Project Info

Team insolvit thumbnail

Project Description

Over 110,000 people in Australia are bankrupt. Whether they share it or not, each one of us knows someone who has been affected…Life doesn’t always go to plan. So we pulled together a team of data scientists, lawyers, coders and engineers to research the problem and create a solution which we call INSOLVIT - a platform that uses real data to predict the likelihood of non-compliance with these obligations, and target resources to the people who need it most. The platform has 3 main features. The compliance dashboard provides an overview of the state of non-compliance in Australia. Our heat map visualises past non-compliance based on locations. We can then overlay other data sets to find correlations and determine why certain areas have high non-compliance. In addition, our cutting edge machine learning model allowed us to find the key attributes that lead to non-compliance. By understanding WHY people are non-compliant, the government can design the right intervention, which best utilises public resources. The second feature is individual risk profiles. The Australian Financial Securities Authority (also known as AFSA) is the trustee and is responsible for around 80% of insolvent individuals. Our machine learning model allows AFSA to calculate the risk of an individual becoming non-compliant with 98% precision. Our third feature is Insolvit Together – a platform clients going through the insolvency process which promotes education and compliance. We also created a community hub.

Data Story

We used the AFSA Non-compliance in personal insolvencies dataset ( ) as the primary dataset underpinning our project. The dataset was analysed through machine learning and through a deep dive manual analysis to identify trends, curation needs, and potential for linkages with other datasets. A key limitation of this dataset was the use of SA3 for location without access to other SA codes or postcodes data. More robust linkage would improve the project. Another key limitation was the size of the relevant dataset – the number of non-compliance entries was less than a statistically significant sample, which limited the ability to correlate trends – though we were able to identify general trends. We also used the GovHackATO dataset ( ) at a greater level of location granularity to compare macroeconomic and ABS (eg: SIEFA) data as an overlay to the AFSA Non-compliance dataset. A number of other datasets were evaluated for linkage (eg: Victorian Government unemployment data from budget information, Australian Government budget information). However, again those datasets contain information at a whole of country or state level, which limits their usefulness at identifying trends with the AFSA data.

Evidence of Work



High-Res Image

Team DataSets

Australia Taxation Office Gov Hack 2018

Description of Use: We overlaid the ATO data with the AFSA data to determine if there were correlations that could be made.Aspects of the data set sit within our heat map.

Data Set

Non-compliance in personal insolvencies

Description of Use: We analysed the AFSA data to determine which factors contribute most to non-compliance in insolvency, we did this both with manual analysis of the data and developing a machine learning algorithm. We then developed visualisations of this data and creates a website and friendly user interface for AFSA employees and other users.

Data Set


Check back here once the first checkpoint passes to see the challenges this team has entered.

Back to Projects