uEFS: An efficient and comprehensive ensemble-based feature selection methodology to select informative features.

Area of Science:

• Machine Learning
• Data Science
• Bioinformatics

Background:

• Feature selection is crucial for identifying relevant data features.
• Existing methods for feature ranking and filtering have limitations.
• Optimal cutoff value selection for feature importance remains a challenge.

Purpose of the Study:

• To propose an efficient and comprehensive univariate ensemble-based feature selection (uEFS) methodology.
• To address limitations in current feature selection techniques.
• To improve the selection of informative features for classifier construction.

Main Methods:

• Developed a unified features scoring (UFS) algorithm for comprehensive feature evaluation and ranking.
• Introduced a threshold value selection (TVS) algorithm to identify optimal cutoff points for filtering irrelevant features.
• Evaluated the uEFS methodology on standard benchmark datasets.

Main Results:

• The uEFS methodology demonstrated competitive accuracy.
• Achieved an average increase of approximately 7% in f-measure.
• Showed an average increase of approximately 5% in predictive accuracy compared to state-of-the-art methods.

Conclusions:

• The proposed uEFS methodology offers an efficient and effective approach to feature selection.
• UFS and TVS algorithms provide a robust mechanism for ranking and filtering features.
• The method significantly enhances classifier performance, outperforming existing techniques.