Advancing Network Security: A Comparative Evaluation of Machine Learning and Ensemble Methods in Intrusion Detection Systems

This paper presents an automated classification framework to address the challenges of imbalanced datasets in machine learning applications across domains such as health, finance, and marketing. The approach integrates data preprocessing, SMOTE for oversampling, and ensemble-based algorithms, including Random Forest, Gradient Boosting, XGBoost, Stacking, and Bagging Classifiers. The use of Scikit-learn pipelines ensures modularity and scalability in addressing classification problems. The study highlights the effectiveness of SMOTE in improving recall for minority classes and demonstrates the superior performance of ensemble methods for classification tasks. Future work includes optimizing hyperparameters, exploring deep learning approaches, and integrating real-time classification systems for practical applications.