Comparison of Feature Selection Methods for Breast Cancer Prediction

Breast cancer is the most common cancer among women worldwide. Doctors are faced with a vast array of multi-dimensional imagery data to diagnose diseases efficiently in real-time and take medical action. Fast and accurate Machine Learning (ML) techniques are used to assist doctors in predicting the severity risk of a patient. In this paper, we propose an experiment based on Feature Selection to reduce dimensionality from the Dataset using different methods such as Principal Component Analysis (PCA), Recursive Feature Selection (RFE), and Recursive Feature Selection with Cross-Validation (RFECV) while applying Machine Learning classifiers: Logistic Regression (LR), Decision Tree (DT), Random Forest (RF), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost) to predict the severity of breast cancer. To assess the predictive model, we evaluate its performance on four principal metrics: accuracy, precision, recall, and F1-score. The results show that using the RFE combined with the XGBoost classifier achieved the best results on all score metrics compared to using the features selected by RFECV or PCA individually. Additionally, XGBoost exhibited good performance, achieving 99.30% accuracy, 98.90% precision, 100% recall, 99.44% F1-score and 99.91% roc-auc.