Feature Selection Using Correlation Analysis and Principal Component Analysis for Accurate Breast Cancer Diagnosis

Area of Science:

• Oncology
• Machine Learning
• Data Science

Background:

• Breast cancer remains a leading cause of mortality in women, necessitating improved diagnostic accuracy.
• The complexity of breast cancer and evolving treatment protocols challenge precise diagnosis.
• Enhanced diagnostic tools are crucial for personalized medicine and reducing cancer recurrence.

Purpose of the Study:

• To develop an improved breast cancer classification model using feature selection and ensemble techniques.
• To identify significant features through correlation analysis and variance for enhanced classification accuracy.
• To evaluate the efficacy of ensemble methods in improving breast cancer diagnosis.

Main Methods:

• Utilized correlation analysis and variance for feature selection, followed by Principal Component Analysis (PCA) for dimensionality reduction.
• Applied and tuned seven well-performing machine learning classifiers.
• Implemented ensemble learning by combining classifiers using hard and soft voting strategies.

Main Results:

• The proposed ensemble approach significantly improved breast cancer classification performance.
• Achieved a high accuracy of 98.24%, with precision of 99.29% and recall of 95.89%.
• Outperformed existing state-of-the-art methods on the Wisconsin Breast Cancer Dataset (WBCD).

Conclusions:

• Ensemble methods combined with rigorous feature selection offer a powerful approach for accurate breast cancer diagnosis.
• The developed model demonstrates potential for clinical application in early detection and treatment planning.
• Further research can explore additional feature engineering and ensemble strategies for even greater diagnostic precision.