Metadata-Version: 2.4
Name: ez-automl-lite
Version: 0.1.0b1
Summary: A lightweight AutoML library optimized for simple APIs and ONNX export.
Author-email: Cristopher Coronado <cgcoronadom@gmail.com>
License: MIT
Project-URL: Homepage, https://github.com/cristofima/auto-ml-lite
Project-URL: Issues, https://github.com/cristofima/auto-ml-lite/issues
Project-URL: Repository, https://github.com/cristofima/auto-ml-lite
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.11
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: flaml[automl]>=2.3.6
Requires-Dist: pandas>=2.2.0
Requires-Dist: numpy>=1.26.0
Requires-Dist: scikit-learn>=1.4.0
Requires-Dist: feature-engine>=1.8.1
Requires-Dist: joblib>=1.3.2
Requires-Dist: lightgbm>=4.3.0
Provides-Extra: onnx
Requires-Dist: skl2onnx>=1.17.0; extra == "onnx"
Requires-Dist: onnx>=1.16.0; extra == "onnx"
Requires-Dist: onnxruntime>=1.18.0; extra == "onnx"
Requires-Dist: onnxmltools>=1.12.0; extra == "onnx"
Provides-Extra: reports
Requires-Dist: sweetviz>=2.3.1; extra == "reports"
Provides-Extra: all
Requires-Dist: ez-automl-lite[onnx,reports]; extra == "all"
Dynamic: license-file

# 🚀 ez-automl-lite

A lightweight, serverless-optimized AutoML library for Python. Build, evaluate, and export high-performance machine learning models with just 3 lines of code.

[![Python Version](https://img.shields.io/badge/python-3.11+-blue.svg)](https://www.python.org/downloads/)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

## ✨ Features

- **3-Line API**: Designed for simplicity and speed across 4 different ML tasks.
- **Serverless-First**: Optimized for AWS Lambda/Azure Functions and low-memory environments.
- **Premium Reports**: Professional HTML/CSS reports for all tasks (No external JS or Internet required).
- **Comprehensive Analytics**: Supports Regression, Classification, Clustering, and Anomaly Detection.
- **ONNX Export**: One-click export for cross-platform deployment.

---

## 📦 Installation

```bash
pip install "ez-automl-lite[all]"
```

---

## 🚀 The 4 Core Modules

### 1. Regression
Automated training with residual analysis and error diagnostics.
```python
from ez_automl_lite import AutoML
aml = AutoML(target="target").fit(df)
aml.report("regression_report.html")
```

### 2. Classification
Visual Confusion Matrices and detailed class-wise performance metrics.
```python
from ez_automl_lite import AutoML
aml = AutoML(target="label").fit(df)
aml.report("classification_report.html")
```

### 3. Clustering (Unsupervised)
Automated optimal K-search using Silhouette and Calinski-Harabasz scores.
```python
from ez_automl_lite import AutoCluster
ac = AutoCluster(max_clusters=8).fit(df)
ac.report("clustering_report.html")
```

### 4. Anomaly Detection
Profile-based detection using Isolation Forest with detailed sample analysis.
```python
from ez_automl_lite import AutoAnomaly
aa = AutoAnomaly(contamination=0.05).fit(df)
aa.report("anomaly_report.html")
```

---

## 📂 Examples & Scripts

Check the `examples/` directory for full implementation scripts:
- `examples/regression_example.py`
- `examples/classification_example.py`
- `examples/clustering_example.py`
- `examples/anomaly_example.py`

---

## 🛠️ Performance & Export

- **ONNX Export**: Cross-platform models in one line: `aml.export_onnx("model.onnx")`.
- **EDA**: Generate pre-training analysis: `aml.eda(df, "eda.html")`.
- **UUIDs**: Every training session generates a unique ID for easy tracking.

---

## 🗺️ Roadmap

- [x] Core Package Refactor
- [x] Premium CSS-only Reports
- [x] AutoCluster & AutoAnomaly implementation
- [x] Cross-platform ONNX support
- [ ] PyPI Automated Release Workflow

---

## 🤝 Contributing & License

Created by [Cristopher Coronado](https://github.com/cristofima). 
Distributed under the MIT License.
