v1.1.0

Changelog

[1.1.0] - 2024-07-23

---

Added

• New Method: rfe_brute_force
  • Implemented the rfe_brute_force method for exhaustive feature elimination.
  • This method iteratively adds features based on Recursive Feature Elimination (RFE) and evaluates model performance to find the best feature subset.
  • Example usage:

    best_features = fs.rfe_brute_force(estimator=RandomForestClassifier(), n_features_to_select=5, force=True)
    print("Best Features:", best_features)

Changed

• Backward Elimination and Forward Selection Enhancements
  • Added a fast parameter to both backward_elimination and forward_selection methods.
  • When fast=True, these methods use a quicker evaluation process, reducing computational time.
  • Example usage:

    selected_features = fs.backward_elimination(significance_level=0.05, fast=True)
    print("Selected Features (Fast Backward Elimination):", selected_features)

  • Example usage for forward selection:

    selected_features = fs.forward_selection(significance_level=0.05, fast=True)
    print("Selected Features (Fast Forward Selection):", selected_features)

Fixed

• Filter Method Improvements
  • Improved the filter_method to raise appropriate errors when using chi2 for regression problems or anova for classification problems.
  • Ensured better handling of data types and target variable compatibility.