Network Intrusion Detection Using RandomForest Algorithm for Enhancing Cybersecurity Accuracy

This study implements a binary Random Forest classifier on the NSL-KDD benchmark dataset for network intrusion detection. The experimental pipeline covers label binarization, LabelEncoder encoding, and StandardScaler normalization applied to 125,973 training samples, with evaluation performed on 22,544 test samples. The trained model achieved 77.22% accuracy, 96.72% attack precision, and 97.22% normal-traffic recall, with a training time of 10.49 seconds. Feature importance analysis identified src_bytes, dst_bytes, and same_srv_rate as the three most discriminative network traffic predictors.