siqili's Blog 要多想

Notes of Causal Inference-2

2021-01-03
Zheng-Mao Zhu

Course Web

Bayesian networks and causal graphs

Bayesian networks discrib statistical models (no causality):

image-1

With local markov assumption, we can do bayesian network factorization:

image-2

Minimality assumption:

image-3

Causal edge assumption:

In a directed graph, every parent is a direct cause of all its children.

The Blocks in Graphs

Here we define the “blocks” in graphs, where the blocked path means $X$ and $Y$ are independent.

image-4

image-5

The proof is easy in chains:

image-6

However, things are different in immoralities, including its descentants:

image-7

image-8

image-9

Here we define the “blocked path”:

image-10

And then we define “d-separation” based on the blocked path:

image-11

We can distinguish causal association and confounding asscociation with causal egde assumption:

image-12


分享到:

Similar Posts

Comments