Heuristic filter feature selection methods for medical datasets.

Area of Science:

• Bioinformatics
• Computational Biology
• Machine Learning

Background:

• Gene selection is crucial for analyzing high-dimensional datasets, common in genomics.
• Effective gene selection aims to improve classification accuracy and computational efficiency.
• Existing methods may not optimally handle datasets with many features and few samples.

Purpose of the Study:

• To propose and evaluate two novel heuristic gene selection methods: Xvariance and Mutual Congestion.
• To assess the performance of these methods on binary medical datasets.
• To identify the most effective method for different data complexities.

Main Methods:

• Developed the Xvariance method, focusing on internal feature attributes for classification.
• Developed the Mutual Congestion method, utilizing a frequency-based approach.
• Applied both methods to eight diverse binary medical datasets for comparative analysis.

Main Results:

• Xvariance demonstrated effectiveness on standard datasets.
• Mutual Congestion significantly improved classification accuracy on high-dimensional datasets.
• The performance varied based on dataset characteristics, highlighting method-specific advantages.

Conclusions:

• Mutual Congestion offers a substantial improvement for gene selection in high-dimensional medical data.
• The choice of gene selection method should consider the specific characteristics of the dataset.
• These heuristic approaches provide valuable tools for feature selection in bioinformatics.