A Comparative Performance of Breast Cancer Patient Classification Using Machine Learning Techniques
Putthipong Sawatdee, Karin Sapanan, Rathapaphon Jaroenklin, Tanachat Chaloermtanaskul, Benjawan Khunprasert, Narumol Chumuang
Abstract
Breast cancer is one of the most significant causes of mortality among women worldwide, where early detection and accurate classification play a vital role in improving survival rates. This study presents a comparative analysis of machine learning techniques for breast cancer patient classification using the Wisconsin Breast Cancer Dataset (WBCD) comprising 569 samples. Seven algorithms were evaluated: Simple Logistic, Logistic Model Tree (LMT), LogitBoost, Iterative Classifier Optimizer, AdaBoostM1, Random Committee, and MultiClass Classifier. Experiments were conducted with different train-test splits (70:30, 60:40, and 50:50). The results indicate that LogitBoost and Iterative Classifier Optimizer achieved the highest accuracy of 97.18% under the 70:30 split, while Simple Logistic and LMT reached 98.59% with the 50:50 split. Furthermore, LogitBoost exhibited superior computational efficiency across all configurations. The findings demonstrate that ensemble learning methods, particularly LogitBoost and AdaBoostM1, provide more robust and reliable performance than single models. This study underscores the potential of machine learning in supporting clinical decision-making for breast cancer diagnosis, with future research recommended to employ larger, more diverse datasets and integrate Explainable AI to improve transparency and trust in medical applications.