Broad intuition · Study map 4 of 85
Probability, uncertainty, and causality
Suggested preparation: Learning from labels, patterns, and feedback
Picture it
A probability model keeps several explanations alive with different weights. New evidence changes those weights. A causal model asks a harder question: what changes if we intervene, rather than just observe?
What to understand
- Bayesian inference and probabilistic graphical models
- Prediction versus intervention; confounding; randomized experiments
Follow the references
1 · Course notes
UC Berkeley · Introduction to Artificial Intelligence ↗Use the Bayes nets and inference chapters. Follow one update of beliefs after an observation.
2 · Course + videos
Brady Neal · Introduction to Causal Inference ↗Use the introduction and confounding lectures; draw the causal graph before reading the notation.
You understand it when…
Ice-cream sales predict swimming accidents. Why does banning ice cream not follow from that relationship?
This is a study map. Follow the reference for the full explanation; use the question to test your mental model. Browse the reference shelf.