Breast cancer is also one of the primary worldwide health concerns and detection at an early stage increases the survival rate considerably. This paper aims at investigating the application of machine learning models in classifying a tumor as benign or malignant. A data set with 569 examples and 30 numeric features derived from FNA images was processed for eminent tumor features like radius, texture, perimeter, area, and smoothness. To optimize model performance, feature selection techniques, including the Chi-square (Chi2) algorithm, were applied to identify the most significant attributes. Standardization was used to ensure uniformity in data distribution, preventing dominant features from influencing model training. Among the models tested, Quadratic SVM and Cubic SVM demonstrated the highest classification accuracy at 98.24%, outperforming other approaches. Neural networks, including Bi-layered and medium configurations, achieved accuracies of 97.54% and 97.36%, respectively. Weighted KNN and Medium Gaussian SVM followed closely with 97.01% and 97.89% accuracy. While ensemble models provided competitive results, their high computational requirements limited their practicality. Fine Gaussian SVM and Ensemble Boosted Trees exhibited the lowest accuracy levels, at 79.44% and 62.74%, respectively, emphasizing the trade-off between model complexity and efficiency. This research highlights the strength of SVM models in breast cancer classification, offering a balance between accuracy and computational feasibility By refining machine learning applications, medical professionals can improve early detection methods, leading to better patient outcomes.