10 - Causal Discovery from Observational Data
Вставка
- Опубліковано 10 чер 2024
- In the 10th week of the Introduction to Causal Inference online course, we cover causal discovery from observational data. Please post questions in the UA-cam comments section.
Introduction to Causal Inference Course Website: causalcourse.com
0:00 Intro
1:14 Outline
1:48 Assumptions for Independence-Based Causal Discovery
7:15 Markov Equivalence and Main Theorem
18:44 The PC Algorithm
32:43 Can We Do Better?
34:17 Issues with Independence-Based Causal Discovery
35:24 No Identifiability Without Parametric Assumptions
39:40 Linear Non-Gaussian Setting
46:48 Nonlinear Additive Noise Setting
Hello and thank you for the course. The guest talk from Jonas Peters is not in the channel's uploads. Can we find it somewhere else? thank you!
Hi Brady, thanks for your course! I appreciate it very much. A question for the beginning example, isn't A and D are d-seperated by B and C? Another question is for the skeleton example, the X1-> X2
Thanks for the nice lecture! Where can I find the guest talk of Jonas Peters?
I'm looking for the same.
Thank you for your wonderful course! I'm wondering if the nonlinear-gaussian assumption works for causal identification is coming from the ''Conjugation properties'' of Gaussian distribution, as gaussian belongs to exponential family. This is just my naive intuition, I will check the proof and refs later! thanks again!
I'm not sure haha
Nice meme at 6:54 😉
How to reach to the solutions of the questions you mention in between?
Hi sir how to interpret undirected edges in essential graphs after executing PC
hi, sir what if after executing the PC it shows the undirected graph, for example, A and B a line with no orientation on either side? How can I interpret this?
Can you explain pcmci algorithm please?
Thank you for your video. btw, where can I find pdf of this chap?
There's the course book: www.bradyneal.com/causal-inference-course#course-textbook
This chapter should come out in 1-2 days. I'll probably send an email out once it's ready on the course mailing list here: www.bradyneal.com/causal-inference-course#course-mailing-list
Alternatively, you can watch the course Slack.
@@BradyNealCausalInference Thanks, Brady!
Is this algorithm available in python?
can i give u 2 likes?