As the statistical agency of the Government of Australia, the Australian Bureau of Statistics provides key statistics on a wide range of economic, population, environmental and social issues, to assist and encourage informed decision making, research and discussion within governments and the community.
ACEMS and the ABS collaborate on cutting edge statistical, methodological, and applied challenges, as the organisations focus on the use of big data and powerful technology to keep Australia at the cutting-edge of statistical analysis.
The breadth of research conducted includes growing expertise in machine learning and data analytics, exploring developments in estimations with other National Statistical Organisations (NSOs), use of remotely sensed data, data linkage, and recommender systems.
Collaborating with the experts within ACEMS has enabled the ABS to explore how statistical machine learning methods can influence the production of official statistics. Bringing together ABS staff and ACEMS academics in a collaborative environment ensures that the organisational culture and methodological knowledge are brought together to develop applicable and implementable insights, not only academic insights.”
Sybille McKeown, Australian Bureau of Statistics
The Australian Bureau of Statistics is a valued long-term partner of ACEMS. As in past years, the engagements and collaborations enabled by this partnership have benefited stakeholders within and beyond both organisations.
As testimony to the strength of the partnership, the level of partnership activity in 2020 exceeded both the forecast and 2019 in-kind value. Additionally, new opportunities and collaborations have been seeded for 2021, and beyond the life of ACEMS as a Centre of Excellence.
Some highlights from 2020 include:
Some of these highlights are elaborated on further below.
ACEMS members serve on the Australian Bureau of Statistics’ Methodology Advisory Committee (MAC), the role of which is “to review and direct research into the collection, estimation, dissemination and analytical methodologies associated with ABS statistics” including Methodology Advisory Committee (MAC) Papers.
Working with ABS Methodology Advisory Committee (MAC) colleagues over the years has been an enriching experience, both professionally and personally. As a leading National Statistics Office, the ABS is engaged in the development and application of rigorous and novel statistical methodologies and technologies. The ABS values academic input, and working with them collaboratively supports the quality of their work in addition to informing further academic research – a virtuous cycle. The ABS is at the forefront of enabling the Australian public service to unlock the potential of public sector data. Working with the MAC is perfect for identifying the next set of nationally important questions and challenges to address.”
ACEMS CI Scott Sisson
The MAC generally meets quarterly to discuss working papers, with presentations on each paper from an academic reviewer and ABS representative, followed by a committee discussion.
ACEMS professors who served on the ABS’ MAC in 2020 were:
ACEMS and the ABS organised an ABS-sponsored workshop In February 2020, before the global pandemic limited travel.
A group of 25 academics and official statisticians met at ACEMS UTS over the course of a week. The workshop drew participants from Australia, Ireland, Italy, Netherlands, New Zealand, and the UK. They shared practical and methodological developments across several countries, in preparation for the 2020/21 round of population censuses, and the future beyond.
The knowledge exchange was valuable for all participants, both academic and professional, and resulted in the publication of a paper entitled Exploring developments in population size estimation.
Former ABS Chief Methodologist who worked at the ABS for over 30 years, Dr Siu-Ming Tam was one of the key organisers of the workshop. He combines a great depth and breadth of knowledge and experience across government, consulting, and now academia. Among his key achievements at the ABS is the publication of ABS statistics online for free public use, to “enable better decision making”; and allowing use of ABS statistics under a creative commons licence to limit bureaucracy.
ACEMS AI James Brown is the ABS Professor of Official Statistics at ACEMS UTS. The ABS provided financial support to first establish this role in 2013 when he joined UTS. James has a long history of collaborating with National Statistical Organisations (NSO), including in the UK and Europe, and expertise providing NSO methodology support. James’ research interests include the: application of multilevel models to the analysis of social data; assessment of census coverage and quality; design and analysis of surveys; and the use of surveys for policy evaluation.
Industry projects and internships offer benefits for both the ABS and ACEMS student researchers. In 2020, the ABS and ACEMS discussed and promoted a number of possible projects and internships for students.
