:bulb: Overview

PyCCEA is an open-source package developed as part of ongoing doctoral research. It provides cooperative co-evolutionary strategies tailored for feature selection in large-scale and high-dimensional problems. The framework adopts a modular, decomposition-based approach and is intended for researchers and practitioners tackling complex feature selection tasks.

Note: PyCCEA is a work in progress. Stay tuned for improvements and new algorithm implementations.

:computer: Installation

To install the PyCCEA package directly from PyPI, use the following command in a Python ≥ 3.10 environment:

pip install pyccea

Alternatively, if you want to install the latest version directly from the GitHub:

pip install git+https://github.com/pedbrgs/pyccea.git

Ensure you have pip and an active internet connection to download dependencies.

:high_brightness: Quickstart

This quickstart demonstrates how to use the CCFSRFG1 algorithm — a CCEA variant with random feature grouping — to perform feature selection on the Wisconsin Diagnostic Breast Cancer (WDBC) dataset.

In this example, you will:

• Load the dataset using the DataLoader utility.
• Configure the dataset and algorithm from .toml files.
• Run the optimization process.

```python import toml import importlib.resources from pyccea.coevolution import CCFSRFG1 from pyccea.utils.datasets import DataLoader

Load dataset parameters

with importlib.resources.opentext("pyccea.parameters", "dataloader.toml") as tomlfile: dataconf = toml.load(tomlfile)

Initialize the DataLoader with the specified dataset and configuration

data = DataLoader(dataset="wdbc", conf=data_conf)

Prepare the dataset for the algorithm (e.g., preprocessing, splitting)

data.get_ready()

Load algorithm-specific parameters

with importlib.resources.opentext("pyccea.parameters", "ccfsrfg.toml") as tomlfile: cceaconf = toml.load(tomlfile)

Initialize the cooperative co-evolutionary algorithm

ccea = CCFSRFG1(data=data, conf=ccea_conf, verbose=False)

Start the optimization process

ccea.optimize() ```

The best feature subset found is stored in the attribute best_context_vector, a binary array where 1 indicates a selected feature and 0 indicates an unselected one.

:books: Documentation

Full documentation, including a comprehensive user guide, step-by-step tutorials, an API reference, and contribution guidelines, is available at PyCCEA docs.

:scroll: Citation info

If you are using these codes in any way, please cite the following paper:

@article{PyCCEA, title = {PyCCEA: A Python package of cooperative co-evolutionary algorithms for feature selection in high-dimensional data}, author = {Venancio, Pedro Vinicius A. B. and Batista, Lucas S.}, journal = {Journal of Open Source Software}, volume = {10}, number = {112}, pages = {8348}, year = {2025} }

:mailbox: Contact

Please send any bug reports, questions or suggestions directly in the repository.