test: Add predict and predict_proba tests for GWLogisticRegression

[jigyasaba]

Summary:

Adds basic test coverage for predict and predict_proba in GWLogisticRegression

Changes:

• Added test for predict
• checks output length
• validates binary predictions (ignoring missing values)
• Added test for predict_proba
• checks output shape (n_samples, 2)
• Added test for keep_models=False
• ensures prediction fails when local models are not stored

Notes:

• geometry is required for prediction and is included in tests
• Some predictions may be NaN, which is handled in assertions

Testing:

• All tests pass locally (pytest)
• Code formatted with ruff
• Checklist

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.35%. Comparing base (a687b09) to head (dcd2b56).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #116   +/-   ##
=======================================
  Coverage   92.35%   92.35%           
=======================================
  Files           6        6           
  Lines         876      876           
=======================================
  Hits          809      809           
  Misses         67       67           

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

🚀 New features to boost your workflow:

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

[jigyasaba]

pre-commit.ci autofix

[martinfleis]

Thanks, though I am not sure this is needed. We test both predict and predict_proba on base models and subclasses do not alter that code in any way. I would rather try to find a way how to trim our test suite to make it quicker, than adding more models to be trained.

[jigyasaba]

You're right that predict and predict_proba are already tested at the base model level.
My intention here was to ensure that the prediction workflow works correctly when used through GWLogisticRegression, particularly with spatial inputs like geometry and the keep_models setting.

That said, I understand the concern about test suite runtime and redundancy. I'm happy to reduce or simplify these tests
Please let me know your preference, and I can update the PR accordingly.

[martinfleis]

I would just not do anything as these features are already covered by existing tests.