feseco 0.4.0

Feature Selection Companion

A consolidated feature selection decision support system (DSS) that employs variance thresholding, missing values ratios, and pairwise correlations.

Link to Web-Based Platform

How to Use

Installation

In your terminal, simply type:

pip install feseco

Importing the Package

The class that you will need to import is CompanionFS which stands for "Companion - Feature Selector".

from feseco import CompanionFS

Example Usage

For example, you have a dataset and you want to identify which features to drop. You can do this simply by calling the filter_features() method of the CompanionFS object.

from feseco import CompanionFS

file_path = "your_dataset.csv"

# You pass the path to your dataset. Optionally, you can pass a feature of the dataset as an argument. This way, the class analyzes the features relative to your target column.
cfs = CompanionFS(file_path, target_col="my_feature")

features_to_drop = cfs.filter_features()

Optionally, you can change the thresholds for each of the feature selection method. See the example below:

cfs = CompanionFS(
    file_path, 
    var_threshold=0.1,      # variance threshold; default is 0.1
    mvr_threshold=0.5,      # missing value ratio threshold; default is 0.5
    corr_threshold=0.75,    # correlation analysis threshold; default is 0.75
    target_col="my_feature"
)

License

This project is licensed under the MIT License.

View the LICENSE

Attribution Requirements

When building upon or reusing this work, please include:

• Author name: Jasper Gomez
• Repository name: Feature Selection Companion
• Repository link: https://github.com/jasper-gomez/feature-selection-companion/tree/refactor/package
• Citation using the CITATION.cff metadata
• A note in your documentation, publication, or code comments acknowledging the original work

Example acknowledgment:

“This work builds on Feature Selection Companion by Jasper Gomez (MIT License).”

Following these guidelines helps support academic and ethical reuse.