BCOVIDOA: A Novel Binary Coronavirus Disease Optimization Algorithm for Feature Selection.

Area of Science:

• Computer Science
• Artificial Intelligence
• Data Science

Background:

• The proliferation of digital devices generates massive datasets, presenting challenges in dimensionality, redundancy, and irrelevance.
• Feature selection is crucial for managing big data by identifying optimal feature subsets.
• Existing feature selection methods struggle with the complexity of large-scale, high-dimensional data.

Purpose of the Study:

• To introduce a novel Binary Coronavirus Disease Optimization Algorithm (BCOVIDOA) for effective feature selection.
• To address the inherent challenges of big data, including dimensionality and irrelevance.
• To evaluate the performance of BCOVIDOA against established feature selection techniques.

Main Methods:

• Development of the Binary Coronavirus Disease Optimization Algorithm (BCOVIDOA), inspired by the replication mechanism of the Coronavirus.
• Evaluation of BCOVIDOA on twenty-six benchmark datasets from the UCI Repository.
• Comparative analysis against nine recent wrapper feature selection algorithms.

Main Results:

• BCOVIDOA demonstrated superior performance across key metrics: accuracy, best cost, average cost (AVG), and standard deviation (STD).
• The algorithm achieved a significant reduction in the size of selected features compared to existing methods.
• Statistical validation using the Wilcoxon rank-sum test confirmed the significance of the results.

Conclusions:

• The proposed BCOVIDOA offers a highly effective and statistically significant approach to feature selection in big data.
• BCOVIDOA provides a promising new tool for optimizing feature subsets, enhancing data analysis and model performance.
• The algorithm's biological inspiration offers a novel perspective on optimization techniques for data science challenges.