Contributing to datamend

Thank you for your interest in improving datamend. Every contribution matters — whether it is a bug report, a repair strategy, a new drift algorithm, documentation, or a test.

Ways to contribute

Type	What to do
Bug report	Open an issue with a minimal reproducible example
Feature request	Open a discussion before opening a PR
New repair strategy	Implement BaseRepairPlugin and open a PR
New validator	Implement BaseValidatorPlugin and open a PR
New drift detector	Implement BaseDriftDetectorPlugin and open a PR
New tracer	Implement BaseTracerPlugin and open a PR
Documentation	Edit files in docs/ and open a PR
Tests	Add tests in tests/ targeting uncovered code paths

Development setup

git clone https://github.com/vignesh2027/datamend.py.git
cd datamend.py
python -m venv .venv
source .venv/bin/activate         # Windows: .venv\Scripts\activate
pip install -e ".[dev]"

Running the tests

pytest                            # Run all tests with coverage
pytest tests/test_repair.py       # Run a specific test file
pytest -k "test_repair"           # Run tests matching a pattern

Coverage must remain above 90% on all PRs.

Code style

datamend uses ruff for linting and formatting:

ruff check datamend/              # Lint
ruff format datamend/             # Format
mypy datamend/                    # Type check

All public functions, classes, and methods must have Google-style docstrings with Args and Returns sections. All public API must have full type hints.

Writing a repair plugin

from datamend.plugins.base import BaseRepairPlugin
from datamend.core.repair import RepairAction
from typing import List, Tuple
import pandas as pd

class MyRepairPlugin(BaseRepairPlugin):
    name = "my_repair"
    description = "Short description of what this repairs."
    version = "0.1.0"
    author = "Your Name"

    def repair(self, df: pd.DataFrame) -> Tuple[pd.DataFrame, List[RepairAction]]:
        # 1. Copy the DataFrame — never mutate in place
        df = df.copy()
        actions: List[RepairAction] = []

        # 2. Detect the issue
        # 3. Fix the issue
        # 4. Log a RepairAction for every change

        return df, actions

Plugins must:

• Never mutate the input DataFrame in place
• Return a RepairAction for every change they make
• Handle errors gracefully (do not crash the pipeline)
• Have unit tests in tests/

Writing a drift detector plugin

from datamend.plugins.base import BaseDriftDetectorPlugin
from typing import Any, Dict
import pandas as pd

class MyDriftPlugin(BaseDriftDetectorPlugin):
    name = "my_drift"
    description = "Custom drift detector."

    def detect(self, reference: pd.Series, current: pd.Series, column: str) -> Dict[str, Any]:
        drift_score = ...  # your logic
        return {
            "drift_score": drift_score,  # 0–100
            "drifted": drift_score > 20,
            "method": "my_method",
            "details": {},
        }

Submitting a PR

1. Fork the repository and create a branch: git checkout -b feature/my-repair-plugin
2. Write your code, tests, and docstrings
3. Run pytest and confirm coverage is above 90%
4. Run ruff check datamend/ and mypy datamend/
5. Open a PR against main with a clear description of what the change does and why

Publishing a plugin as a separate package

If your plugin is domain-specific (e.g. medical, financial), publish it as a standalone package and declare the datamend entry-point:

[project.entry-points."datamend.plugins"]
my_plugin = "my_package.plugins:MyRepairPlugin"

datamend will auto-discover it when the package is installed.

Code of conduct

Be respectful and constructive. datamend is a community project and everyone is welcome.