1

I have access to 10,000 comments for a mobile app, and I want to run some interesting statistical analysis on them.

What I have done so far: Look at the frequency of each word in all the comments. Then look at a subset of these words that are relevant for performance feedback (ex: "crash", "freeze", "slow") and simply note the frequency of appearance.

Would you guys suggest any other methods to extract relevant information from these comments? I have access to excel and SQL.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
David Alisha
  • 11
  • 1
  • It kind of depends on what kind of information you would find relevant. #data dredging. How about apply zipf's law to estimate the total number of words in the mobile app vocabulary (including the words not in your data)? – zkurtz Sep 15 '13 at 23:06
  • Thanks for the comment. I just looked into zipf's law and I can see the application. What I'm mostly interested in is segmenting the comments into "negative", "neutral" and "positive", where negative comments show the flaws of the app, neutral comments just state that the app is satisfactory and positive comments praise the app. – David Alisha Sep 15 '13 at 23:23

0 Answers0