Metadata-Version: 2.4
Name: ethos_ai_guardian
Version: 0.1.0
Summary: Ethos AI Guardian: Automated bias detection and mitigation for machine learning models.
Author: Ethos AI Team
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Intended Audience :: Science/Research
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: pandas
Requires-Dist: numpy
Requires-Dist: torch
Requires-Dist: scikit-learn
Dynamic: author
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: license-file
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# Ethos AI Guardian

Ethos AI Guardian is a powerful, automated bias detection and mitigation toolkit for machine learning models. Built specifically for data scientists, it provides a seamless pipeline to identify historical, confounding, and selection biases in your datasets, and mitigates them using state-of-the-art techniques.

## Features

- **Automated Bias Detection:** Quickly scan datasets for Statistical Parity, Disparate Impact, and Equal Opportunity differences.
- **Deep Contextual Debiasing:** Pro-level Tabular Transformers to understand contextual bias in categorical and numerical features.
- **Adversarial Mitigation:** Uses an adversarial debiasing architecture to enforce fairness without sacrificing predictive power.
- **Pipelines:** Easy-to-use Scikit-Learn style pipelines for integrating into your existing workflow.

## Installation

You can install `ethos_ai_guardian` using pip:

```bash
pip install ethos_ai_guardian
```

## Quick Start

```python
import pandas as pd
from ethos_ai_guardian import AntiBiasPipeline

# Load your dataset
df = pd.read_csv("your_data.csv")

# Initialize the pipeline
pipeline = AntiBiasPipeline(model_type='logistic')

# Run automated bias mitigation
results = pipeline.train_and_evaluate(
    df=df, 
    features=['age', 'income', 'education'], 
    sensitive_attr='gender', 
    target_col='approved'
)

print(results["fairness_metrics"])
```

## License

MIT License
