Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Fwiw, I believe the best estimator has been improved on since HyperLogLog, with a more recent result that is provably optimal (and slightly faster asymptotically, dropping the loglog factor), which perhaps more importantly can also process streamed data online: http://people.seas.harvard.edu/~minilek/papers/f0.pdf


A friend mentioned the existence of this, but I couldn't find it myself. Thanks for pointing it out.

All the algorithms can process data in a streaming fashion, though - they only require a single pass.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: