A question for any stats folk out there...
What method would I use to determine auto-correlation in a time series? The only problem is it’s not really a time series, but events at time t, which have integer values?
Conversation
Replying to
You have data on the times of up/downvotes? Cool! If you have sufficiently large numbers of votes then ACF is probably fine.
2
Lewis -- this is something that would be interesting to look at. My thoughts here are that we could ping the same 10,000 comment ids over a span of time to get information on upvote movement. You could get creative and put better logic in to detect popular submissions, etc.