A Naïve Performance Metric Parameter for an Intrusion Detection System Model

Researchers employ a variety of metrics to assess the intrusion detection system (IDS) model’s performance quantitatively. These metrics help gauge an IDS’s ability to differentiate between normal network traffic and malicious activities and the performance of IDS models. Most researchers evaluate IDS model performances based on conventional metrics, which does not explore minor differences immensely due to their range that lies between 0 and 1. The present study introduces a new performance metric that distinguishes differences in the classification performance of the machine learning-based classification model, especially in scenarios where performance differences are minimal. The present study shows that an ensemble boosted tree with a TPFC value of 1134.4 is the IDS model with high conventional metrics-based values.