TRANSPARENT INTELLIGENCE: A COMPARATIVE STUDY OF MACHINE LEARNING MODELS FOR BREAST CANCER DIAGNOSIS

Abstract

This research presents a comprehensive investigation into the application of machine learning techniques for breast cancer prediction using the Wisconsin Diagnostic Breast Cancer (WDBC) dataset. The study systematically compares the performance of five widely used supervised learning algorithms—Logistic Regression, Random Forest, Support Vector Machine (SVM), Decision Tree, and K-Nearest Neighbors (KNN)—to evaluate their diagnostic accuracy and robustness. In addition to predictive performance, particular emphasis is placed on model interpretability, achieved through the integration of SHapley Additive exPlanations (SHAP), which enables transparent interpretation of feature contributions for the best-performing model. Moreover, Principal Component Analysis (PCA) is employed to reduce dimensionality, visualize data structure, and highlight class separability between benign and malignant cases, thereby offering deeper insight into the intrinsic patterns of the dataset. The findings of this research contribute to the growing body of knowledge on the use of artificial intelligence and data-driven methodologies in medical diagnostics, with a dual focus on predictive precision and explainability—two essential pillars for clinical adoption. Given that breast cancer remains one of the leading causes of mortality among women globally, the development of reliable and interpretable diagnostic systems is of paramount importance. Traditional diagnostic methods, though valuable, often suffer from limitations in speed, scalability, and objectivity. Machine learning, by contrast, provides a powerful analytical framework capable of identifying subtle and complex nonlinear patterns within biomedical data, enabling earlier and more accurate disease detection. The WDBC dataset, derived from fine-needle aspirates of breast masses, offers a high-quality and well-structured foundation for this exploration. Its detailed set of features describing cellular characteristics serves as an ideal basis for training, validating, and interpreting machine learning models that hold promise for enhanced clinical decision support and improved patient outcomes.