Breast Cancer Prediction Using Support Vector Machine Ensemble with PCA Feature Selection Method

Abstract

Breast cancer is the most prevalent cancer among women worldwide and ranks second in cancer-related mortality, comprising 11.6 percent of all cancer cases. Given that survival outcomes are largely contingent on the stage at which the disease is detected, early detection plays a pivotal role in securing the best prognosis for patients. Machine Learning algorithms are increasingly employed in breast cancer diagnosis due to their accuracy and capacity to anticipate the likelihood of recurrence. In this research, Support Vector Machine (SVM) was chosen as one of the classifiers, recognized for its precise predictive capabilities in cancer prediction and prognosis. To enhance model accuracy and mitigate variance, Principal Component Analysis (PCA) feature selection was incorporated into the study. Among the methods explored, the boosting ensemble method utilizing SVM as the base classifier demonstrated superior performance in breast cancer prediction. SVM as the base classifier with boosting ensemble method has outperformed the SVM models by increasing the accuracy value from 94% to 96% with a precision and recall score of 97%. Consequently, this research contributes to the advancement of patient diagnosis by implementing a classification algorithm tailored for breast cancer prediction.