demo
When 99% accurate isn't
You take a 99%-accurate medical test. It comes back positive. The probability you're actually sick? Maybe 2%. Welcome to Bayes.
Why this matters in ML
Bayes' theorem isn't just for medical tests. It's the framework every ML model uses to update beliefs from evidence. Naive Bayes classifiers, Bayesian inference, MCMC, prior + likelihood = posterior — all the same shape. Understanding the disease-test paradox makes every other Bayesian construction click.
Anchored to 01-math-foundations/probability-statistics.