News for hackers. Stuff that matters.

What algorithm should I use for trending detection?

What algorithm should I use for detecting trending items? (Whether that is toics, or articles)

I am tempted to just look for articles read which have a moving average period M which is high than their moving average period N. (Where M is smaller than N).

( 2 comments | 2 points | on HN )
Re:What algorithm should I use for trending detection? by pjonesdotca on Monday January 18, 2010 @07:53PM (#1061208)
I would suggest checking out "Programming Collective Intelligence" and review their chapter on decision trees. That way you would be able to build a 'model' of 'trends' based on "reads" per "time quotient" and use that for your classifier.
Re:What algorithm should I use for trending detection? by azeemazhar2 on Monday January 18, 2010 @08:16PM (#1061239)
thanks. slaps head. goes to own library.