents with the ground truth value? $\endgroup$ Commented Jul 26, 2018 at 14:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.