Risk, Uncertainty, and Decision-Making

Today, both society and the economy generate a complex torrent of data. If this unprecedented flow of information is to be made useful, we require new tools and methods for its analysis.

The measurement, interpretation, and communication of complexity and risk is a key part of modern science. Te Pūnaha Matatini researchers working within the Risk, Uncertainty, and Decision-Making theme (formerly called Complex Data Analytics) are developing tools for understanding and dealing with complex systems by developing the underlying theory. This includes work on optimising stochastic systems from supply chains to healthcare, inferring numbers of New Zealand birds from their calls (AviaNZ), and building a library of New Zealand soils from their spectral signatures. Public engagement with science is also a key part of Te Pūnaha Matatini’s work, and the researchers in this theme are working on ways to improve scientist-public interactions.



Equilibria and dynamics in networks and supply chains

Equilibrium models have been applied to networks in diverse economic settings including supply chains, telecommunications, electricity, and transportation networks. Supply chain strategy theory is far from complete in explaining why many different supply chain network structures co-exist. For example, different cooperative network structures exist within New Zealand: Fonterra is vertically integrated, while Zespri maintains a network of independent packing houses; the red meat industry is particularly fragmented with significant public debate on why this is the case. Risk is also an important factor in equilibrium, since variation in risk attitudes by agents leads to varying strategic decisions. The research proposed in this project will bring together ideas from these separate fields and extend current models to incorporate significant features that are not currently captured, such as risk and uncertainty in electricity networks. We will improve existing equilibrium concepts (such as supply function equilibrium and Markov perfect equilibrium), and generate new models, which can be used by policy makers to inform infrastructure policy, improve incentives, and guide policy development towards efficient outcomes.

Three-year outcome: models that are used by policy makers and regulatory authorities to inform improved infrastructure and industrial policy


Optimisation of complex information networks

Modern engineering is undergoing a data revolution. Increasingly complicated engineering and management systems are being designed and built to incorporate sensors that gather and communicate data about the state of the system. The potential for data-driven decision making depends on the value of information these sensors collect. Consider the value of information obtained from sensors placed in the buildings and vehicles of a modern city such as proposed by Christchurch’s Sensing City project. With such information, controlling traffic congestion becomes much easier with timely estimates of point-to-point trip demand. Sensors allow a taxi fleet to be made of entirely electric vehicles supplied with inductive electric power at certain sites. Public transport also becomes much more efficient if it can adapt routes and timetables based on observation of current demand. Such information has a value that accrues from enabling more flexible and sustainable operation of the city. We will devise and build data-driven models that will enable engineers and software architects to design and build intelligent sensing networks. The models will make use of stochastic optimisation techniques that trade off the value of information available from sensors with the cost and difficulty of obtaining that information. The research will focus on three fields, electricity networks, transport networks, and geothermal reservoirs.

Three-year outcome: a data-driven decision framework that is used to optimise the design and control of complex engineered systems (electricity systems, transportation systems, civil infrastructure, geothermal and other industries, water resource systems)


Healthcare analytics

Per capita health spending over 2000-09 is estimated to have grown, in real terms, by 4% annually on average across the OECD. This is driven partly by an aging population, as well as the development of new and expensive medical procedures. There are opportunities for controlling this expenditure by applying analytics to healthcare. Last year, Memorial Sloan-Kettering Cancer Center in New York and IBM agreed to collaborate on the development of a tool based on IBM’s Watson computer that will assist the hospital group’s doctors with the diagnosis of cancer and suggest treatment. New Zealand will need to forge similar collaborations here, focusing on both improving medical treatment, and efficient health care delivery. Our research team will apply methods from queuing networks, stochastic optimisation, and simulation to derive improved models for healthcare delivery. A healthcare system can be viewed as a network of services provided to the sick or injured. Patients traverse the nodes of this network in different orders depending on their condition and the outcome of the tests and procedures they undergo. Limitations in resources at the nodes generate queues (waiting lists). We will extend results from telecommunications queuing networks (in which the traffic typically does not gain new information as it travels) to the healthcare case. This will form a conceptual framework for the development of algorithms targeted at improving healthcare outcomes in particular settings.

Three-year outcome: healthcare analytics applications that deliver improved healthcare outcomes in the New Zealand health system