I have a log with event name and timestamp of all the action users are doing with an application for exemple Photoshop

I would like extract frequent event sequences using Sequential pattern mining so i can find what are the mistake (bad sequence of action) that user do often.

What is the best algorithm to achieve that.

asked Oct 04 '11 at 23:54

maxime%20caron's gravatar image

maxime caron
51447

I think what you are looking for is motif detection. I have no pointers, but maybe you can google yourself somewhere from there.

(Oct 05 '11 at 05:12) Justin Bayer

One Answer:

I don't think I understand your question.

If you want to get the most likely sequences, you could try using Conditional Random Fields or Markov Chains, which I used once to do a validation of user activities before doing the log on the database.

If you already have the database, then you can use Markov hidden models to obtain likely sequences of events.

answered Oct 05 '11 at 04:33

Leon%20Palafox's gravatar image

Leon Palafox ♦
40857194128

yes i already have the database and just want to extract common usage pattern. I will look into markov hidden models thanks. i will accept your response as the answer in the next few day.

(Oct 06 '11 at 13:01) maxime caron
Your answer
toggle preview

powered by OSQA

User submitted content is under Creative Commons: Attribution - Share Alike; Other things copyright (C) 2010, MetaOptimize LLC.