Breast cancer presents a major health challenge globally, emphasizing the necessity for improvements in diagnostics and therapies. As medical data expands, it offers chances to improve outcomes but complicates extracting actionable insights. This study investigates machine learning (ML) and deep learning (DL) applications for classifying breast cancer, utilizing data from the METABRIC. The METABRIC data processing included label encoding, managing dominance in categorical features, imputation of missing values, and normalization. It assessed eight ML algorithms: Random Forest, Logistic Regression, Gradient Boosting, Extra Trees, Linear Support Vector Classifier, Extreme Gradient Boosting, Light Gradient Boosting, and Category Boosting, with Category Boosting leading at an accuracy of 84.16%. Using SMOTE enhanced model performance. A 10-layer neural network using ReLU and Sigmoid activation functions achieved 86.91% accuracy, while a hybrid model optimized with a genetic algorithm for logistic regression on survival predictions reached 98.43% accuracy.