Jmafoster1/visualisation

[jmafoster1]

This PR adds a visualisation dashboard to the CTF to serve as a nice pretty GUI. (Sorry it ended up being a bit of a monster PR!)

Closes #411

If you want to see the dashboard in action, check out this branch, install the dependencies to a virtual environment of your choice, and run causal-testing dashboard. You can then navigate to https://localhost:5006 and "upload" use dag/data from examples.

Main changes

• Moved visualisation code out of abstract_discovery.py and into a separate dedicated visualisation package
• Expanded this out into an entire visualisation dashboard package
• Moved effect_direction into CausalTestResult class. Ideally, this would be part of EffectEstimate, but that leads to circular imports with CausalEffect :(
• Bug fix on NoEffect estimate and making the other causal effects more consistent and robust
• Renamed EffectEstimate.value to EffectEstimate.effect_estimate (this touches a lot of files)
• Visualisation and dashboard tutorial.

[github-actions]

🦙 MegaLinter status: ⚠️ WARNING

Descriptor	Linter	Files	Fixed	Errors	Elapsed time
⚠️ PYTHON	black	35		1	1.11s
✅ PYTHON	pylint	35		0	6.34s

See detailed report in MegaLinter reports

MegaLinter is graciously provided by

[jmafoster1]

Bug fix: This should have been .any() all along

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.00%. Comparing base (b7c8e1e) to head (cb3f170).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #403      +/-   ##
==========================================
- Coverage   98.06%   98.00%   -0.07%     
==========================================
  Files          26       29       +3     
  Lines        1605     1901     +296     
==========================================
+ Hits         1574     1863     +289     
- Misses         31       38       +7     

Files with missing lines	Coverage Δ
causal_testing/__main__.py	99.31% <100.00%> (+0.06%)	⬆️
causal_testing/causal_testing_framework.py	100.00% <100.00%> (ø)
causal_testing/discovery/abstract_discovery.py	98.38% <100.00%> (-0.67%)	⬇️
causal_testing/discovery/hill_climber_discovery.py	90.19% <100.00%> (ø)
causal_testing/estimation/effect_estimate.py	100.00% <100.00%> (ø)
causal_testing/testing/causal_effect.py	97.56% <100.00%> (ø)
causal_testing/testing/causal_test_case.py	100.00% <100.00%> (ø)
causal_testing/testing/causal_test_result.py	100.00% <100.00%> (ø)
causal_testing/testing/data_adequacy.py	100.00% <100.00%> (ø)
causal_testing/visualisation/geometry.py	100.00% <100.00%> (ø)
... and 2 more

... and 1 file with indirect coverage changes

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 373a413...cb3f170. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jmafoster1]

Right, this should be fine now. Patch coverage is 100%, but Codecov sometimes gets confused when you move/delete code that was previously covered, which is what's happened here.

[jmafoster1]

TODO: Double check the kurtosis interpretation ("suspiciously stable"/"suspiciously unstable") before merging

See: https://fastercapital.com/content/Kurtosis--Kurtosis--Peeking-into-the-Tails-of-Normal-Distribution.html#Leptokurtic--Mesokurtic--and-Platykurtic-Distributions

EDIT: I think it's fine as is