Predicting crowd density for public transport in real time

CK Delta, the data-innovation unit within CK Hutchison Holdings, worked with M1 — Singapore's first digital network operator — to forecast crowd density at train stations in real time, and feed it to commuters through a mobile app. With public funding and a hard deadline, there was no room to miss.

The challenge

M1 secured public funding to deploy Wi-Fi routers across Singapore's transport network, on the condition that citizens benefit from the data collected. CK Delta needed a predictive analytics solution to forecast crowd density at stations and power an app that tells commuters about waiting times and congestion — especially important for maintaining social distancing in the post-COVID environment. Building an in-house streaming system would have risked the deadline.

The solution

Rather than build from scratch, CK Delta chose Quix as a production-ready, full-stack yet modular stream-processing platform. Data scientists could process data in Python and deploy ML models to production autonomously — with no re-engineering from Python into Java or Scala — while platform engineers integrated cleanly with M1 and the Land Transport Authority's existing IT infrastructure.

Security was handled through deployment into a dedicated AWS account with VPN access and strict access controls, separate development, staging and production environments, and regular security scans and penetration testing. The parallel workflow paid off: three data scientists built the algorithms while two platform engineers handled the integration, without any code refactoring between them.

The results

The system launched on schedule, in four months, and exceeded expectations — handling nearly 40GB of data a day, quadruple the originally planned capacity, while staying reliable and secure. CK Delta now plans to extend the solution to additional cities and regions.