ARCHIVES

Breast cancer continues to be one of the most critical public health challenges worldwide, significantly contributing to cancer-related mortality among women. Early-stage diagnosis substantially improves survival outcomes and reduces treatment complexity. However, conventional diagnostic procedures, including mammography and biopsy, are often dependent on clinical expertise and may suffer from inter-observer variability and delayed decision- making. This research proposes a robust machine learning-driven framework for automated breast cancer classification using structured clinical datasets. The methodology integrates advanced preprocessing techniques, dimensionality reduction strategies, and multiple supervised learning algorithms, including Logistic Regression, k- Nearest Neighbors, Support Vector Machine, and Random Forest. Feature optimization through Principal Component Analysis (PCA) and Recursive Feature Elimination (RFE) enhances predictive efficiency and minimizes redundancy. Experimental findings reveal that ensemble-based classifiers, particularly Random Forest, achieve superior classification accuracy exceeding 97%, with improved precision and recall metrics. The proposed system demonstrates strong generalization capability and provides a scalable, interpretable, and computationally efficient solution suitable for clinical decision support systems.