Contact Us

Drop us a message and we’ll be in touch!


    Live Tube Train Monitoring

    Data analysis for Transport For London

    Data Investigation with TfL

    In collaboration with the Open Data Institute, React AI were commissioned to investigate the possibilities opened up by the prototype instrumentation of underground trains on the Victoria Line.


    This included accessing high-volume live and historic data from the TfL servers, into a cloud-based custom analytics engine to establish canonical data points from the array of readings, which in turn exposed an API to our partner Vizicities who produced a web UI and visualisations of the data, in close collaboration with React AI.

    Services Provided

    Initial Report and Options

    React AI analysed the multiple data sets available, and produced a report covering:

    • Quality, quantity and potential uses of data sets.
    • Possible analysis approaches, including an overview of how the techniques could be applied, summary of the state of the art, and pros and cons for traditional approaches, such as data regression at points in time, data regression across time; Machine Learning approaches; network simulations; and other (Bayesian techniques; root cause analysis).
    • Advice on selection, variation and segregation of the data sets.
    • Input on other complementary sources of data.
    • Considerations for train-level, track-level and network-wide analyses.
    • Possible future capabilities of an analytics system.

    Data Analysis

    Key activities undertaken:

    • Analysis of high-volume signal data to extract key information for visualisation
    • Build current state from signal history.
    • Code to cross-reference train data and timetables
    • Combination of data sets (live data, incidents, timetables)


    The React AI analytics took data from the TfL Elasticsearch servers and processed them into React AI messages with the aim of distilling out the key information. The resulting messages were loaded to React AI’s cloud-hosted servers. The Vizicities UI (or other users) can connect via websockets and request either a pull of data from a specified time period, or an ‘as-live’ replay.

    React AI are registered as a TfL preferred supplier.