Software-Defined Networking (SDN), with its centralized control architecture, enhances network management flexibility but remains vulnerable to Distributed Denial of Service (DDoS) attacks, which can disrupt communication between the application and data planes. Effective attack detection is therefore essential. Feature Selection (FS) plays a vital role in improving intrusion detection systems by reducing computational overhead while maintaining high accuracy. This article proposes a hybrid FS method, Filter-Wrapped Evaluation (FWE), which combines Pearson Correlation and Recursive Feature Elimination (RFE) using XGBoost to improve DDoS detection in SDN environments. The proposed FWE method enhances detection performance by selecting the most relevant features and eliminating redundancy, thereby boosting accuracy and reducing processing time. Experimental evaluations on four benchmark datasets demonstrate that using the remaining features, FWE achieved 99.99% accuracy on CICIDS2017 and 100% accuracy on CICIoT2023 when paired with Decision Tree and Random Forest classifiers. When reduced to the top 10 selected features, FWE maintained high performance, achieving 99.99% accuracy on the APA-DDoS dataset with significantly lower execution time as low as 0.0871 s with Decision Tree. These results confirm FWE’s effectiveness in balancing detection accuracy with real-time efficiency. Future work will focus on optimizing feature thresholds and evaluating performance with deep learning models.