Vision

From Data to Insight

Data analytics derives patterns from a large quantity of unprocessed data to develop predictive capability from the dominant patterns. Data analytic capability is important to demand forecasting, incident management and prediction, and congestion management in general. Machine learning techniques are the basis for data analytics development.

Machine learning is a subfield of computer science. It constructs systems that can learn from data, rather than follow explicitly programmed instructions. The general procedure of machine learning includes:

  • Data Collection: collect the observed data and corresponding labels as the training data
  • Model Learning: Learn model structures and parameters for the specified type of the model.
  • Prediction: Use the learned model to predict the label for a new observation
  • Feedback loop: Present the prediction results to domain experts and get feedback regarding the prediction results.

For transport industry, data analytics leverages ubiquitous data generated from surveys, traffic control system, transit system and telecommunication systems to built insight into the transport systems. This allows the transport authorities, operators and travellers to make the most informed decisions as soon as patterns emerge from the real-time data. In particular, data analytics build valuable insights into:

  • Demand: how is demand generated and distributed temporally and spatially
  • Behaviour: how is travel behaviour being influenced by congestion, pricing and public transport accessibility
  • Supply: whether the supply of capacity from infrastructure meets demand; what is the impact of incident and planning on the demand and behaviour

Machine Learning brings disparate data sets together into useful and coherent transportation predictions.

An Integrated Platform for Congestion Management

Recurrent congestion is a product of three forces: travel demand, travel behaviour and infrastructure capacity supply. There are multiple factors that can influence either one, or all of the forces, and consequently determine the level of congestion.

Recurring road congestion is a function of demand, travel behaviour and infrastructure capacity.

To address congestion we propose working with transport agencies and the travelling public to build the best possible system for understanding and improving both the real-time and long-term performance of the road and wider transport network.

This system will be the foundation for understanding how, where, when and why congestion occurs and how to mitigate its effects in real-time. This will inform future transport infrastructure planning and investment. The system will be a long-term transport asset for the people of NSW and their government.

The core innovation of the system is a platform to integrate and fuse transport data from all current and future data sources and to incorporate this fused data into transport models built using the most advanced analytic techniques.

A big data computing platform can fuse data into transport models, for situational awareness, predictive analytics and active traffic management.