Breast Cancer Prediction: A Comparative Study of Different Machine Learning Algorithms Across Multiple Data Sets

Early detection is critical in the treatment of breast cancer, which is a major cause of worldwide cancer related deaths. In this study, we report results from applying four machine learning techniques – Logistic Regression, Decision Tree, Random Forest, and Support Vector Machines – on four publicly available data sets – Diagnostic Wisconsin Breast Cancer Database, Original Wisconsin Breast Cancer Database, Mammographic Mass Database, and Prognostic Wisconsin Breast Cancer Database – to compare the predictive performance of these techniques. We found Support Vector Machines to outperform all other techniques, reaching an accuracy as high as \(98.25\%\) in the Diagnostic Wisconsin Breast Cancer Database. Our results can complement medical services in the diagnosis and management of breast cancer.