Develop and Test PhilanthroPy

To ensure the reliability of donor analytics models, PhilanthroPy maintains a rigorous testing suite that mirrors our GitHub CI environment. Contributors and maintainers should run these tests locally before proposing changes.

Prerequisites

Ensure you have the development dependencies installed:

pip install -e ".[dev]"

Running Tests Locally

PhilanthroPy uses pytest alongside hypothesis for property-based testing and pytest-cov for coverage analysis.

1. High-Level Unit Tests & Coverage

Run the core test suite and verify that code coverage remains above 85%:

pytest tests/ --cov=philanthropy --cov-fail-under=85

2. Scikit-Learn API Compliance

Verify that all transformers and estimators strictly adhere to the scikit-learn API:

pytest tests/test_sklearn_compat.py -v

3. Property-Based Testing

Verify the mathematical robustness of transformers using randomized data generation:

pytest tests/test_transformers_property.py -v --hypothesis-seed=12345

4. Running Doctests

Verify that all code examples in docstrings are functional:

pytest philanthropy --doctest-modules