Yes, but the output of the phase in question is *not* a count. It can't be combined. You could argue that this is the problem!
On Mon, Jul 18, 2011 at 10:38 PM, Ted Dunning <[email protected]> wrote: > Most of these metrics are based on counts so a combiner would definitely help. > > On Jul 18, 2011, at 12:31, Sean Owen <[email protected]> wrote: > >> The combiner question is interesting. Many metrics happen to be built >> up from some commutative function of smaller parts, like addition. I >> think the answer is 'yes' in many cases -- but not in all cases. For >> example I don't think you can construe log-likelihood this way. I'm >> also not sure how Pearson would come out since you need the mean from >> the start, though I wouldn't be surprised if there's a clever way to >> compute it online. >
