Predictive analytics methods for building a proactive network monitoring system

Abstract

Predictive analytics methods for building a proactive network monitoring system

Klychkov I.A.

Incoming article date: 16.05.2025

Modern predictive analytics methods significantly enhance the capabilities of network monitoring systems by enabling early detection of anomalies and potential failures. This article presents the results of a study on approaches to building a proactive network monitoring system using machine learning and statistical analysis methods. It is demonstrated that the use of combined models based on recurrent neural networks and autoregressive models provides the most accurate network traffic forecasting with a prediction horizon of up to 10 time intervals. The practical implementation of the proposed approach allows for a 27% reduction in unplanned downtime and a 35% decrease in incident response time compared to traditional reactive monitoring systems.

Keywords: predictive analytics, network monitoring, machine learning, statistical analysis, anomaly detection, traffic forecasting, recurrent neural networks, autoregressive models, proactive systems, fault tolerance