logoalt Hacker News

stuxnet79 • yesterday at 5:50 PM • 3 replies • view on HN

This is the kind of pedantry I come into HN for. Poisson is such an under-utilized distribution for these kinds of phenomena.

Replies

tmoertel • yesterday at 8:31 PM

> Poisson is such an under-utilized distribution for these kinds of phenomena.

Yeah, Poisson distributions show up everywhere if you are paying attention. You can even use the theory as the basis for a fast, elegant algorithm to take parallel, weighted random samples. You can even do it in SQL!:

    SELECT *
    FROM Population
    WHERE weight > 0
    ORDER BY -LN(1.0 - RANDOM()) / weight
    LIMIT 100  -- Sample size.
For more on the Poission connection, see: https://blog.moertel.com/posts/2024-08-23-sampling-with-sql....
8note • yesterday at 11:42 PM

possion distribution is what i learned in university for describing this kind of event?

the book is underutilized?

➕ show 1 reply
Karrot_Kream • yesterday at 9:03 PM

Isn't it a natural modeling choice? If your events arrive on exponentially distributed deltas, you end up having poisson arrival counts.