•  
  •  
 

Abstract

Breast cancer remains one of the leading causes of mortality worldwide, emphasizing the critical need for accurate and efficient diagnostic tools. This study investigates the effectiveness of combining linear and non-linear feature selection methods-Principal Component Analysis (PCA), Pearson Correlation Coefficient (PCC), and Backpropagation Neural Networks (BNN) to improve breast cancer classification using machine learning models. We utilized the Wisconsin Breast Cancer Dataset to evaluate the performance of five classifiers-Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Logistic Regression (LR), Decision Tree (DT), Naïve Bayes (NB) and Artificial Neural Network (ANN). The results demonstrated that BNN-selected features consistently outperformed PCA and PCC across all classifiers, with SVM achieving the highest classification accuracy. The results showed that BNN-selected features consistently outperformed PCA and PCC, with SVM achieving up to 97.3% accuracy and 98.1% precision, and KNN reaching 97.1% accuracy and 96.9% precision. Stacking ensemble models further improved performance: the non-linear SVM-KNN-ANN ensemble attained perfect classification metrics of 100% accuracy, precision, recall, specificity, and F1-score demonstrating the superior synergy between BNN and ensemble learning. These findings highlight the diagnostic advantage of combining non-linear feature selection with meta-learning approaches and support the development of robust, high-accuracy breast cancer detection systems.

Download

Included in

Engineering Commons

Share

COinS