Add examples for GWLinearRegression and GWLogisticRegression

[aman-coder03]

Summary

Closes #85

This PR adds two new example notebooks to the User Guide demonstrating complete workflows for GWLinearRegression and GWLogisticRegression.

Changes

• docs/source/gw_linear_regression_example.ipynb - new example notebook
• docs/source/gw_logistic_regression_example.ipynb - new example notebook
• docs/source/index.md - added both notebooks to the User Guide toctree

What the examples cover

GWLinearRegression example:

• Real-world dataset: US per-capita income by state, 1929-2009 (libpysal us_income)
• Research question: does the income convergence effect vary spatially across US states?
• Global OLS baseline for comparison
• Spatially varying coefficients (local_coef_) mapped and interpreted by region
• Local R² (local_r2_) and residual (resid_) analysis

GWLogisticRegression example:

• Creating a binary target from a continuous variable
• Fitting the model and checking prediction_rate_
• Focal accuracy and pooled metrics (y_pooled_, pred_pooled_)
• Per-model metrics via local_metric
• Mapping local log-odds coefficients (local_coef_)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.01%. Comparing base (5b8382d) to head (aed1f6d).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #87      +/-   ##
==========================================
+ Coverage   91.36%   92.01%   +0.65%     
==========================================
  Files           6        6              
  Lines         799      814      +15     
==========================================
+ Hits          730      749      +19     
+ Misses         69       65       -4     

☔ 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.

[martinfleis]

I was anticipating something else. This is mirroring what is already in the user guide, just reshuffled. I was more hoping for an example on how to use gwlearn on a real-world case, something like we have in momepy. This still uses the default example dataset and does not add anything on top of what the documentation already covers.

[aman-coder03]

Thank you for the feedback @martinfleis
Could you point me toward the kind of real-world dataset and question you had in mind? For example,
A publicly available spatial dataset(from census, OpenStreetMap, or maybe a PySAL dataset)
A meaningful research question(predicting house prices, exploring income inequality spatially)
This would help me write an example that demonstrates gwlearn solving a genuine analytical problem rather than repeating the tutorial dataset

[martinfleis]

No. If you want to contribute an example I would expect it comes from you. It takes time to come up with a good example, I don't know what is going to work well.

[aman-coder03]

@martinfleis i have replaced the linear regression example with a real world case study using US state income data(1929-2009) investigating the income convergence hypothesis. I will work on coming up with a similarly meaningful real world case for the logistic regression example as well

[aman-coder03]

I explored the Tokyo mortality dataset from libpysal - 262 municipalities with socioeconomic variables(unemployment, home ownership, elderly population, technical occupations) and observed vs expected deaths for ages 25-64. I am planning to use it to classify municipalities as high/low mortality(observed > expected, 40/60 split) and examine whether the socioeconomic predictors of high mortality risk vary spatially across the city

[aman-coder03]

@martinfleis can you please have a look and suggest changes if needed?

[martinfleis]

The fact that I haven't responded two days, which also happened to be a weekend made you think it is fine to remind me?

[aman-coder03]

you are right, I apologize. I should have waited longer before pinging