A hybrid feature selection model based on improved squirrel search algorithm and rank aggregation using fuzzy...

Area of Science:

• Biomedical data analysis
• Machine learning
• Computational biology

Background:

• Feature selection is crucial for handling large biomedical datasets.
• Existing methods like filters, wrappers, and embedded approaches have limitations (computational complexity, performance, classifier dependency).
• Hybrid approaches offer a balance but often suffer from unstable and biased filtering metrics, especially with diverse biomedical data types.

Purpose of the Study:

• To propose a stable filtering method for hybrid feature selection in biomedical datasets.
• To address the instability and metric bias issues in traditional filtering techniques.
• To improve classification accuracy and computational efficiency for biomedical data analysis.

Main Methods:

• Developed a hybrid feature selection model incorporating a stable filtering method based on rank aggregation.
• Integrated the Improved Squirrel Search Algorithm to optimize the feature selection process.
• Evaluated the model's performance on nine diverse biomedical datasets using three different classifiers.

Main Results:

• The proposed model demonstrated superior performance compared to well-established and state-of-the-art feature selection methods.
• Achieved significant improvements in classification accuracy across multiple biomedical datasets.
• Showcased reduced computational time, indicating enhanced efficiency.

Conclusions:

• The novel rank aggregation-based filtering method provides a stable and effective approach for hybrid feature selection in biomedical contexts.
• The Improved Squirrel Search Algorithm integration further optimizes the selection process, leading to better predictive model performance.
• The model's robustness is validated across various datasets and classifiers, highlighting its applicability in real-world biomedical research.