tau-community-detection 0.2.5

Community detection via Louvain/Leiden + Genetic Algorithm

TAU: Parallel Genetic Clustering with Louvain and Leiden

TAU is a high-performance Python package for modularity-based community detection using a hybrid of genetic algorithms and graph clustering methods (Louvain and Leiden). It is built for scalability, parallelism, and usability across research and applied machine learning settings.

---

Features

• Genetic Initialization: Enhances clustering quality by optimizing initial conditions with evolutionary search.
• Pluggable Clustering Engines: Supports Louvain and Leiden algorithms.
• Parallel Execution: Uses Python's multiprocessing/threading for speed-up across CPUs.
• Flexible Graph Input: Read from adjacency lists, edge lists, CSVs, or pandas DataFrames.
• CLI and API Access: Run from the terminal or call programmatically in Python.
• Reproducible Runs: Optional random seed parameter for repeatable experiments.

---

Installation

From PyPI

pip install community_TAU

From Source

git clone https://github.com/HillelCharbit/community_TAU.git
cd community_TAU
pip install .

---

Quick Start

Command-Line Interface

python -m community_TAU --graph data/example.graph --size 80 --workers 4 --max_generations 300

Command-Line Arguments

Argument	Description	Default
--graph	Path to graph file (adjacency list, edge list, etc.)	Required
--size	Population size for genetic algorithm	60
--workers	Number of parallel workers	All available cores
--max_generations	Maximum number of generations	500
--seed	Random seed (optional)	None

---

Python API Example

from community_TAU import community_TAU
import networkx as nx

G = nx.read_edgelist("example.graph")
result = community_TAU(G, size=100, workers=8, max_generations=400)

# Access results
print(result.partition)
print(result.modularity)

---

Input Formats Supported

• Adjacency list (.graph)
• Edge list (.csv, .txt)
• Adjacency matrix (CSV or DataFrame)

Conversion tools are available in the graph_loader module.

---

Output

• Final partition (dictionary of node → community)
• Final modularity score
• Optional generation logs and fitness history

---

Testing

To run unit tests:

pytest tests/

Test coverage includes input parsing, genetic algorithm logic, clustering evaluation, and parallel execution.

---

Contributing

We welcome contributions! To get started:

1. Fork this repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Make your changes and commit (git commit -am 'Add new feature')
4. Push to the branch (git push origin feature/my-feature)
5. Open a pull request

Please review the contributing guidelines before submitting changes.

---

License

This project is licensed under the MIT License.

---

Acknowledgments

• Louvain and Leiden algorithms based on work by Blondel et al. and Traag et al.
• Graph handling inspired by NetworkX and igraph interfaces.

---

Project Status

TAU is actively maintained and under continuous development. Feedback and issues are welcome on the GitHub issue tracker.