Real-Time Urban Traffic Intelligence Platform

A real-time traffic control-room platform built as a student project that runs on an 8 GB laptop CPU. Loop-detector records travel through a Kafka-style message broker, a stream processor builds 5-minute event-time windows with a watermark, and an online machine-learning scorer forecasts the speed of every station 15, 30 and 60 minutes ahead with LightGBM and detects incidents. A React dashboard shows the live map, corridor travel times, a congestion heat map, the alert workflow and the pipeline monitor. The data come from a simulated city with 48 detector stations.

Category: Big Data / Stream Processing / Machine Learning / Intelligent Transportation Systems

CSE Project 34 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.