Presented By O'Reilly and Cloudera
Make Data Work
December 1–3, 2015 • Singapore

Adaptive, demand-driven shared transit through crowdsourcing and big data

Feng-Yuan Liu (Infocomm Development Authority of Singapore)
11:50am–12:30pm Wednesday, 12/02/2015
Hadoop & Beyond
Location: 328-329 Level: Non-technical
Average rating: ***..
(3.91, 11 ratings)

Prerequisite Knowledge

Layman understanding of Apache Spark


With public transportation there is often a trade-off between:

  1. A cheaper but lengthier – and sometimes uncomfortable – ride via public transport, or
  2. A faster but costlier “personalized transport” (e.g., taxis and private cars)

We think technology – data science and mobile technology – can help overcome these trade-offs and provide direct, data-driven transportation options. Our work at Singapore’s Government Analytics department offers an alternative direct transport mode that lies between both options. We see it as democratizing the Uber experience.

By aggregating demand, direct express bus routes quickly and comfortably ferry riders with minimal stops. At the same time, cost is reduced by aggregating demand into larger buses, as opposed to limited capacity taxis.

To achieve this, we bring together crowd-sourced insights and historical data analytics: commuters are empowered to suggest and book routes via the Beeline app, while farecard (tap-in/tap-out) transactions and taxi data are analyzed to find historical demand. Following that, an optimization algorithm identifies a set of viable express routes, each constrained to at most sis stops. Bus operators around Singapore then offer these routes for booking via the Beeline app.

In this session, we will cover the following:

  • How crowdsourcing is combined with historical data to optimise bus routes
  • What operational issues arise when transitioning analytics-proposed routes into real-world deployment
  • Whether our experimentation succeeded, or failed, in changing the commuting patterns of Singapore
  • How we utilized Apache Spark to process terabytes of farecard transaction and taxi data (using a graph-based algorithm)

Feng-Yuan Liu

Infocomm Development Authority of Singapore

Feng-Yuan is director, government analytics at the Infocomm Development Authority (IDA) of Singapore. He heads a multidisciplinary team of data scientists including data analysts, social scientists, computer scientists, and data visualizers that help government agencies in Singapore make sense of their data and to enable evidence-based decision making. He has worked in various government agencies and public policy roles, most recently in public transport regulation at the Land Transport Authority, in climate change and energy policy at the Ministry of Trade and Industry, and in the Ministry of Finance.