Transport / Real-time Analytics
Live Tube Train Monitoring
CLIENT — Transport for London (with the Open Data Institute)
A cloud-based analytics engine that turned prototype instrumentation on Victoria Line Underground trains into live and historical insight, exposed through an API for web visualisation.
The challenge
Prototype instrumentation on Victoria Line trains produced high-volume live and historical data on TfL's servers. The data had to be accessed, reconciled against timetables and signal history, and turned into canonical, queryable data points — in real time and at scale.
Our approach
We built a custom cloud analytics engine that pulled live and historical data from TfL's Elasticsearch servers, processed it into canonical messages on React AI's cloud, and reconstructed current state from signal history cross-referenced with the timetable. A websocket API let partner Vizicities retrieve historical data or replay it live for web UIs and visualisations. The initial phase assessed the available datasets and weighed approaches — regression, machine learning, network simulation and Bayesian techniques — across train-, track- and network-level analysis.
Results
The work established React AI as a registered TfL preferred supplier.
Technologies
- Cloud
- Docker
- Real-time data processing
- Elasticsearch
- WebSockets
- Bespoke development
Turning prototype Underground-train instrumentation into a real-time analytics engine for Transport for London.