Evaluation of Mutiple Classical Machine Learning Models for Breast Tumor Classification

Globally, breast cancer continues to be among the most commonly diagnosed cancers in women. Accurate and early detection plays a critical role in enhancing treatment outcomes and lowering death rates. As structured clinical datasets become more accessible and Artificial Intelligence (AI) technologies evolve, Machine Learning (ML) is increasingly utilized to support diagnostic decision-making. This study methodically evaluated three supervised learning models commonly used in clinical practice, namely Logistic Regression (LR), Support Vector Machine (SVM), and Random Forest (RF). All models were uniformly trained and evaluated on the WDBC dataset, comprising 569 samples with 30 numerical features. To evaluate model performance, the study used four core indicators: accuracy, precision, recall and F1 score. Results show that SVM achieved the highest accuracy (97.36%), followed by Random Forest (96.49 %) and Logistic Regression <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(95.61 \%)$</tex>. The outcomes indicate that margin-based and ensemble classifiers provide reliable approaches for early-stage cancer identification. To build a reproducible machine learning process, this study also explored key issues in the practical application of the model, such as the impact of feature selection, the importance of model interpretability, etc. The study's results emphasize the value of traditional machine learning methods in clinical feasibility and lay the foundation for the development of future multimodal or deep learning systems.