Tuesday, April 20, 2021

Bayes

I found out about nomograms VERY recently, here applied to delta-V needs on your route to some planetoid. Turns out - there's more! Apparently back in 1975 one TJ Fagan used one to simplify Bayes' Theorem, for doctors. doi 10.1056/NEJM197507312930513.

There's some online froo fah rah ongoing about whether Bayes is (still) obscure. It's not obscure in astronomy; it's used all the time in reducing noise from data so we can tell what periodic forces are affecting starlight. (That and something called "Monte Carlo".) What is obscure - to me - is what they're even talking about. I tend to skip those parts of a paper. Like I used to skip Lambert Problems and how to solve Kepler equations, until I got sick of being a moron found a concrete problem to solve to which end I taught myself, which took me about all March.

So yeah, if you're not actively applying your math, Bayes is going to be something someone else does, so will be obscured to you. I question how many <3Science! reporters know how to do Bayes. I suspect: few. Although the NYT seems decent at it.

Well - Fagan's where to start, if you are so inclined. One might look also at Carvalho - Page - Barney, 10.1214/18-BA1112.

No comments:

Post a Comment