Comprehensive examples and best practices for testing data science code with pytest.

pytest/
├── README.md                    # This file
├── get_started/                 # Basic pytest concepts
│   └── sentiment.py
├── parametrization/            # Parametrized testing
│   ├── process.py
│   ├── process_fixture.py
│   └── sentiment.py
├── test_structure_example/     # Project organization
│   ├── src/
│   └── tests/
├── advanced_fixtures/          # Advanced fixture patterns
│   ├── session_scoped.py
│   ├── autouse_fixtures.py
│   ├── conftest.py
│   └── README.md
├── temporary_files/           # Safe file I/O testing
│   ├── file_operations.py
│   ├── data_pipeline.py
│   └── README.md
├── numerical_testing/         # NumPy/DataFrame testing
│   ├── numpy_arrays.py
│   ├── dataframe_testing.py
│   └── README.md
├── mocking/                   # External dependency mocking
│   ├── api_mocking.py
│   ├── database_mocking.py
│   ├── requirements.txt
│   └── README.md
├── custom_markers/            # Test organization with markers
│   ├── pytest.ini
│   ├── marked_tests.py
│   └── README.md
└── project_config/           # Complete project configuration
    ├── pytest.ini
    ├── conftest.py
    ├── test_with_fixtures.py
    └── README.md

pip install pytest

# For advanced features
pip install pytest-cov pytest-xdist pytest-benchmark

# Basic examples
pytest get_started/
pytest parametrization/

# Advanced features
pytest advanced_fixtures/
pytest numerical_testing/
pytest mocking/

# Full project configuration
cd project_config && pytest

• Simple test functions and assertions
• Parametrized tests for multiple test cases
• Basic fixtures for data reuse

• Session-scoped fixtures: Load expensive datasets once
• Autouse fixtures: Automatic setup for all tests
• Shared fixtures: Common test data via conftest.py

• tmp_path fixture: Isolated temporary directories
• File I/O testing: CSV, JSON, model serialization
• Pipeline testing: End-to-end data processing

• NumPy arrays: Floating-point comparison with tolerance
• Pandas DataFrames: Proper DataFrame equality testing
• Statistical validation: Testing model outputs and data properties

• API mocking: Test without hitting real APIs
• Database mocking: Test queries without databases
• Error simulation: Test failure scenarios safely

• Test organization: Group tests by speed, requirements, domain
• Selective execution: Run specific test categories
• CI/CD integration: Different test suites for different stages

• Complete setup: Production-ready pytest configuration
• Centralized fixtures: Project-wide test utilities
• Best practices: Logging, warnings, reproducibility

# Fast feedback during development
pytest -m fast

# Before committing changes
pytest -m "fast or (integration and not slow)"

# Full test suite
pytest

# Unit tests (fast feedback)
pytest -m "unit and fast"

# Integration tests  
pytest -m "integration and not gpu and not expensive"

# Performance tests (separate stage)
pytest -m "slow or expensive"

# Test data processing pipelines
pytest -m data_processing

# Test model training
pytest -m model_training

# Test without external dependencies
pytest -m "not api and not database"

• Reproducible results: Consistent random seeds
• Isolated tests: No interference between tests
• Proper numerical comparison: Handle floating-point precision

• Fast feedback: Separate fast/slow test categories
• Efficient fixtures: Load expensive data once
• Parallel execution: Run tests concurrently

• Clear error messages: Detailed assertion information
• Test organization: Easy to find and run specific tests
• Comprehensive logging: Track test execution

• Standardized setup: Consistent test environment
• Shared fixtures: Common test data and utilities
• Documentation: Clear examples and best practices

def test_data_cleaning(tmp_path):
    # Use temporary files for safe testing
    input_file = tmp_path / "dirty_data.csv"
    # Test cleaning pipeline...

@pytest.fixture(scope="session")
def trained_model():
    # Train once, test many aspects
    return expensive_model_training()

def test_model_accuracy(trained_model):
    # Test with proper numerical comparison
    assert model.accuracy > 0.9

@patch('requests.get')
def test_api_integration(mock_get):
    # Mock external calls for reliable testing
    mock_get.return_value.json.return_value = {'data': 'test'}
    # Test your logic...

Each directory contains detailed README files with:

• Specific feature documentation
• Running instructions
• Best practices
• Troubleshooting guides

Start with the examples that match your current testing needs, then explore advanced features as your test suite grows.

• Article: Pytest for Data Scientists
• Video Series: YouTube Playlist
• Official Docs: pytest.org

These examples are designed to be practical and educational. Feel free to adapt them for your specific data science testing needs.