I ran a survey asking this question: 'What do you like the most about your favorite character in the game?' and I got 23K responses; most of them are phrases/sentences; what's the best way to analyze these results such as to find out what the main responses are? A simple word cloud may not do because it strips out the context by isolating words; but I'm open. Would love help. Didn't know how to attached the dataset (my first time on Stack-Ex) so [here][1] it is. Thanks in advance fellow coders/analysts.

  • $\begingroup$ Check out Latent Dirichlet Allocation (LDA). It's a probabilistic model for words that sorts words and responses into a pre-determined number of topics. $\endgroup$
    – TBSRounder
    Mar 23, 2016 at 19:48
  • $\begingroup$ Sentiment Analysis might be another option, check out the syuzhet package in R. $\endgroup$
    – DaBenski
    Mar 23, 2016 at 21:08
  • $\begingroup$ you can add the link as [link](http://example.com), what is your base point to decide the main responses are ?, do you want classification of the sentences into positive and negative sentence? $\endgroup$ Apr 23, 2016 at 3:34
  • $\begingroup$ you can always compute n-grams for your corpus and determine their frequency so that they can be displayed in a word cloud (or more accurately a phrase cloud). $\endgroup$ Jun 21, 2016 at 23:07
  • $\begingroup$ Consider the discussion to this question: datascience.stackexchange.com/questions/9886/… $\endgroup$
    – mapto
    Oct 29, 2018 at 8:16


Your Answer

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

Browse other questions tagged or ask your own question.