Public Transport Optimisation

We built a smart-app solution that helps public transport agencies match drivers to high-demand areas and give passengers accurate travel time estimates so they can plan smarter their travel journeys.

Comment 1

Inspiration

Public transport often suffers from poor planning. Some stops have too many vehicles, others leave passengers stranded. Drivers aren't always deployed where demand is highest, and passengers often can’t predict how long a trip will take, especially with multiple boarding/onboarding stops. While tools like Google Maps provide ETAs, they don't account for stop-based delays in public transport. Our solution fills this gap with smarter planning and better travel time estimates.

What it does

Our web app helps transport agencies deploy vehicles to areas with high demand and gives passengers more accurate ETAs based on actual trip behaviour, including intermediate stops. Using predictive models, the app estimates:

  • Where and when demand will peak
  • How long a journey will take, factoring in real passenger delays

How we built it

We built two regression models:

  • Demand model to predict trip volume by location and day
  • ETA model to predict travel time based on route, distance, day, time, and stop activity
    We deployed the models as APIs on Google Cloud Platform and connected them to a locally hosted interface for demo purposes.

Challenges we ran into

  • Data was messy and inconsistent. Merging and cleaning it took significant effort.
  • Serving ML models on GCP was technically demanding under time pressure.
  • Coordinating team efforts remotely required strong communication and motivation.

Accomplishments that we're proud of

  • Built a working ML-backed transport planning tool in limited time.
  • Deployed end-to-end model pipelines.
  • Maintained strong teamwork throughout. Every member contributed meaningfully.

What we learned

  • Every complex problem becomes clearer once you start solving it.
  • Clean data is key, but so is a collaborative team.
  • You don’t need perfect tools to build real, impactful solutions.

What's next for Public Transport Optimisation

  • Integrate real-time GPS data via Google Maps API
  • Expand dataset coverage for more accurate forecasts
  • Deploy on a production-ready web app (e.g. GCP App Engine or Cloud Run)
  • Partner with transport authorities to pilot in real-world settings

Built With

Share this project:

Updates