About The Company

A major government transport agency in South-East Asia servicing a population of 5.7 million with services like buses and metro. Responsible for planning, designing, building and maintaining the land transport infrastructure and systems, they aim to bring about a more inclusive public transport system. They aim to leverage technology to strengthen their rail and bus infrastructure to provide exciting options for future land transport.


The agency sought to explore graph analytics to optimize the scenario prediction of the public transport network to predict the potential impact of bus service modifications, to address possible issues of over-provisioning. Specifically, the client wishes to accurately predict measurable impacts such as the number of passengers affected, average journey time increase, if they were to modify, reduce, or remove existing bus services.

Our Solution

  • Created an Origin-Destination graph to illustrate bus and train connectivity.
  • Created a routing graph to model current bus boarding behaviour and understand overlaps and potential substitution effects.
  • Provided a custom-built a scenario planning tool for the client powered by Lynx’s proprietary graph analytics engine.
  • Created impact simulations using the interactive, flexible simulation tool for predicting single, multiple, and bus-train modification scenarios.
  • Produced preliminary results closely tailored to the client’s operational requirements within 4 weeks. These results were trained and produced based on 1 month of travel card data from bus and train journeys.

The Outcome?

  • Provided a simulation which measured 40% accuracy in terms of matching the factual journey
  • Able to pinpoint specific scenario outcomes including but not limited to the following outputs:
  1. Number of average affected passengers daily
  2. The increase of average journey time in minutes or seconds
  3. Average Daily Public Transport Ridership
  4. Maximum vehicle load vs capacity and top substitute buses and trains

Screen Shot 2019-09-16 at 12.55.59 AM