Evaluating the Effectiveness of Random Forest + XGBoost versus Support Vector Machines in Achieving High Accuracy and Precision for Breast Cancer Classification

Aim: The aim of the current work is to compare and contrast the performance of classification models in diagnosing breast cancer, i.e., the hybrid model Random Forest + XGBoost vs the Support Vector Machine (SVM) model. Materials and Methods: Two models were debated. Group 1 used the SVM algorithm with manually specified hyperparameters, e.g., kernel and regularization, and validated it on the Wisconsin Diagnostic Breast Cancer (WDBC) data. Group 2 used an ensemble of Random Forest and XGBoost with a hybrid approach consisting of feature selection, normalization, and boosting to classify more effectively. The RF + XGBoost model created an ensemble by averaging multiple decision trees and gradient boosting to enhance prediction accuracy. The hybrid model always showed better precision and stability, and it worked well in noise and class imbalance. Results: The RF + XGBoost hybrid model achieved a total best accuracy of 97.8% and outperformed the SVM model with a best accuracy of 84.2%. The hybrid model also performed better than the SVM model in precision (0.961 vs. 0.813), recall (0.960 vs. 0.860), and F1-score (0.940 vs. 0.835), in addition to improved accuracy. Statistical analysis revealed that there was a significant difference in performance (p < 0.05), validating the enhanced classification ability of the ensemble model. Conclusion: RF + XGBoost hybrid model is an excellent extension of SVM for breast cancer detection with improved accuracy, precision, recall, and F1-score. The improvement is statistically significant, and the model is appropriately fit for real-time and high-level clinical diagnostic systems.