ACEMS PhD student Prosha Rahman commenced an Industry internship project with the ABS on a topic related to unbiasing big data. Big data offers distinct advantages for use in official statistics, but when originally collected for other purposes, there may be issues such as bias. Conversely, surveys with suitable random sampling methodology can address the aforementioned issued, but with a bigger margin of error. This project investigates how to harness the best of both approaches, to unbias big data. Prosha is being supervised by ABS’ Dr Tam and ACEMS CI Scott Sisson and AI Boris Beranger, and they are working on publishing a paper based on work in this project.
The ABS also proposed short-term supervised industry projects/internships for ACEMS students on a number of topics.
ACEMS has enjoyed a range of engagements with ABS colleagues in 2020, which have benefited both organisations, including by building personal relationships and knowledge exchange, which strengthens the partnership and the impact it realised. This has best been articulated in the industry keynote presentation at ACEMS retreat by the ABS’ Sybille McKeown profiled below.
|Building Strong Relationships for Successful Partnerships|
|“Relationships are key to successful partnerships between research and industry. Building relationships, through engagements, helps researchers to develop a better understanding of the organisational context, needs, and conducting research that’s relevant to and can be successfully translated by industry for meaningful impact”|
Sybille McKeown, Program Manager for Methodology at the ABS, has had a long-standing active relationship with ACEMS. Sybille has served on the ACEMS Governance Advisory Board (GAB) until a role change at the ABS meant she stepped down at the end of 2020. Kirsten Stone from the ABS has taken her place on the GAB.
As Program Manager for Methodology, Sybille has had responsibility for developing and deploying statistical methodologies that underpin ABS transformation objectives and enable delivery of new statistical solutions. She is passionate about building methodological capability required for the ABS to take advantage of opportunities, such as advances in technology, big data, and use of administrative information for statistical purposes.
Highlight engagements between the ABS and ACEMS in 2020 are outlined below:
ABS’ Sybille McKeown delivered the industry keynote at the opening of the main retreat on the topic “What’s at the core of an ACEMS legacy”, which Sybille was well-placed to deliver, as a valued ACEMS collaborator since the launch of ACEMS Centre of Excellence.Watch Sybille’s presentation
ABS Assistant Director Claire Clarke spoke on the topic “Machine Learning in National Statistical Offices”, including NSO’s interest in ML methods, challenges related to applying ML in NSOs, and opportunities for ACEMS frontier research to benefit NSOs including ABSWatch Claire’s presentation
“This partnership creates benefits all round.
These projects will enable QUT researchers to sink their teeth into real world problems, and the ABS will build data science capabilities that modernise current operations.
PhD researchers from QUT will be able to collaborate with ABS specialist teams on joint projects that provide industry experience and put research into practice
Through data science innovations, we can continue our progression from a substantial reliance on sample surveys to making better use of the vast array of data that already exists via other sources”
ACEMS PI and ABS Chief Methodologist and General Manager, Methodology Division, Dr Anders Holmberg
The ABS has been a valued partner of ACEMS since inception of the Centre of Excellence. Over this time, the ABS and ACEMS have developed a strong and synergistic partnership, with collaborations across multiple ACEMS nodes.
ACEMS is keen to see new collaborations seeded which extend beyond the Centre of Excellence, and this new partnership is part of that legacy.
By working together, we can use data to inform decisions and make a real difference to government services,
“Our research is at the centre of developing new statistical and computational methods to solve complex problems, including food security, business and government processes as well as social challenges.” – ACEMS CI Kerrie Mengersen
The relationships developed as part of the ABS and ACEMS have laid strong foundations for ensuring a legacy beyond the Centre of Excellence.
ABS Assistant Director Bernadette Giuffrida has been appointed co-chair and will be a visiting fellow to QUT.
The ABS and ACEMS researchers from QUT node’s Centre for Data Science have already committed to continuing collaborations by co-funding an Associate Professor in Data Science and Government Statistics Chair, commencing in 2021.
The Chair will be based at ACEMS QUT and work in partnership between the ABS and QUT to:
The ABS and QUT have committed to co-funding three PhD positions at QUT for work to commence in 2021 on the following projects: