Current and past data fellows and their work
Participants in our Data Fellowship Program have carried out a variety of data-related projects. Here’s a list of who they are and what they’ve done.
The most recent data fellows
Understanding the healthcare journey of high-cost patients
Robin Stevenson — Department of Health
Robin’s project will bring together data from the Medical Benefits Scheme, the Pharmaceutical Benefits Scheme and hospital admissions. Diagnostic and treatment information will be used to find associations between the codes that doctors, pharmacists and other medical professionals assign to this data. This will then be used to understand the healthcare journeys of high-cost patients and to support evidence-based policy development.
Robin is a data scientist and physicist and works at the Department of Health. Robin has a Bachelor of Philosophy (Honours) in physics and a PhD in physics.
Modelling household and family income from tax returns
Sam Bye — Australian Bureau of Statistics
It is known that income is most often shared at a household/family level. However, at the moment administrative income data is only available at the individual level. Sam will use the (de-identified) information collected on tax returns to model household and family income. This information can then be used to make policy decisions such as where to spend public money in areas such as health, education and social welfare.
Sam is an Analyst at the Australian Bureau of Statistics. He has also worked in international banks and holds a Bachelor of Commerce and a Graduate Diploma in Applied Finance and Investment.
Encouraging the productivity of Australian exporters
Shea Houlihan — Department of Prime Minister and Cabinet
Shea will encourage the productivity and competitiveness of Australian exporters by:
- identifying a sub-set of low-performing businesses and
- designing a behavioural intervention to help improve their performance.
The project will identify businesses that are most likely to respond to behavioural intervention, and aims to integrate behavioural science and machine learning to understand exporter behaviour.
Shea is a statistician at the Department of Prime Minister and Cabinet, in the Behavioural Economics Team of the Australian Government – BETA for short. Shea has also worked for the government in the UK, is a trained economist and has a PhD in policy evaluation.
Applying machine learning to intelligently impute missing values
Marcus Suresh — Department of Industry
Datasets with missing values is a common and often unavoidable problem which can bias research. Marcus will develop an application to alleviate this bias by using machine learning to intelligently estimate missing values and substitute them into datasets used to forecast the trajectories of Australian firms through the economy.
Marcus is currently an economist at the Department of Industry and Innovation and Science, and has previously worked at the Commonwealth Treasury, Prime Minister and Cabinet and at the ANZ Bank. Marcus has a Bachelor of Economics (Honours), a Bachelor of Commerce (Accounting) and a Master of Public Policy (Economic Policy).
Using supervised machine learning to assign national Major Vegetation Groups to data
Jeremy Groves — Department of Environment and Energy
The way that the states and territories collect and supply native vegetation data, to the commonwealth, differs from state to state. This project will use supervised machine learning to assign national Major Vegetation Groups (MVGs) so there will not be the need for the states and territories to standardise their data. This project uses the NSW database, to keep the scope limited to start with, but model will be used on all state and territory data. This would mean that the states and territories won’t have to spend time standardising their data to the national system as consistency will be automated. The data is used for national and international legislation and vegetation maps.
Jeremy works in Environmental Resource and Information Network (ERIN) at the Department of the Environment and Energy. He has also worked in universities and the mining sector. He has an Associate Diploma in Geoscience, a Bachelor of Science (Environmental Science) and a PhD and Post-Doctoral research in freshwater ecology.
Past data fellows
Checking financial condition reports
Todd Campbell — Australian Prudential Regulation Authority
Todd looked at how to use text analytics to help identify insights and risks in prudential supervisory reviews of insurance companies’ financial condition reports. He also looked at identifying efficiencies and compiled an overview of the industry to assist supervisors.
These documents are currently reviewed manually to identify issues and feed into other risk assessments. Specialist teams also review the documents to help supervisors, and more systematic analysis could assist this process.
Todd is responsible for delivering an Innovation Centre Lab for the Australian Prudential Regulation Authority.
The goal of the lab is to test modern analytics approaches that could provide new insights and support decisions relating to the authority’s prudential supervision role.
Automating land-use delineation
Rakhesh Devadas — Australian Bureau of Agricultural and Resource Economics and Sciences (Department of Agriculture and Water Resources)
Rakhesh developed techniques to automate the way agricultural land uses are mapped at a national scale.
His project improved methods for building the Land Use of Australia data series for understanding current and long-term changes in Australian land use.
He did this by developing advanced geospatial data-analysis and modelling techniques. These techniques combine satellite-derived information and various national datasets.
Rakhesh has post-graduate qualifications in agricultural economics and a doctorate in spatial data applications.
His 16 years’ of professional experience include implementing operational projects for the NSW and Queensland state governments and research projects involving satellite time-series data at RMIT University in Melbourne and University of Technology Sydney.
Analysing government buying patterns
Patrick Drake-Brockman — Digital Transformation Agency
Patrick’s project provided data about the way government agencies buy products and services and the sellers they buy from.
He used longitudinal network analysis to look at datasets held by AusTender, the website where many contract details are posted. This analysis showed the effects policy changes have on forming networks between government agencies and sellers.
Patrick is a senior adviser in the Digital Transformation Agency’s investment office, providing advice to the government on ICT procurement proposals.
He’s spent 19 years in the Australian Government in roles including ministerial ICT support, critical infrastructure policy and national security information sharing policy.
Detecting non-compliance in regulatory schemes
Gabriella Duddy — Clean Energy Regulator
Gabriella used machine learning to help the Clean Energy Regulator (CER) detect non-compliance in its regulatory schemes.
She developed a process for the Small-Scale Renewable Energy Scheme to help the CER to use its resources more efficiently and strengthen the integrity of Australia’s Renewable Energy Target.
Gabriella has worked across the CER’s intelligence and analytics functions, helping to develop the agency’s approach to increasing data capability.
She’s in her final semester of a Master of Energy Change at the Australian National University in Canberra.
Targeting biosecurity risks at airports
Dipangkar Kundu — Department of Agriculture and Water Resources
Dipangkar developed an empirical model for identifying and targeting potential non-compliance biosecurity risks at Australian airports.
His model helped to find risk patterns and identify international travellers who carry a higher biosecurity risk.
Dipangkar is a senior data analyst at the Department of Agriculture and Water Resources, where he was honoured with the Secretary Award of Achievement for his contribution to biosecurity risk profiling.
He has a PhD in computational hydrology from the University of Sydney.
Improving survey accuracy
Daniel Merkas — Australian Bureau of Statistics
Daniel developed machine-learning models to improve the quality and efficiency of the address register held by the Australian Bureau of Statistics.
The register contains more than 10 million addresses and helps to improve surveys and link datasets to inform Australia’s important decisions.
He used machine-learning algorithms to automate decisions that currently need people to carry them out and are resource-intensive.
Daniel is a statistician in the bureau’s address register unit and is studying for a Master of Science (Applied Statistics) at Swinburne University.
Simulation model for call-centre activities
Jack Xu — Department of Human Services
Jack buildt a simulation model that replicated the day-to-day telephone activities of the Department of Human Services.
He showed information including customer wait times, number of transferred calls, busy signals, level of staff occupancy and more. Jack’s model enabled his team to answer important business questions with more confidence than ever before.
Jack is a data analyst whose job is to look for ways DHS operational staff can be more efficient in their work. This includes helping to ensure the department achieves its telephone and claims processing deliverables.
He previously worked at analytics software firm SAS Institute and holds a Bachelor of Actuarial Studies.
Using location data from business numbers
Oliver Berry — Australian Bureau of Statistics
Oliver used machine learning to explore new ways of using data from an Australian Business Number (ABN). This meant that business information could be analysed based on location.
Improving the timing of social media posts
Ross Ryan — Department of the Prime Minister and Cabinet
Ross’s project aimed to predict and improve attendance at work-like activities in remote Australia after they were promoted through social media.
These are activities some job-seekers take part in to benefit their communities.
Assessing research funding levels
Tash Roujeinikova — National Health and Research Medical Council
Tash used machine learning techniques and text mining to show the level of interdisciplinary research funded by the National Health and Medical Research Council.
Real-time household consumption
Alex Kelly — Treasury
This is a new way of measuring household spending using social media resources.
Alex used data on trends including travel, retail, home and car sales to provide a real-time indicator of household consumption.
Expenditure models for policy design
Yingsong Hu — Department of Finance
Yingsong used health-related datasets to create expenditure models for evidence-based policy design.
Comparing different GDP modelling in Australia
Audrey Lobo-Pulo — Treasury
Audrey used machine learning techniques to create a new way of predicting economic growth in Australia.
Expanding a real-time file identification system
Janis Dalins — Australian Federal Police
Janis expanded on a real-time file identification system that supports digital forensics.
Machine learning allowed for predictive analysis based on lightweight features such as metadata. Similarity checks were done with fuzzy hashing.
Read more about Janis Dalins’ work on our blog.
Container terminal model for data analysis
Thomas Rutherford — Bureau of Infrastructure, Transport and Regional Economics
Thomas designed an agent-based model of a single container terminal.
This model uses vessel and trade data to create synthetic inputs and to validate outputs. This will later scale up to the wider intermodal supply chain.
Streamlining data publication
Dominic Love — Australian Bureau of Statistics
Dominic used machine learning to streamline how the Australian Industry publication is compiled and produced. This included improvements to the processing cycle of the Economic Activity Survey.
Simulation for health risk
Imaina Widagdo — Department of Health
Imaina designed a microsimulation model to assess hospitalisation risk in chronic disease patients.
Estimating greenhouse gas emissions
Senani Karunaratne — Department of the Environment and Energy
Senani developed an empirical model to predict the changes of terrestrial soil carbon.
This model monitors soil carbon in Australia’s crop and grasslands and can be used as a validation tool for the official estimates of greenhouse gas emissions.
Harmful trading detection techniques
Tariq Scherer — Australian Securities and Investments Commission
Tariq used ASIC data to develop techniques for detecting harmful trading. These techniques find patterns of repeated misconduct and relationships between entities of interest.
Analysis of truck and road freight activity
Richard Green — Bureau of Infrastructure, Transport and Regional Economics
Richard created a process and platform for analysing GPS data from road freight vehicles. This provided insights into congested areas of the road network, rest patterns of truck drivers, and changes in road freight activity.Back to top