Frequency based feature selection method using whale algorithm.

Area of Science:

• Computational biology
• Machine learning
• Data science

Background:

• Feature selection is crucial for high-dimensional datasets, particularly in medical applications.
• Identifying the most impactful features improves predictive model performance and interpretability.

Purpose of the Study:

• To propose a novel filter feature selection method for high-dimensional binary medical datasets.
• To evaluate the effectiveness of combining the whale algorithm with Mutual Congestion and majority voting.

Main Methods:

• A three-stage filter feature selection approach was developed.
• Stage 1: Whale algorithm for irrelevant feature removal.
• Stage 2: Mutual Congestion for feature ranking, followed by forward feature selection with majority voting.

Main Results:

• Mutual Congestion demonstrated strong predictive capability on its own.
• The whale algorithm enhanced Mutual Congestion's accuracy in most cases.
• The proposed method achieved improved prediction with a reduced feature subset compared to existing methods.

Conclusions:

• The proposed hybrid feature selection method is effective for high-dimensional medical data.
• This approach offers a more efficient and accurate way to select relevant features for classification tasks.