Penerapan Recursive Feature Elimination pada Support Vector Machine untuk Klasifikasi Kanker Payudara

Abstract

Breast cancer is a prevalent type of cancer among women worldwide and can have fatal consequences if not detected early. Errors in breast cancer diagnosis can occur due to the use of irrelevant features or attributes, leading to misclassification. To minimize this possibility, this study applies the Recursive Feature Elimination (RFE) feature selection method to the WDBC (Wisconsin Diagnostic Breast Cancer) dataset to select the most relevant features in distinguishing benign and malignant tumor classes. SVM (Support Vector Machine) algorithm was used as the classification model with a data sharing ratio of 90:10, resulting in an accuracy of 0.98, precision of 1.00, recall of 0.94, and F1-score of 0.97. The implementation of RFE successfully reduced 50% of the features without reducing the performance of the model compared to the use of all features.