Thoughts by Your Hommie
Statistics worked out in the open — with the code that produced every number.
Three things happen here, and they all share one rule: if a claim can be simulated, it gets simulated before it gets believed.
Stats & ML cases
Real problems worked end to end — the modelling decision, what it cost, and what the alternative would have produced.
Simulations
Beliefs checked before the data arrives — sampling behaviour under assumptions that are actually wrong.
Book revisions
Statistics textbooks rewritten topic by topic, with the derivation replaced by a simulation wherever one makes the point better.
How to read the code here
Posts are written in either R or Python — whichever suits the problem, one per post. Every code block is visible by default and can be folded away, and the source of any post is a click from its page. Interactive figures are client-side: they work from a saved copy of the page with no server running.