ficaria 0.1.0

Feature selection and missing data imputation

Fuzzy Imputation and Critical Attribute Reduction for Intelligent Analysis

About The Package • Prerequisites • Setup • Usage • License • Authors

🔷 About The Package

The ficaria package is a Python package providing custom, scikit-learn–compatible transformers for data imputation and feature selection. The transformers are designed to integrate seamlessly with scikit-learn pipelines, making them easy to use in real-world machine learning workflows and straightforward to extend for custom or research-oriented use cases.

The package was developed as part of a Bachelor’s degree thesis at the Warsaw University of Technology, Faculty of Mathematics and Information Science. All implemented methods are fuzzy-based, leveraging concepts from fuzzy set theory to handle uncertainty, vagueness, and incomplete data in a principled and interpretable manner. This makes ficaria particularly suitable for datasets where classical crisp methods may be insufficient or overly restrictive.

⚙️ Prerequisites

The ficaria package depends on the following Python libraries:

• NumPy
• Pandas
• SciPy
• scikit-learn
• kneed

🛠 Setup

Ficaria can be installed from PyPI:

pip install ficaria

All dependencies are automatically installed when installing the package via pip.

🚀 Usage

Ficaria provides scikit-learn–compatible transformers for data imputation and feature selection. All transformers implement the standard fit / transform interface, so they can be used directly in pipelines alongside scalers, estimators, and cross-validation tools.

Example 1 — Feature Selection with FuzzyGranularitySelector

from ficaria import FuzzyGranularitySelector

selector = FuzzyGranularitySelector(n_features=5, eps=0.3)
selector.fit(X_train, y_train)
X_reduced = selector.transform(X_test)

Example 2 — Data Imputation with FCMKIterativeImputer

from ficaria import FCMKIterativeImputer

pipeline.fit(X_train, y_train)
X_transformed = pipeline.transform(X_test)

Example 3 — Combining Transformers in a Pipeline

Since all transformers implement fit and transform, they can be combined:

from sklearn.pipeline import Pipeline
from ficaria import FuzzyGranularitySelector, FCMKIterativeImputer

pipeline = Pipeline([
    ("imputer", FCMKIterativeImputer()),
    ("selector", FuzzyGranularitySelector(n_features=5, eps=0.3)),
])

pipeline.fit(X_train, y_train)
X_final = pipeline.transform(X_test)

📄 License

This project is licensed under the MIT License. See the LICENSE file for details.

👥 Authors

Aleksandra Kwiatkowska
Email: [e-mail]
Github: @kwiatkowskaa

Małgorzata Mokwa
Email: [e-mail]
Github: @malgosiam2

Bogumiła Okrojek
Email: [e-mail]
Github: @szostkawron