Fixed propensity example

[FrancescMartiEscofetQC]

As explained in #54 we want to support some method of passing "fixed" propensity scores. This PR does not change any in the implementation but rather shows how one can do it with the existing implementation creating a custom model.

Rendered version

Checklist

• Added a CHANGELOG.rst entry

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 94.88%. Comparing base (a1ee27c) to head (f620c69).
Report is 19 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #70   +/-   ##
=======================================
  Coverage   94.88%   94.88%           
=======================================
  Files          15       15           
  Lines        1544     1544           
=======================================
  Hits         1465     1465           
  Misses         79       79           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[kklein]

Sweet - thanks!

[kklein]

I think I got the point but it might be worth replacing the pronoun 'this' with the subject you are referring to reduce the likelihood of a misunderstanding. :)

[FrancescMartiEscofetQC]

1b77bdc

[kklein]

FYI @ArseniyZvyagintsevQC @MatthiasLoefflerQC :)

[kklein]

Thank you!

[apoorvalal]

Would using DummyClassifier also work? With strategy prior, this spits out the empirical propensity score (so this isn't strictly a 'known' pscore estimator but in moderate-to-large samples will be basically accurate).

The econml analogue works well.

[kklein]

Absolutely - great point!