Year: 2025
Type: article
Abstract: Network reliability is essential for faultless communication in the telecommunication sector. The paper, Network Fault Severity Detection Using Log Data, focuses on predicting fault severity in the network of Telstra based on machine learning. The research establishes a predictive model to predict fault severity into three categories: 1 denotes a few errors, 2 numerous faults in them, and 0 denotes no defects. The approach requires heavy data preprocessing, feature design, and data exploration to establish patterns in logs. Machine learning algorithms like CatBoost, Random Forest, XgBoost, and LightGBM are utilized for accurate prediction. Feature importance analysis is also used to further improve model explainability by isolating major factors of failure. The study highlights the predictive analytics contribution to enhancing network reliability, minimizing downtime, and maximizing customer satisfaction. The method aids Telstra and other telecommunication providers in optimizing service quality, resource efficiency, and maintenance. The scalable approach suggested guarantees proactive fault detection, thereby reducing operational costs and enhancing overall network performance.
Source: Journal of Neonatal Surgery
Institution: Bharathiar University
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7219245578
