A Customizable Network Traffic Analytics and Security Dashboard using Machine Learning and Local Packet Capture

This paper presents a locally hosted network traffic analytics and security dashboard that combines live packet capture using Wireshark, machine learning-based traffic classification, and a real-time Streamlit dashboard. A Random Forest model trained on the CIC-DDoS2019 dataset is deployed to detect potential attacks from extracted flow features. Traffic is continuously logged to an SQLite database, providing users with data ownership and privacy. This implementation offers an alternative to cloud-based security solutions, suitable for SMEs and individuals seeking robust, on-premise network defense.