LECTURE HUB

🧮Lecture 14 - Updating beliefs with Bayes

🚦 Status🌟 Mastered

🧠 Mastery🦅 Adept

📅 Last Review7 Dec 2025

📐 Bayes' rule: posterior is proportional to prior times likelihood. Start from what was plausible, then weigh how well each explanation fits what you saw.
In odds: multiply the prior odds by the likelihood ratio. One multiplication, no table.
A 🔤 Prior is not a prejudice. It is the base rate from lecture 3 with a new name and permission to be updated.
Strong evidence about a rare thing still leaves it unlikely. A second independent piece of evidence changes that fast.
The picture is a square cut into strips. After the video it stayed.

back. two weeks missed: Hamburg, then the cold. Rasmus's notes for week 4 are in my inbox, unread
Hale: you were ill, Bayes was not, sit down
🔤 Prior = what you believed before. likelihood = how well each story explains the data. posterior = what you believe after
it is the metal detector from February. THE SAME SUM. I got it wrong then
odds form is easier: prior odds × likelihood ratio = posterior odds
evidence does not replace the prior. it moves it
two people with different priors can see the same data and still disagree, correctly
still coughing. Adaeze passed me a lozenge without looking up

Watch 🎬 Bayes theorem, the geometry of changing beliefs tonight while this is warm
Write 🔤 Prior in one sentence
Redo the February metal detector with 📐 Bayes' rule in odds form
Process the week 4 notes from Rasmus
Ask Hana what to watch next

📝LECTURE NOTES