cleanoutliers 0.1.1

A small Python library to detect and remove outliers from datasets.

cleanoutliers

A lightweight Python library to detect and remove outlier rows from arrays and DataFrames.

Features

• Remove outliers from NumPy arrays and pandas DataFrames.
• Multiple methods: IQR, Z-score, Modified Z-score.
• Optionally return the outlier mask for inspection.

Installation

pip install cleanoutliers

For pandas support:

pip install "cleanoutliers[pandas]"

Quick Start

import numpy as np
from cleanoutliers import remove_outliers

data = np.array([10, 11, 12, 13, 300])
cleaned = remove_outliers(data, method="iqr", threshold=1.5)
print(cleaned)  # [10. 11. 12. 13.]

With pandas DataFrame

import pandas as pd
from cleanoutliers import remove_outliers

df = pd.DataFrame({"age": [20, 21, 22, 99], "score": [88, 90, 91, 89]})
clean_df = remove_outliers(df, method="iqr", columns=["age"])

API

detect_outliers(data, method="iqr", columns=None, threshold=1.5, z_threshold=3.0)

Returns a boolean mask where True means the row is an outlier.

remove_outliers(data, method="iqr", columns=None, threshold=1.5, z_threshold=3.0, return_mask=False)

Returns cleaned data with outlier rows removed. If return_mask=True, returns (cleaned_data, outlier_mask).

Publish to PyPI

1. Update metadata in pyproject.toml (name, author, email).
2. Install publishing tools:

   pip install -U build twine
   

3. Build package:

   python -m build
   

4. Check distributions:

   twine check dist/*
   

5. Upload to TestPyPI (recommended first):

   twine upload --repository testpypi dist/*
   

6. Upload to PyPI:

   twine upload dist/*
   

Publish with GitHub Trusted Publishing (No API Token)

This project includes GitHub Actions workflows for Trusted Publishing:

• .github/workflows/publish-testpypi.yml
• .github/workflows/publish-pypi.yml

One-time setup on PyPI/TestPyPI

1. Create accounts on PyPI and TestPyPI.
2. In each project settings page, add a Trusted Publisher.
3. Provider: GitHub.
4. Fill in:
   • GitHub owner/user
   • Repository name
   • Workflow filename (publish-testpypi.yml for TestPyPI, publish-pypi.yml for PyPI)
   • Environment: leave empty unless you use one

Release flow

1. Commit and push your changes to GitHub.
2. For a TestPyPI pre-release, create a release-candidate tag:

   git tag v0.1.0-rc1
   git push origin v0.1.0-rc1
   

3. For a PyPI release, create a stable tag:

   git tag v0.1.0
   git push origin v0.1.0
   

The corresponding GitHub workflow will build and publish automatically using OIDC.

License

MIT