[DOC] Adding basic conceptual example of causal graphs

[adam2392]

I've been merging in some PRs without reviews for the sake of getting upstream PRs to work. Now I think we're at a stage where we can improve the codebase, but also not be a bottleneck for super common algorithms upstream (e.g. FCI). This PR is now a reviewable PR for the sake of converging on the communication of concepts to end-users.

Addresses partially: #3

Changes proposed in this pull request:

• Based on the previous PRs: [DOC, ENH] Cleanup repo and docs and add typing #8 , [ENH] Add PAG class #9 , [ENH] PAG algorithms to enable FCI algorithm #10 this example now walks through the concepts of fundamental graphs that are used in constraint-based structure learning and causal inference

Before submitting

• I've read and followed all steps in the Making a pull request
  section of the CONTRIBUTING docs.
• I've updated or added any relevant docstrings following the syntax described in the
  Writing docstrings section of the CONTRIBUTING docs.
• If this PR fixes a bug, I've added a test that will fail without my fix.
• If this PR adds a new feature, I've added tests that sufficiently cover my new functionality.

After submitting

• All GitHub Actions jobs for my pull request have passed.

[adam2392]

Here is the example rendered: https://output.circle-artifacts.com/output/job/ea0363b2-a2fb-420d-8834-fada0ce23cf4/artifacts/0/dev/auto_examples/intro_causal_graphs.html#sphx-glr-auto-examples-intro-causal-graphs-py

[petergtz]

Hey @adam2392, I've done a very superficial review (given my lack of theoretical background in the science topic) and would expect others to weigh in too.

A general comment: to make these kinds of reviews easier for reviewers, please don't mix different changes into one PR. I think the changes in pag.py and admg.py should not be in this PR (and probably those in pyproject.toml neither).

I know that it's often annoying to fiddle around with different branches, separating different changes into those different branches, etc. But there's one huge benefit I've observed in the past: reviews and especially approvals will happen faster, because there's less to review, and one requested change does not block all the other changes you made. So I'd encourage you to try it out.

[petergtz]

Is "c_components" a well-known expression in the science community or would this benefit from the verbose name "confounded_components".

[adam2392]

I think the string can be more explicit, but in general anyone working in causality from graphical model perspective would get exposed to c-components (imo), but open to changing the property name too.

[adam2392]

I know that it's often annoying to fiddle around with different branches, separating different changes into those different branches, etc. But there's one huge benefit I've observed in the past: reviews and especially approvals will happen faster, because there's less to review, and one requested change does not block all the other changes you made. So I'd encourage you to try it out.

Hey @petergtz sure I can do that. I figured small changes on top of the main contributing example were fine, but I guess they kind of stacked up to a significant # LOC :/.

I'm okay with waiting on this one to merge in. This is an example really trying to communicate core concepts, and I think in general will be one of the first things a user might look at, so def a higher bar.

I'll try to be more cognizant of breaking up the proposed changes more.

[codecov-commenter]

Codecov Report

Merging #11 (08d0d50) into main (8c2ff2e) will decrease coverage by 1.10%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##             main      #11      +/-   ##
==========================================
- Coverage   68.27%   67.16%   -1.11%     
==========================================
  Files          14       14              
  Lines         788      801      +13     
  Branches      208      212       +4     
==========================================
  Hits          538      538              
- Misses        204      217      +13     
  Partials       46       46              

Impacted Files	Coverage Δ
pywhy_graphs/viz/draw.py	0.00% <0.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more