The Internet has emerged as a crucial resource for mankind, underscoring the paramount importance of information security. Intrusion Detection Systems (IDS) play a pivotal role in safeguarding networks against cyber threats. However, the imbalanced nature of data distribution and high dimensionality pose significant challenges in developing effective IDS. This paper presents a novel approach addressing these challenges through an innovative oversampling technique and feature selection method. Our proposed method, HOK-SMOTE, leverages an Ordered Weighted Averaging (OWA) approach for feature selection from the KDD Cup 99 dataset and employs K-Means SMOTE for imbalanced learning. Additionally, we compare our hybrid algorithm against an ensemble model comprising Support Vector Machine (SVM), K Nearest Neighbor (KNN), Gaussian Naïve Bayes (GNB), and Decision Tree (DT) classifiers, utilizing weighted average voting for output prediction. Extensive experimentation on various oversampling techniques and traditional classifiers demonstrates the superior accuracy of our proposed approach. Precision, recall, F-measure, and ROC curve analyses confirm the effectiveness of HOK-SMOTE coupled with ensemble learning in mitigating imbalanced learning in IDS. Our findings shed light on the dominance of ensemble modeling and oversampling techniques in addressing intrusion detection challenges, providing a precise solution for robust network security.